Diabetic kidney disease: Radiological assessment and clinical correlations

Abstract

Diabetic kidney disease (DKD) is the leading cause of chronic kidney disease and end-stage kidney disease worldwide. It affects approximately 20%-50% of individuals with diabetes mellitus, with a substantially higher prevalence among patients with type 2 diabetes, in whom it is the primary indication for renal replacement therapy. In type 1 diabetes, DKD typically develops after a latency period of 5-15 years following disease onset, whereas in type 2 diabetes it may already be present at diagnosis, reflecting the often prolonged, subclinical course of hyperglycemia preceding clinical recognition. The pathogenesis of DKD is multifactorial and involves the complex interplay of chronic hyperglycemia, oxidative stress, persistent inflammatory activation, and intrarenal hemodynamic alterations. These mechanisms promote endothelial dysfunction, activation of profibrotic signaling pathways, and structural remodeling of the renal parenchyma, ultimately leading to glomerular hyperfiltration, progressive albuminuria, and a gradual, irreversible decline in estimated glomerular filtration rate. While traditional diagnostic approaches rely heavily on biochemical markers and clinical parameters, radiological imaging has emerged as a crucial complement in the assessment of DKD. This review examines the role of various imaging modalities, including ultrasonography, computed tomography, magnetic resonance imaging, and new emerging techniques, in the evaluation of DKD. Ultrasonography remains the first-line imaging tool, providing valuable information about renal size, echogenicity, and structural abnormalities that correlate with disease progression and functional decline. Advanced imaging techniques offer enhanced capabilities for assessing renal perfusion, fibrosis, and microstructural changes, potentially enabling earlier detection and more precise monitoring of disease progression. The integration of radiological findings with clinical and laboratory data could significantly improve diagnostic accuracy, risk stratification, and therapeutic decision-making. This comprehensive review highlights the current applications, limitations, and future directions of radiological assessment in DKD, emphasizing the importance of multimodal imaging approaches in optimizing patient care and outcomes.

Key Words: Diabetic kidney disease; Diabetic nephropathy; Ultrasonography; Magnetic resonance imaging; Computed tomography; Renal imaging

Core Tip: Radiological imaging plays an increasingly important role in diabetic kidney disease assessment, effectively complementing traditional biochemical markers. Ultrasonography remains the primary imaging modality, while advanced techniques including magnetic resonance imaging and computed tomography provide enhanced evaluation of renal perfusion, fibrosis, and microstructural changes. Integration of multimodal imaging with clinical data improves diagnostic accuracy, enables earlier disease detection, and optimizes therapeutic decision-making, ultimately enhancing patient outcomes in this leading cause of chronic kidney disease.

INTRODUCTION

Diabetes mellitus is an increasing global public health burden, a widespread non-communicable disease with significant medical and socioeconomic consequences. Among its many complications, diabetic kidney disease (DKD) remains the leading cause of end-stage kidney disease (ESKD) in many high-income countries[1,2]. The challenge in DKD is late detection: Traditional diagnosis relies on widely available clinical and biochemical parameters, such as albuminuria and estimated glomerular filtration rate (eGFR), which have significant limitations. They do not allow direct assessment of renal structural changes, have poor ability to distinguish DKD from other nephropathies in diabetic patients, and often identify the disease at an advanced stage when treatment options are limited[1]. In recent years, advanced imaging techniques have played an increasingly important role in the non-invasive characterization of DKD, overcoming some limitations and enabling direct visualization of anatomical and functional changes in the kidney. In this review, we critically analyze the main imaging methods applicable to DKD, from ultrasound with elastography and contrast-enhanced ultrasound (CEUS) to magnetic resonance imaging (MRI) with functional sequences such as diffusion-weighted imaging (DWI) and blood oxygen level-dependent (BOLD) imaging, as well as emerging spectral computed tomography (CT) techniques. We evaluate their diagnostic and prognostic potential and future prospects for integration into clinical practice.

THE GLOBAL BURDEN OF DKD

Globally, DKD is the leading cause of ESKD, accounting for approximately 50% of cases in developed countries[2]. The prevalence of DKD varies significantly depending on the population studied, type of diabetes [type 1 diabetes mellitus (T1DM) or type 2 diabetes mellitus (T2DM)], country of residence, gender, and ethnicity[2]. The global incidence of T2DM-related chronic kidney disease (CKD) more than doubled from 1992 to 2021, rising from approximately 801, 870 cases to over 2 million cases per year. In 2021, the estimated global prevalence of diabetes among people aged 20 to 79 was 11%, and it is expected to increase to 12% by 2045[3]. These data are supported by the increasing prevalence of major risk factors for the onset of T2DM, primarily obesity and longer average life expectancy[4]. Typically, DKD is defined by the presence of CKD characterized by persistent albuminuria (> 30 mg/24 hours, urine albumin-creatinine ratio r > 30 mg/g), associated with a gradual reduction in eGFR (eGFR < 60 mL/minute/1.73 m²) in a person with diabetes[5]. DKD is associated with a very high cardiovascular risk, as supported by the significant incidence of premature death from cardiovascular complications, even before reaching ESKD[6]. Several risk factors contribute to the development and progression of DKD, including longer duration of diabetes, poor glycemic control, hypertension, dyslipidemia, and genetic susceptibility[7].

PATHOPHYSIOLOGY AND CLASSIFICATION OF DKD

The pathophysiology of DKD is complex, arising from the interplay of several interconnected processes. Metabolic disturbances such as hyperglycemia, hemodynamic stress from increased intraglomerular pressure, and persistent inflammation all contribute to kidney damage. These effects are reinforced by pro-fibrotic signaling that promotes scarring, as well as genetic and epigenetic factors that influence individual susceptibility. Together, these mechanisms drive the development and progression of DKD[8].

DAMAGE INDUCED BY HYPERGLYCEMIA

Excessive and persistent glucose concentration in the bloodstream triggers a series of harmful metabolic pathways within target kidney cells, such as mesangial, endothelial, and podocyte cells. The polyol pathway is enhanced due to excess glucose being converted to sorbitol. Intracellular accumulation of sorbitol causes osmotic stress, depletion of the nicotinamide adenine dinucleotide phosphate hydrogen cofactor, and consequently, an increase in oxidative stress[9]. Another characteristic damage caused by the hyperglycemic state is the generation of advanced glycation end products (AGEs). Hyperglycaemia promotes the non-enzymatic glycation of lipids and proteins, resulting in the formation of AGEs. When these compounds bind to their specific receptors (RAGE), they trigger intracellular signaling pathways, such as nuclear factor kappa-B. This leads to inflammation, increased production of reactive oxygen species, and greater extracellular matrix deposition, which result in fibrosis and mesangial expansion[10]. Hyperglycemia also promotes the activation of protein kinase C (PKC). This activation is supported by increased biosynthesis of diacylglycerol, which in turn activates the PKC enzyme. This can compromise vascular function and permeability, ultimately contributing to an increase in the extracellular matrix[11].

HEMODYNAMIC ALTERATIONS

Changes in circulation and vascular pressure within the kidneys are a central pathogenic element in DKD. In the early stages of diabetes, an increase in eGFR is frequently observed. This phenomenon is primarily caused by dilation of the arteriole that carries blood to the glomerulus (afferent arteriole) and, secondarily, by narrowing of the arteriole that exits it (efferent arteriole), a process facilitated by hyperactivity of the renin-angiotensin system (RAS)[9]. The resulting hypertension at the glomerulus increases mechanical stress (shear stress) and the permeability of the glomerular capillary membrane, causing damage to the podocytes and endothelial lining[10]. Moreover, DKD is characterized by hyperactivation of the RAS. The RAS within the kidney is excessively active in DKD. This hyperactivity stimulates vasoconstriction, particularly of the efferent arteriole, increases inflammatory and fibrotic processes, and induces hypertrophy[12].

INFLAMMATION AND FIBROSIS

DKD is characterized by a condition of persistent low-grade inflammation, as already reported in CKD[13]. Metabolic pathways activated by excess glucose, such as AGEs/RAGE and oxidative stress, promote the release of pro-inflammatory cytokines, including tumor necrosis factor-α and interleukin-6, and chemokines by resident renal cells, such as podocytes, mesangial cells, and tubular cells. These signaling molecules recruit and activate immune system cells, such as monocytes and macrophages, in the kidney tissue, exacerbating the damage[14]. Moreover, DKD is characterized by a progressive fibrotic stimulus. Transforming growth factor beta (TGF-β) plays a central role in fibrosis, being stimulated by both hyperglycemia and AGEs. TGF-β induces mesangial cells to synthesize excessive amounts of extracellular matrix, such as collagen and fibronectin, leading to enlargement of the mesangial area and, in advanced cases, nodular glomerulosclerosis (Kimmelstiel-Wilson lesions)[15]. It is important to note that oxygen deficiency (hypoxia) in the tubulointerstitial tissue also contributes significantly to the development of fibrosis in this specific renal area[16].

STAGES AND HISTOLOGICAL CLASSES OF DKD

The pathophysiological complexity of DKD, encompassing metabolic derangements, hemodynamic alterations, chronic inflammation, and progressive fibrosis, underscores the inadequacy of relying solely on biochemical markers for comprehensive disease assessment. While albuminuria and eGFR remain cornerstones of clinical evaluation, they provide limited insight into the underlying structural changes within the renal parenchyma, such as mesangial expansion, glomerulosclerosis, tubulointerstitial fibrosis, and microvascular remodeling, which characterize the histological progression from class I to class IV DKD (Table 1)[17-20]. More importantly, these serological markers often indicate a pathological condition only after irreversible nephron loss has occurred, leaving a critical window of potentially reversible disease undetected. Advanced imaging modalities offer the potential to bridge this diagnostic gap by enabling direct, non-invasive visualization of structural and functional alterations within the kidney. In the following sections, we examine how ultrasonography, CT, and MRI can complement conventional biomarkers by providing anatomical, hemodynamic, and microstructural information that may facilitate earlier detection, more accurate phenotyping, and improved prognostic stratification in DKD.

Table 1 Histological classes of diabetic kidney disease.

Histologic class	Glomerular	Tubulointerstitial	Vascular
Class I	Isolated thickening of the GBM. Minimal or absent mesangial alterations[18]	Normal or tubular hypertrophy (secondary to hyperfiltration)	Normal or mild arteriolar hyalinosis (more common in the efferent arteriole)[17]
Class II	Diffuse mesangial matrix expansion, mild (IIa) or severe (IIb). Absence of nodular lesions (Kimmelstiel-Wilson) and global glomerulosclerosis < 50%[18]	Focal tubular atrophy and interstitial fibrosis (correlated with the severity of mesangial expansion). Glycogen accumulation in tubular epithelial cells (Armanni-Ebstein lesions)[19]	More evident arteriolar hyalinosis, often involving both afferent and efferent arterioles[17]
Class III	Presence of at least one Kimmelstiel-Wilson nodular lesion. Nodules are acellular and PAS-positive. Global glomerulosclerosis < 50%[19]	Moderate-to-severe IFTA. Infiltration of inflammatory cells[19]	Marked arteriolar hyalinosis. Atherosclerosis of interlobular arteries (characteristic of DKD, but not specific)[17,20]
Class IV	Diffuse global glomerulosclerosis > 50%. The most advanced form of diabetic glomerulosclerosis, with widespread involvement of all remaining glomeruli[18]	Severe and diffuse IFTA (a key predictor of progression to end-stage renal disease[19]	Extensive vascular sclerosis and vascular wall hypertrophy[20]

This table summarizes the histopathological classification of diabetic kidney disease, highlighting the progressive glomerular, tubulointerstitial, and vascular alterations that characterize each class. Class I is defined by isolated glomerular basement membrane thickening with minimal mesangial involvement and usually mild or absent vascular changes. Class II is marked by diffuse mesangial expansion, mild (IIa) or severe (IIb), accompanied by early tubulointerstitial injury and more evident arteriolar hyalinosis. Class III is characterized by the appearance of Kimmelstiel-Wilson nodules and moderate-to-severe interstitial fibrosis and tubular atrophy, together with marked arteriolar involvement. Class IV represents the most advanced stage, defined by diffuse global glomerulosclerosis affecting more than half of the glomeruli, extensive interstitial fibrosis and tubular atrophy, and widespread vascular sclerosis. GBM: Glomerular basement membrane; IFTA: Interstitial fibrosis and tubular atrophy; DKD: Diabetic kidney disease; PAS: Periodic acid-Schiff.

ULTRASONOGRAPHY

Ultrasonography is a key technique for studying DKD patients with kidney impairment. Modern ultrasonography is continually evolving with new techniques. Previously limited to doppler and B-mode imaging, the clinical landscape is now shifting towards more advanced methods such as CEUS and shear wave elastography (SWE).

B-MODE

B-mode ultrasound is the backbone of ultrasound imaging due to its simplicity, low cost, and non-invasive assessment of the kidney. It provides measurements of renal dimensions, parenchymal thickness, and echogenicity[21]. The physiological measurements of adult kidneys are 10-12 cm by 8 cm, with a cortical thickness of 7-10 mm. Renal diseases are typically characterized by a reduction in kidney volume, and deviations from these dimensions have been correlated with CKD[22]. This explains why ultrasound plays a crucial role in evaluating morphological modifications and relative changes at different stages of DKD. Indeed, in DKD, an increase in kidney volume has been observed, likely secondary to hypertrophic changes. This probably reflects hyperfiltration and renal injury resulting from elevated intraglomerular pressure[23]. To confirm the relationship between DKD ultrasound characteristics, Rigalleau et al[24] analyzed a cohort of 75 T2DM patients over a 5-year follow-up. These patients included 6 normoalbuminuric, 39 macroalbuminuric, and 30 macroalbuminuric individuals. Moreover, patients were divided into two groups based on kidney size. During the follow-up period, researchers observed that 9 of the 11 patients who started dialysis had larger kidneys. They concluded that, in DKD patients, an increase in kidney size could serve as a marker to predict CKD progression. A similar result was observed by Mancini et al[23] compared renal volume between 88 DKD patients and 73 controls. They observed that renal volume in DKD patients was increased compared to controls. Moreover, they reported that renal hypertrophy in T2DM patients could be observed even without proteinuria. Ultrasound findings in DKD could also be useful for predicting the prognosis of kidney dysfunction progression. Ham et al[25] used eGFR and a score based on ultrasound findings, collected from 252 patients, to develop a model for predicting renal prognosis. Their results showed that increased renal echogenicity correlated with declining renal function (eGFR decrease of ≥ 10%). Also, Derchi et al[26] compared ultrasound imaging of 85 patients with T2DM and 42 healthy patients. In the T2DM group, renal volume was increased more than in the controls. Controversially, another study[27] demonstrated that commonly used kidney measures have limited specificity for prognostic stratification and do not show a significant association with unfavorable renal outcomes.

DOPPLER ULTRASOUND

Doppler ultrasound is a non-invasive ultrasound technique used to evaluate renal arterial and venous flux[28]. The principal and most reproducible parameter is the renal resistance index (RRI). It is calculated as (peak systolic velocity end diastolic velocity)/peak systolic velocity. Under physiological conditions, its value ranges from 0.47 to 0.7[29]. It provides information about intrarenal vascular resistance and its variations in relation to pathological conditions affecting the kidneys. RRI analysis could help identify DM2 patients at high risk of developing renal damage. However, this technique has several limitations, including operator dependence, signal attenuation in obese patients, and increased intrarenal resistive index in other pathological or physiological conditions, such as hypertension and advanced age[30,31]. According to several studies, no significant difference was found between RRI values and albuminuria levels in their ability to detect DKD[32,33]. Their data showed a progressive increase in RRI associated with rising levels of albuminuria. Therefore, elevated RRI appears to be closely linked to microalbuminuria and macroalbuminuria, making it an important ultrasound marker in patients with T2DM. However, the exact pathophysiological mechanism is not clear. Some authors believe that it is secondary to diabetic glomerulopathy with arteriosclerotic lesions[34], whereas others indicate a non-specific renal scarring process, such as interstitial fibrosis (IF), loss of capillaries and glomeruli, with a relative loss of intrarenal vessels[35]. Jung et al[36] studied 59 patients with diabetes who underwent renal doppler and renal biopsy. Patients were sequentially divided into DKD and non-DKD groups. An increase in RRI was observed in patients with DKD (0.71-0.76 vs 0.66-0.72; P = 0.032). However, in the pathophysiology of DKD, a central role is played by vascular wall lesions and decreased tissue perfusion. More specifically, Lin et al[37] analyzed vascular perfusion in 54 DKD patients compared with 36 non-DKD patients. Their results showed that both maximum and minimum blood velocity of the main, segmental, and interlobular arteries were reduced with increasing DKD grade (respectively P < 0.05, P < 0.001, P < 0.0001), while, at the same time, the RRI of intrarenal arteries increased as DKD advanced. Controversially, Nickavar et al[38] reported in a cohort of 49 T1DM patients, 21 with DKD and 28 without DKD, that RRI variation was not significantly different between the two groups (DKD 0.61 ± 0.04 vs no DKD 0.59 ± 0.04; P = 0.09), concluding that RRI does not represent a reliable marker for screening early DKD.

CEUS

CEUS uses as a contrast agent tiny gas-filled microbubbles similar in size to red blood cells[39]. This agent follows the blood flow in the microcirculation, and through a time-intensity curve, it is possible to evaluate the dynamic process and intensity changes of the contrast agent based on tissue perfusion[40]. CEUS is used to diagnose renal diseases such as carcinoma lesions and vascular diseases[41,42] but the information on renal perfusion could also be useful for DKD. In a study by Zhang et al[43] on haemodynamic changes in the cortical region of DKD patients, the results demonstrated that the overall mean arterial velocity of all segments is lower than that of the controls. In another study, Wang et al[44] compared CEUS results from 55 DKD patients and 26 patients without DKD. They found that rise time, fall time, and mean transit time were higher in DKD patients. These findings indicate that perfusion enhancement was reduced in DKD. These results are consistent with those of the study by Dong et al[45], in which 46 DKD patients were compared with 25 healthy individuals, and all underwent CEUS. Moreover, these preliminary data suggest that RRI could be replaced or, more appropriately, complemented by CEUS for the evaluation of renal perfusion in DKD patients. CEUS can represent microvascular renal perfusion in real time and in dynamic modalities. This altered renal perfusion may be secondary to DKD-related changes such as Kimmelstiel-Wilson nodules, glomerular sclerosis, and renal IF[1].

SWE

SWE is an ultrasound technique that evaluates tissue elasticity. It measures the velocity of the transmitted shear wave through the tissue in metres per second, and the software processes the data into a tissue stiffness parameter, converting it into kilopascals (kPa)[46]. Low speed corresponds to a soft medium, while high speed corresponds to a stiff medium. Therefore, SWE could be considered an important tool for analyzing chronic morphological alterations and for evaluating the severity of cortical rigidity in DKD[47]. Several studies have investigated optimal cut-off values to identify DKD with adequate sensitivity and specificity, aiming to differentiate it from other conditions. For example, Yuksekkaya et al[48] analyzed the different SWE levels in a cohort of 108 DKD patients and 17 healthy controls. They noted higher levels of SWE in patients with DKD than in the control group (10.156 ± 1.75 kPa vs 8.241 ± 1.4 kPa; P < 0.001) and higher SWE in DKD stage 3 compared to DKD stage 1 (P < 0.05). Their results also identified a SWE cut-off level of 9.23 kPa to predict early DKD (sensitivity 67%, specificity 82%). Interestingly, Koc and Sumbul[49] reported increased SWE levels in DKD patients compared to those without DKD, while RRI was similar in both groups. These data confirm the value of SWE in identifying cellular and microvascular changes in early DKD. A more recent study[50] evaluated SWE in relation to creatinine in 60 patients with DKD and 20 controls. They found an increase in SWE and RRI in the DKD groups (P < 0.001) and set the SWE cut-off for differentiating DKD from non-DKD at > 10.5 kPa (100% specificity and 100% sensibility), and > 41 kPa for distinguishing DKD stage 3 from stage 2 (sensitivity 62%, specificity 64.9%). In contrast, Gunduz et al[51], in a cohort of patients (40 with T2DM and 17 controls), did not find a significant SWE difference between groups (P > 0.05). A similar result came from a systematic analysis by Cè et al[52]. Their study included 69 studies with a total of 6728 patients, showing no significant SWE difference between CKD and healthy individuals, while a relationship emerged between SWE and the degree of fibrosis. A single ultrasound parameter may not be sufficient for the evaluation of DKD. Therefore, the main ultrasound techniques (B-mode, doppler, and SWE) can be used together as part of multimodal ultrasound imaging technology (MUIT). MUIT integrates images obtained from these modalities, providing comprehensive, higher-resolution information to support diagnosis and treatment planning. MUIT could be used in many medical fields, such as cancer[53] and liver fibrosis[54].

CT

CT is an attractive and widely available imaging technique. Modern CT infrastructure enables precise quantification of tissue attenuation, volume, and fat distribution, and can be combined with advanced image analysis. However, ionizing radiation and the potential nephrotoxicity of iodinated contrast require more judicious use, particularly in advanced kidney dysfunction or in other organ dysfunctions such as heart failure and cirrhosis, which predispose to acute renal decompensation in CKD[55,56]. Non-contrast and low-dose CT protocols, including those assisted by artificial intelligence (AI), are being developed to reduce radiation exposure[57]. Another possible workaround could be to analyze scans obtained for other indications, in order to reduce inappropriate radiation exposure and improve understanding of DKD, as has already been done in other population[58]. In this setting, different potential parameters for early detection, phenotyping, and prognostication are being investigated.

For example, microvascular changes are established pathophysiological features of DKD, and CT may help identify signs of DKD. This hypothesis was recently investigated by Zheng et al[59], who evaluated 34 T2DM patients and 19 non-diabetic controls using spectral CT. In this study, spectral CT imaging revealed altered cortical kidney perfusion in T2DM patients, demonstrated by a faster iodine wash-in phase. These results potentially support the use of spectral CT as an alternative protocol for early identification of kidney injury in diabetic patients, regardless of proteinuria and eGFR. CT scans can also be useful for evaluating the musculoskeletal system in DKD patients. Recently, musculoskeletal attenuation has been explored as a possible CT-related biomarker. In a retrospective study by Fan et al[60], which included 267 T2DM patients with or without DKD, lower skeletal muscle CT attenuation values were independently associated with a higher risk of early DKD compared to those with higher values. Another parameter gaining attention is the evaluation of abdominal adipose tissue. Abdominal adipose tissue can usually be classified into different fat depots according to their position and function: Visceral, subcutaneous, perirenal, and renal sinus adipose tissue. The study of these tissues can be accurately conducted using CT and MRI. In the retrospective study by Peng et al[61], which included 151 healthy controls and 579 T2DM patients (281 non-DKD, 298 DKD), disease progression and risk stratification were explored using CT attenuation changes of the renal parenchyma, perirenal fat, and renal sinus fat as parameters. These parameters have been identified as possible imaging biomarkers with significant correlations to key clinical parameters[61]. Accordingly, Fan et al[62] studied the association between CT-derived perirenal adipose tissue characteristics and DKD risk categories in 404 patients with T2DM. Their results showed that, in T2DM patients, higher perirenal fat attenuation values are correlated with an increased risk of higher DKD Kidney Disease: Improving Global Outcomes risk categories[62]. The growing interest of the scientific community in recent years suggests that future developments are highly likely. For example, the development of CT-centred radiomics based on machine learning models could represent pragmatic strategies to enhance the strengths of CT[63,64]. Indeed, the use of AI for guiding the phenotyping as being already conceptualized for other diseases and DKD[65-67]. However, external validation across different centres and CT infrastructure is needed.

MAGNETIC RESONANCE

MRI shows great potential in assessing renal disease, particularly due to its ability to examine soft tissues without the use of contrast media or ionizing radiation, and to obtain multiple biomarkers using techniques that assess different properties of living tissue. Several studies suggest that these biomarkers, such as T1 mapping, the apparent diffusion coefficient (ADC) from DWI, and perfusion metrics, can detect changes in renal microstructure that are not detectable with conventional markers, or can provide information complementary to traditional markers such as eGFR and albuminuria[68].

Additional advantages of using MRI in renal investigation include the ability to quantify specific pathophysiological processes, allowing repeated scanning and longitudinal re-evaluation without exposure to ionizing radiation in both healthy volunteers and patients. In addition to its primary use in macrostructural morphological and dimensional assessment of the kidneys, MRI can provide different parameters to evaluate various aspects of renal function, such as hemodynamics, perfusion, oxygenation, and microstructural changes (fibrosis, inflammation). In the specific case of DKD, it also represents a potentially valuable complement or alternative to kidney biopsies, facilitating comprehensive bilateral assessment that is not affected by sampling bias and can capture regional heterogeneity[69]. The technique’s high spatial resolution allows clear visualization of cortical and medullary structures, making it particularly valuable in clinical trials for elucidating drug mechanisms of action. The ability to obtain various imaging biomarkers that can be combined has led to the development of the concept of multiparametric MRI, which is also applied to renal MRI.

The field of renal MRI biomarkers has experienced significant growth through interdisciplinary collaboration among diverse medical and scientific specialists. This collaborative momentum led to the establishment of “PARENCHIMA”, a pan-European research initiative launched in 2017 under the European Cooperation in Science and Technology framework, which sought to address obstacles hindering the clinical implementation and commercial development of renal MRI biomarkers[70]. The initiative’s legacy continues through “renalmri.org”, which disseminates clinical guidelines for MRI biomarker applications in CKD. Growing scholarly interest in this field, exemplified by broad participation in the 2021 International Society for Magnetic Resonance in Medicine Workshop on Kidney MRI Biomarkers, has driven the development of systematic reviews and consensus-based technical protocols, with renal MRI biomarkers gaining increasing prominence in nephrology research[71-78].

PHASE CONTRAST MRI

Phase-contrast MRI (PC-MRI) is a non-contrast technique that directly measures renal blood flow (RBF) throughout the cardiac cycle by multiplying the sectional area of the renal artery by the mean blood velocity[69]. PC-MRI demonstrates significant clinical utility in CKD, revealing decreased RBF in both mixed CKD and DKD populations compared to healthy volunteers (normal RBF approximately 1.1 L/minute)[79,80]. The technique shows strong diagnostic performance, differentiating between DKD stages G3 and G4/5 with an area under the curve (AUC) of 0.88 (P = 0.004)[79], and correlates well with gold-standard pulmonary arterial hypertension infusion methods[81]. Measurement reproducibility is good, with coefficients of variation ranging from 6% to 13% across different populations and study intervals. Global kidney perfusion, calculated by normalizing RBF to kidney volume, shows highly significant reductions in both diabetic and non-diabetic CKD populations.

Beyond basic flow measurements, PC-MRI provides hemodynamic biomarkers throughout the cardiac cycle, including end diastolic velocity, peak systolic velocity, and renal arterial resistive index. Although acquisition requires only one breath-hold per kidney, the planning process can be time-consuming and technically demanding.

MRI FROM MACROSTRUCTURE TO MOLECULAR IMAGING

Compared to ultrasound, MRI volume measurements demonstrate superior accuracy and precision, as shown in comparative studies of healthy individuals and DKD patients with documented progressive renal hypertrophy across the glycemic spectrum from normal to diabetic states, as assessed by MRI volumetry[82] and a paradoxical initial kidney enlargement followed by accelerated volume loss in T2DM, potentially related to hyperfiltration and oxidative stress[83]. A reversible increase in size occurs in acute kidney injury in non-DKD patients, in contrast to advanced DKD, in which total kidney volume reduction is highly reproducible[79,84].

While being widely consolidated as a technique for macroscopic uses such as morphologic assessment, non-ionizing characterization of focal lesions and acute benign pathology assessment and for surveillance, recent technological developments have made possible the use of MRI for the focused assay of microstructural changes. The development of new MRI techniques brought to the ability to investigate microscopic organic properties of renal tissue (such as parenchymal fibrosis) and to quantify metabolic parameters such as blood oxygenation and parenchymal blood flow both with and without contrast medium use. These techniques, derived from neuroradiological applications offer valuable and accurate information with the advantage of operator independence and the ability to assess each kidney separately.

The sum of these techniques is nowadays referred with increasing frequence as functional MRI (fMRI)[85,86]. Each of these techniques offers an overview aimed to specific functional mechanisms or microstructural measurements. MRI may find a precious role in the longitudinal investigation of patients with repeated imaging, to confirm disease progression, or also find a decisive role in prognostic assessment with a sensitivity that by far anticipates changes in bio-humoral markers, such as creatinine, that is indicative of structural damage has irreversibly impaired function.

Synoptically fMRI techniques are: (1) T1 mapping, that analyzing changes in T1 relaxation time provides and assessment of either fibrosis of inflammation; (2) BOLD, that measures kidney tissue oxygenation by detecting changes in deoxyhemoglobin (a sign of early damage in DKD); (3) Arterial spin labeling (ASL), that uses blood as an intrinsic contrast agent to map blood flow (perfusion) and assess kidney function and progression; (4) DWI, that maps water molecule diffusion to assess microscopic changes, water movement, and tissue microstructure in the kidneys, revealing fibrosis; and (5) Magnetic resonance elastography (MRE)-MRI that measures kidney stiffness, indicating fibrosis, a key feature of DKD progression.

The use of these techniques stays at the moment at a research level as homogeneous consensus and proof of accountability of the results are required with prospective randomized studies. In addition, availability of these techniques does not involve systematic technological update but certainly involves focused education of dedicated physicians.

T1 mapping

T1 mapping is a quantitative, non-contrast-enhanced MRI technique derived from MR relaxometry that reflects tissue-specific relaxation properties. T1 relaxation time is a tissue specific measurement and by assessing it, it is possible to derive information on the composition of tissues, in particular the presence of collagen or inflammation. T1 maps show the rate at which nuclear spin magnetization returns to equilibrium after radiofrequency excitation. T1 values can be obtained using various acquisition and post-processing methods, and standardized recommendations for their generation have recently been provided by the PARENCHIMA initiative[72].

All pathologic conditions that involve inflammatory change through the TGF-beta signaling, involve collagen deposition in tissues, consequently developing parenchymal fibrosis. As collagen is characterized by a microstructure which involves organized molecules with a stable thermodynamic balance, at a molecular level it is characterized by a shorter T1 relaxation time. Consequently, T1 mapping may be used to assess fibrotic changes in tissues[85,86]. On the opposite, acute inflammatory conditions (such as IgA nephropathy) involve an increase of T1 relaxation time due to inflammatory infiltrate.

T1 relaxation time (or its reciprocal, R1) is influenced by the tissue microenvironment, particularly water content, making it a potential biomarker for pathological changes. Altered T1 values have been reported in liver and kidney diseases, including decreased cortical T1 in cirrhosis and increased T1 in IgA nephropathy and advanced CKD[80,87,88]. In CKD, T1 mapping, similar to the ADC derived from DWI, has shown potential for distinguishing lower from higher degrees of renal fibrosis, supporting its role as a promising prognostic biomarker[80].

Despite these advantages, native T1 lacks specificity, as it may be affected by various pathological processes. Although T1 correlates with IF, it is also influenced by tubular atrophy, chronic vasculopathy, and transplant glomerulopathy[89,90]. However, when combined with other MRI-derived biomarkers such as ADC, T1 improves the detection of fibrosis and provides complementary information on disease progression, including DKD. T1, alone or in combination with other biomarkers, has shown predictive value for renal outcomes, including albuminuria, decline in kidney function, and medium-term prognosis[69,89,90].

Elevated T1 values have also been observed in acute kidney injury, with partial normalization during recovery, although persistent T1 elevation in some patients may indicate progression toward chronic kidney damage[80]. Importantly, multiple studies have demonstrated excellent repeatability of renal T1 measurements, with low coefficients of variation for cortical and medullary assessments. This high reproducibility supports the potential utility of T1 mapping as a reliable imaging biomarker in both clinical research and drug development.

The same relaxation maps may be obtained for T2 relaxation time, a measurement dependent on the energy dissipation between protonic spins. T2 maps show high sensitivity to presence of oedema and inflammation[91].

BOLD MRI

BOLD-MRI is a non-invasive technique that uses deoxygenated hemoglobin as an endogenous contrast agent to assess tissue oxygenation: Transported oxygen bound to haemoglobin is released in the tissues; during oxygen release oxyhaemoglobin is converted to deoxyhaemoglobin, that is characterized by paramagnetic properties, able to increase local magnetic susceptibility gradients within and around blood vessels. High local concentrations of deoxyhaemoglobin accelerate transverse signal decay, resulting in a shortening of the effective transverse relaxation time (T2-star) in T2-star-weighted magnetic resonance images and a corresponding increase in the transverse relaxation rate R2-star (R2-star = 1/T2-star, expressed in s^-1). R2-star is commonly used as the primary quantitative outcome measure in BOLD MRI therefore higher R2-star values indicate lower local tissue oxygenation, whereas lower R2-star values reflect higher oxygenation[85]. The physiopathological link between tissue oxygenation and DKD has been found in the mitochondrial dysfunction secondary to hyperglycaemia that reduces transmembrane ionic transportation with consequent tissue hypoxia[85].

It must be clarified that interpretation requires careful consideration of fractional blood volume and hematocrit, which also affect measurements[92,93]. Recent research has demonstrated that when fractional blood volume is properly accounted for, kidney cortex appears normoxemic in healthy individuals but hypoxemic in CKD patients, despite similar R2-star values between groups[94].

Renal oxygenation reflects the balance between oxygen supply and consumption, with most oxygen used for sodium reabsorption from the filtered fluid. As increased RBF results in more filtered sodium and therefore greater oxygen consumption, the filtration fraction (the ratio of glomerular filtration rate to renal plasma flow) is a key determinant of oxygenation status, rather than blood flow alone[95].

BOLD-MRI has been successfully used to evaluate the renoprotective effects of various medications, including sodium-glucose cotransporter 2 inhibitors, glucagon-like peptide-1 (GLP-1) agonists, and other therapeutic agents in clinical trials[96,97]. However, careful interpretation is essential, as shown by studies in which increased R2 following empagliflozin treatment may have reflected elevated hematocrit rather than worsening oxygenation[98].

Recently a prospective comparative study on the effect of depaglifozin alone against depaglifozin and sacubitril/valsartan therapy over 12 weeks employed BOLD MRI along with serologic markers to assess response to therapy, in one of the first studies employing an MRI biomarker along with the standard parameters in T2DM with impaired renal function[99].

Importantly, BOLD R2-star has shown promise as a prognostic biomarker for predicting kidney outcomes and progressive functional decline in CKD patients[69]. The technique’s utility in drug development is supported by excellent test-retest repeatability (less than 7% variation), and ongoing standardization efforts through PARENCHIMA are establishing consistent protocols for patient preparation, data acquisition, and analysis across multiple imaging centres[73,94].

ASL

Microvascular injury endothelial dysfunction ASL perfusion reduction. Another technique used to assess blood flow is ASL, which measures perfusion by tagging blood volumes flowing to the renal artery as a natural contrast agent, creating quantitative perfusion maps of organs[100].

The sensitivity to pathologic changes of this technique is explained with the rarefaction consequent to microvascular injury secondary to endothelial diabetic dysfunction that not only allows to link hypoperfusion to the ASL perfusion biomarker but also represents a predisposing factor for renal hypoxia[85].

The PARENCHIMA consensus established that only cortical perfusion measurements should be reported for kidney studies, as they provide more reliable results[77].

Clinical studies have shown that ASL cortical perfusion is reduced in DKD patients compared with healthy individuals[79], however, interpreting ASL results requires caution, as perfusion values can change due to variations in kidney volume, even when total RBF remains constant. This has been observed following blood volume expansion treatments and potentially with GLP-1 agonist[96]. The measurement variability is significant, with intra-subject coefficients of variation ranging from 9% to 33%, depending on the studied population[79,87,101].

ASL offers an advantage over PC-MRI by visualizing perfusion heterogeneity that may reveal lesions; however, it faces challenges such as susceptibility to physiological motion from breathing and peristalsis, as well as longer acquisition times. Its limited adoption is due to inconsistencies across MRI vendors, who offer different ASL techniques, and the fact that many centres develop their own custom solutions under research agreements, making standardization across sites difficult.

DWI

Renal DWI is an MRI sequence sensitive to free diffusion of water molecules within extracellular compartments. Its restriction corresponds to a reduction in ADC values. This allows DWI to be sensitive to tissue microstructural changes, including fibrosis, collagen accumulation, oedema, neoplastic infiltration, and perfusion alterations. The DWI-derived ADC has been demonstrated a useful biomarker in distinguishing between relatively low and high levels of renal fibrosis[80,89,90,102-104], supporting its utility as a surrogate marker of fibrosis and, by extension, DKD progression.

ADC has been shown to be way more performing than simple GFR in assessing fibrotic involution in kidney allografts undergoing seriated biopsies[104], while consistent correlation between ADC and IF have been reported in several studies with pathologic correlation[105-107].

In DKD, ADC has been associated with established biochemical markers of kidney function, including GFR[79]. However, ADC has shown limited sensitivity to short-term therapeutic changes and does not consistently correlate with eGFR, serum creatinine, renal hypoxia, or inflammation[105]. Recently in order to support clinical translation and to harmonize renal DWI protocols and facilitate multicenter studies PARENCHIMA has published consensus-based recommendations[108], while other clinical trials are still investigating relationships between DWI biomarkers and renal fibrosis in diabetes[109].

Although traditional serological measurements remain the clinical standard for assessing and stratifying risk in kidney disease, including DKD, and are strongly associated with renal and cardiovascular outcomes[110-112], MRI approaches provide additional information on renal structure and perfusion that cannot be obtained with conventional markers. A recent meta-analysis assessed how fMRI detects structural and functional alterations in DKD, addressing the limitations of conventional biomarkers and the problem of the scarce number of renal biopsy in diabetic patients[113]. A total of 24 studies, encompassing 1550 participants, using five fMRI techniques and seven quantitative parameters. ASL-MRI showed the clearest distinction between DKD and healthy volunteers: RBF was significantly lower in DKD patients, with a weighted mean difference of -99.03 mL/100 g/minute [95% confidence interval (CI): -135.8 to -62.27, P < 0.00001], reflecting the microvascular impairment typical of DKD, suggesting early damage to cortical perfusion. Diffusion tensor imaging (DTI) MRI also revealed significant differences, with fractional anisotropy (FA) reduced by -0.02 in DKD compared to controls (95%CI: -0.03 to -0.01, P < 0.0001). The decrease in FA in diabetic patients indicates a loss of microstructural tissue integrity, probably linked to IF, tubulo-interstitial alterations, and disorganization of the anisotropic structures of the cortex. This makes DTI a sensitive tool for detecting structural changes even in the absence of clinically evident changes in renal function. In contrast, BOLD-MRI and intravoxel incoherent motion imaging (IVIM-DWI) showed no statistically significant differences between groups, although IVIM parameters demonstrated some potential for detecting early DKD changes. The absence of differences in BOLD parameters suggests that renal oxygenation is not a sensitive marker in the stages of DKD analyzed, or that BOLD suffers from methodological limitations (respiratory variability, hematocrit dependence) that reduce its discriminatory capacity.

The integrated interpretation of parameters derived from ASL, DTI, IVIM, and BOLD reflects the pathophysiological complexity of DKD, which involves perfusion, microstructure, diffusivity, and oxygen metabolism. A multiparametric approach allows simultaneous detection of vascular, fibrotic, and functional alterations, overcoming the limitations of individual techniques. This strategy is in line with the evolution of renal diagnostics towards non-invasive biomarkers capable of anticipating damage compared to traditional clinical measures such as creatinine and eGFR. The study proposal is particularly relevant in light of the difficulty of obtaining renal biopsies in diabetic patients, making the multiparametric profile one of the most promising candidates for early assessment, progression monitoring, and prognostic stratification[113].

MRE

MRE provides a non-invasive assessment of tissue stiffness by measuring the propagation of externally induced mechanical waves, with wave velocity increasing in stiffer tissue[114]. Compared with renal biopsy, MRE enables whole-organ assessment and has been widely applied to monitor fibrotic progression in liver disease, suggesting potential relevance for DKD[115,116]. Although renal stiffness measurements show good reproducibility[116,117], studies in native CKD and DKD kidneys have demonstrated decreased stiffness with advancing disease despite increasing fibrosis, a pitfall possibly due to renal perfusion changes[118,119]; this suggests that elastographic assessment in DKD still requires validation as concurrent pathological processes may overlap with confounding effect on fibrosis-specific signals.

CONTRAST-ENHANCED DYNAMIC MRI

Gadolinium-based dynamic contrast-enhanced MRI (DCE-MRI) is a technique that tracks intravenous gadolinium agents through the kidney to measure cortical perfusion, filtration fraction, tubular volume, and blood volume[120]. While it shows promise for assessing kidney function, DCE-MRI faces significant barriers to widespread adoption in DKD, mainly due to variability and low concordance between measurements, which significantly limit the reproducibility of results, and due to restrictions on its use in patients with advanced CKD, as brain deposition and toxicity make gadolinium unsuitable for repeated administrations over time.

Studies have reported that DCE-derived eGFR can range from 50% underestimation to 6% overestimation compared to reference methods, with errors varying across patient populations[69]. Intra-subject variability ranges from 15% to 22%, and comparisons with other MRI techniques, such as ASL and PC-MRI, show poor agreement at the individual patient level[92].

MULTIPARAMETRIC MRI

While single MRI biomarkers may show limited accuracy secondary to technical limitations still not overcome or to overlapping biochemical, thermodynamic or microenvironmental changes in hydration, perfusion inflammation, the use in combination of two or more MRI biomarkers may improve the accuracy in the detection of parenchymal changes in DKD, especially in longitudinal monitoring over time with an increasing trust towards multiparametric MRI.

Several studies in patients with CKD have shown that lower cortical ADC and perfusion values, as well as higher T1 values, are associated with a greater degree of structural damage and, in some cases, a more rapid progression of renal function decline[68,117,118].

In their prospective study Hua et al[68] evaluated MRI markers for non-invasive assessment of CKD and renal IF in 43 CKD patients and 20 controls: T1 mapping and diffusion-based MRI parameters demonstrated progressive abnormalities with worsening CKD and were strongly correlated with eGFR and IF. Machine-learning models combining T1 mapping and diffusion metrics accurately distinguished CKD patients from controls and effectively assessed fibrosis severity, achieving high accuracy (0.84), sensitivity (0.70), and specificity (0.92) (AUC = 0.96). These findings clearly demonstrate that various MRI parameters, particularly those derived from T1 mapping and diffusion techniques, change progressively as CKD advances. Reductions in cortical diffusion coefficients and increases in T1 relaxation times reflect renal parenchymal structural alterations associated with functional loss and IF, as indicated by their strong correlations with eGFR and IF. Integrating these parameters into machine-learning models provides high diagnostic accuracy, highlighting the potential of multiparametric MRI as a non-invasive tool for CKD evaluation, although its utility remains diagnostic rather than long-term predictive.

In their longitudinal study, Buchanan et al[121] followed 22 patients with CKD G3-4 using annual multiparametric renal MRI. Patients who later showed faster CKD progression had higher cortical and medullary T1 and lower cortical perfusion at baseline. Over two years, progressors exhibited declining kidney volume and ADC, along with rising cortical T1 and reduced perfusion, while stable patients showed only reduced kidney volume. Findings suggest that elevated T1 and reduced perfusion may predict CKD progression, and that T1, kidney volume, and ADC are promising markers for monitoring disease over time. The results show that multiparametric MRI can provide sensitive indicators for predicting and monitoring the progression of CKD. Already at baseline, patients who subsequently worsened had higher T1 values in the cortex and medulla and reduced cortical perfusion, signs of early structural and microvascular changes. At follow-up, progressors showed a significant decrease in total renal volume and ADC, along with a further increase in T1, consistent with parenchymal deterioration. On the opposite T2 relaxation time did not show significant modifications, suggesting that this parameter is not very sensitive to the evolution of CKD. Overall, the study indicates that T1, perfusion, TKV, and ADC could be promising biomarkers for early identification of at-risk patients and non-invasive monitoring of CKD progression.

However, it should be noted that the strength of the evidence varies between different MRI parameters. Among the available longitudinal studies, the cortico-medullary ADC difference (ΔADC) has shown the most robust prognostic outcome. In a prospective cohort of patients with CKD or kidney transplantation, reduced ΔADC independently predicted, after adjustment for eGFR and proteinuria, a higher risk of rapid decline in renal function or initiation of dialysis (hazard ratio = 4.6)[122].

Recently, a prospective study evaluated the usefulness of BOLD-MRI and DWI for early detection of DKD. Forty participants were enrolled, including 20 diabetic patients subdivided by albuminuria level and 20 healthy controls[123]. Analysis of cortical and medullary R2-star (M-R2-star) and ADC values revealed that BOLD-MRI was highly effective in identifying early renal alterations, whereas DWI did not distinguish between diabetic and control groups. The absence of significant differences in ADC values suggests that microstructural changes detectable by DWI occur at more advanced stages of DKD, when IF is already present. This supports the view that DWI is not a reliable early marker, but rather a late indicator of structural damage. Receiver operating characteristic analysis showed that M-R2-star and the medullary-to-cortical ratio (MCR) were strong predictors of DKD, with optimal cutoff values (M-R2-star > 30.31/second; MCR > 1.61 for mild and > 1.69 for severe DKD) achieving 100% sensitivity and high specificity. Cortical R2-star also predicted moderate-to-severe DKD with good sensitivity. Overall, BOLD-MRI outperformed DWI, demonstrating strong potential as a non-invasive biomarker for detecting early hypoxic changes and predicting the progression of DKD.

The data showed that BOLD has greater potential in the early stages due to its ability to detect functional changes, such as hypoxia, before anatomical damage occurs. Although DWI does not discriminate in the early stages, it remains useful at a more advanced stage to characterize the progression of fibrosis. The combined use of the two techniques could therefore allow a more complete temporal assessment of the pathophysiological continuum of DKD.

Recently an interesting study investigated the diagnostic performance of a combined protocol including T1 mapping, DWI-DTI and DWI diffusion kurtosis imaging in DKD. On 63 subjects from 3 groups with normal renal function, early DKD and advanced DKD the protocol combining the 3 parameters was able to distinguish between patients with normal renal function and early DKD patients with an AUC of 0.982 and sensitivity and specificity of respectively 95% and 81.8%[124].

Despite this potential, the routine integration of MRI biomarkers into clinical practice requires further large-scale prospective studies, standardization of protocols, inter-centre validation, and demonstration of cost-effectiveness[125] (Figure 1).

Figure 1 Multiparametric magnetic resonance imaging framework in diabetic kidney disease. ADC: Apparent diffusion coefficient; ASL: Arterial spin labeling; MRI: Magnetic resonance imaging; CKD: Chronic kidney disease; eGFR: Estimated glomerular filtration rate.

CLINICAL INTEGRATION: A PRAGMATIC FRAMEWORK FOR IMAGING SELECTION IN DKD

The selection of imaging modalities in DKD requires deliberate clinical integration rather than an exhaustive enumeration of available technologies. The expanding repertoire of advanced imaging techniques has increased technical capability but has not proportionally clarified their role in routine care. Consequently, the central clinical question is not which imaging studies can be performed, but which should be performed, necessitating careful consideration of diagnostic yield, resource utilization, and the likelihood that imaging findings will meaningfully influence management decisions.

Ultrasonography remains the first-line imaging modality in DKD because it adequately addresses the primary structural questions encountered in most clinical settings, including kidney size, parenchymal appearance, and exclusion of obstructive pathology. Typical sonographic features, such as reduced renal length or cortical thickness and increased echogenicity, are consistent with chronic parenchymal disease and assist in assessing chronicity, although they are not disease-specific and offer limited incremental prognostic information beyond established clinical markers such as eGFR trajectory and albuminuria.

Multiparametric renal MRI has demonstrated the capacity to provide quantitative assessments of renal perfusion, oxygenation, and tissue microstructure, offering insights into pathophysiological processes that are not captured by conventional imaging modalities. Systematic reviews and expert consensus documents have highlighted its potential to non-invasively characterize fibrosis and functional heterogeneity in DKD and other CKD[108,113]. However, despite promising longitudinal associations with disease progression[121], multiparametric MRI remains largely confined to research and early translational settings. This limitation reflects the absence of validated, outcome-driven thresholds that link imaging biomarkers to specific therapeutic interventions, as well as persistent challenges related to inter-scanner variability, acquisition standardization, and analytical harmonization across centers[108,125].

Advanced imaging is clearly warranted in selected clinical scenarios where findings may directly alter management. An unexpectedly rapid decline in kidney function should prompt evaluation for superimposed and potentially reversible pathology, such as obstructive uropathy, nephrolithiasis, or renovascular disease, where cross-sectional imaging may be informative. Atypical clinical presentations raising suspicion for non-DKD may require imaging to assess renal anatomy and procedural feasibility prior to biopsy. In advanced DKD approaching kidney replacement therapy, imaging priorities increasingly include vascular assessment to support access planning.

It must be recognized that at the moment this framework reflects expert consensus rather than formal evidence-based guidelines, therefore the limited comparative effectiveness research evaluating imaging strategies in DKD must be acknowledged. To date, no randomized trials have demonstrated improved clinical outcomes attributable to any specific imaging approach, and observational studies linking imaging biomarkers to prognosis rarely address the methodological challenges inherent in treatment stratification based on those same findings. Accordingly, the guiding principle remains imaging with intent: Studies should be obtained when they address specific diagnostic questions whose resolution is likely to influence clinical decisions. Broader clinical adoption of advanced imaging techniques will require both prospective evidence that imaging derived information leads to management changes that translate into measurable improvements in patient outcomes, a benchmark not yet met in routine DKD assessment along with standardization of pathological thresholds across different manufacturers and worldwide[125] (Table 2).

Table 2 Practical imaging approach in diabetic kidney disease.

Clinical scenario	First-line imaging (clinical practice)	Key parameters to assess	Clinical decision impact	Advanced techniques (research/selected cases)	When to consider advanced imaging
Initial DKD assessment (new diagnosis of diabetes + albuminuria or declining eGFR)	B-mode US + color doppler	Kidney length (9-12 cm normal); cortical thickness (> 7 mm normal); echogenicity (grade 0-III); RI (< 0.70 normal); CMD preservation	Exclude alternative diagnoses (obstruction, masses, cysts); establish baseline renal size; identify vascular abnormalities	DWI-MRI, BOLD-MRI	Atypical presentation, rapid progression, young patient, consideration for clinical trial enrollment
Routine monitoring (stable DKD, predictable progression)	B-mode US (annually or when clinically indicated)	Progressive cortical thinning; increasing echogenicity; loss of CMD; changes in kidney size	Confirm expected progression; detect unexpected changes requiring further investigation	None typically indicated	N/A
Rapid/unexplained eGFR decline (> 5 mL/minute/1.73 m2/year or > 25% drop)	US + doppler. If inconclusive: Non-contrast CT	Hydronephrosis; renal artery stenosis (RI asymmetry > 0.05); stones, masses; perinephric abnormalities	Identify treatable causes: Obstruction, RAS, AKI superimposed on CKD	CT angiography (if RAS suspected and intervention considered), MRI with DWI (if AIN suspected)	High clinical suspicion for renovascular disease or acute interstitial nephritis; surgical/interventional planning needed
Discordant clinical findings (albuminuria without retinopathy, hematuria, rapid onset, age < 40 years)	US; consider nephrology referral + kidney biopsy	Assess for non-diabetic kidney disease: Asymmetric kidneys; cortical scarring; masses	Imaging alone insufficient; biopsy often required for definitive diagnosis	Multiparametric MRI (if biopsy contraindicated or to assess fibrosis burden pre-biopsy)	Biopsy contraindicated (bleeding risk, anticoagulation), need to quantify fibrosis non-invasively
Pre-dialysis planning (stage 4-5 CKD, eGFR < 30 mL/minute/1.73 m2)	US (bilateral kidney assessment + vascular mapping for access)	Kidney size (often < 9 cm); severe echogenicity; vascular anatomy (cephalic/basilic veins); arterial inflow	Dialysis modality selection, vascular access planning (AVF vs AVG)	Venography or US elastography (vessel compliance assessment)	Complex vascular anatomy, previous failed access, central venous stenosis suspected
Post-transplant DKD surveillance (transplant recipients with recurrent DKD)	US + doppler of allograft	Allograft size; cortical thickness; RI; exclude hydronephrosis/collections	Monitor allograft function, exclude rejection vs recurrent DKD	Protocol allograft biopsy (gold standard); DWI-MRI or elastography (if biopsy contraindicated)	Rising creatinine, new proteinuria, need to distinguish rejection from recurrent disease
Research/clinical trial setting (characterizing early DKD, treatment response monitoring)	Multiparametric MRI (DWI, BOLD, T1 mapping, ASL) or CEUS	ADC values; cortical/medullary R2; T1 relaxation times; perfusion parameters; tissue stiffness	Quantify subclinical damage, monitor response to novel therapies (SGLT2i, GLP-1 RA, endothelin antagonists)	Photon-counting CT, MR elastography, AI-based texture analysis	Specialized research centers, clinical trials, biomarker development studies
Complications (suspected papillary necrosis, emphysematous pyelonephritis, infected cyst)	Non-contrast CT (first-line for complications)	Gas in renal parenchyma; filling defects; perinephric stranding; abscess formation	Urgent intervention (drainage, antibiotics, possible nephrectomy)	Contrast-enhanced CT (only if eGFR > 30 and potential surgical benefit outweighs risk)	Severe infection requiring source control, surgical planning

This table provides practical, expert opinion-based recommendations for imaging selection in diabetic kidney disease (DKD) across common clinical scenarios. First-line imaging reflects widely available techniques appropriate for routine clinical practice in most centers. Advanced techniques are generally reserved for research applications, clinical trial enrollment, or highly selected cases where conventional imaging is inadequate or contraindicated. Clinical decision impact describes how imaging findings influence patient management. The framework emphasizes the central role of ultrasonography as the primary imaging modality for DKD, with judicious use of advanced techniques when clinical circumstances warrant additional investigation beyond standard care. Recommendations are based on expert consensus, acknowledging that local availability, expertise, and healthcare system resources will influence individual practice patterns. DKD: Diabetic kidney disease; eGFR: Estimated glomerular filtration rate; US: Ultrasonography; CMD: Corticomedullary differentiation; BOLD: Blood oxygen level-dependent; MRI: Magnetic resonance imaging; DWI: Diffusion-weighted imaging; N/A: Not applicable; RAS: Renal artery stenosis; CT: Computed tomography; RI: Resistive index; AKI: Acute kidney injury; CKD: Chronic kidney disease; AIN: Acute interstitial nephritis; AVF: Arteriovenous fistula; AVG: Arteriovenous graft; ASL: Arterial spin labeling; CEUS: Contrast-enhanced ultrasound; ADC: Apparent diffusion coefficient; GLP-1 RA: Glucagon-like peptide-1 receptor agonist; SGLT2i: Sodium-glucose cotransporter-2 inhibitor; AI: Artificial intelligence.

CONCLUSION

Renal imaging in DKD is evolving rapidly, driven by the urgent need to overcome the limitations of traditional biochemical markers and to enable earlier, more precise disease detection. Today, physicians have access to a wide range of imaging modalities, each providing unique insights into the structural, hemodynamic, and microstructural changes that characterize DKD progression. While single-modality approaches have shown promising results, heterogeneity in protocols, processing, and validation standards has slowed widespread clinical adoption. International collaborative initiatives, such as PARENCHIMA and physicians’ societies, have attempted to standardize protocols and establish reproducible biomarkers. These standardization efforts are essential not only for ensuring consistency but also for facilitating the large-scale adoption necessary to demonstrate clinical utility. Looking to the future, the integration of AI and machine learning algorithms holds significant promise for transforming DKD imaging. Radiomics, combined with deep learning models, can reveal subtle patterns and imaging signatures that are not detectable by visual analysis, potentially enabling automated disease phenotyping, risk stratification, and prediction of therapeutic response. As demonstrated in preliminary studies, AI-driven approaches can improve diagnostic performance while identifying novel imaging biomarkers. However, clinical translation to AI-based tools still requires rigorous external validation. In conclusion, advanced imaging techniques offer unprecedented opportunities to improve the detection, characterization, and monitoring of DKD. Ongoing development and validation of these techniques, together with the emergence of radiomic signatures and molecular imaging approaches, are indispensable tools in the global effort to reduce the burden of DKD and its progression to end-stage renal disease.