Diabetic kidney disease (DKD) is a microvascular complication caused by diabetes, which occurs in more than 40% of diabetic patients and is the leading reason for kidney failure worldwide (1). The prevalence of DKD is increasing due to the lack of early diagnosis and effective interventions (2). Renal biopsy is the gold standard for diagnosing kidney disease. Still, this method has limitations, such as being invasive and having sampling errors, so it is only used for a small percentage of patients with unclear diagnoses. Most clinical guidelines use the results of blood and urine biochemical markers as the preferred method for diagnosing and evaluating DKD (3). However, there are some inaccuracies in these blood and urine biochemical markers from the several clinical phenotypes that have been confirmed (4). This requires new, reproducible, non-invasive biomarkers to assess renal pathology in DKD.

Functional magnetic resonance imaging (fMRI) has shown great potential in assessing renal pathology in DKD. It is a non-invasive way to obtain more kidney structure and function biomarkers without exogenous contrast agents (5). Compared to kidney biopsy and serum or urine bioinformatic markers, fMRI has unique advantages (6).

Nowadays, the main fMRIs that have been used for structural or functional assessment of the kidney in diabetic kidney disease include magnetic resonance elastography (MRE-MRI), arterial spin labeling imaging (ASL-MRI), blood oxygen level-dependent imaging (BOLD-MRI), dilated weighted tensor imaging (DWI-MRI). MRE-MRI is a method to assess the degree of kidney elasticity and fibrosis by obtaining changes in tissue stiffness through shear waves generated by external vibrations with different conduction speeds in tissues of varying stiffness (7); ASL-MRI uses blood in the body as a contrast agent to track water protons in the blood and obtain tissue perfusion as a way to determine the progression and prognosis of diabetic kidney disease (8); BOLD-MRI shares similarities with ASL-MRI, except that BOLD-MRI uses deoxyhemoglobin to assess tissue oxygenation levels and assess renal hypoxia (9); DWI-MRI is an imaging that uses the diffusion of water molecules in vivo without contrast injection to assess microscopic changes in the kidney by mapping the movement of water within the tissue (10), Meanwhile, based on DWI-MRI theory, DTI-MRI and IVIM-MRI were born to address the deficiencies of DWI-MRI in microscopic orientation, distribution, diffusion and microperfusion effects of water molecules, and they enriched the acquisition of bioinformation markers (11). We summarize the advantages, disadvantages, and main parameters of fMRIs, which are currently used to assess DKD. Detailed information can be found in Table 1 .

The use of fMRI to assess renal structural and functional alterations in DKD is not yet widely available in the clinic, and relevant systematic reviews are limited. To investigate the value of functional magnetic resonance imaging for assessing renal structure and function in diabetic kidney disease, we performed this meta-analysis of several fMRIs.

This Meta-Analysis was guided and performed by the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines (12) and Meta-analysis Of Observational Studies in Epidemiology(MOOSE) (13, 14). A prospective protocol was developed and registered with PROSPERO (https://www.crd.york.ac.uk) under (ID: CRD42023409249). Informed consent was obtained from all authors for all included studies, and ethics committee approval was not required for further evaluation of published articles.

Nowadays, more and more studies have shown many different phenotypes for the progression of DKD. For phenotypes other than the classical phenotype, using eGFR with proteinuria to assess DKD no longer seems appropriate (4). Although renal puncture biopsy can avoid such issues, it is not absolute. In the early stages of DKD, atypical biopsy findings may be present. In contrast, fMRI is cheaper, less painful, and more acceptable to patients than a renal biopsy and can avoid the risk of sampling errors. While compared to serum and urine biomarkers, patients’ results are less subject to external influences, with more biomarker information and a more comprehensive and detailed assessment of the kidney. Despite its drawbacks, fMRI remains a promising non-invasive, contrast agent-free method of detecting kidney morphology or function.

Many studies have shown that fMRI has unique advantages in detecting and differentiating chronic kidney disease (CKD). Zeng et al. (45) found that renal cortical RBF, f, D, and D* parameters were significantly and positively correlated with eGFR in patients with chronic kidney disease using ASL and IVIM, and the f values even exceeded eGFR in distinguishing CKD from HV on the areas under the curve (AUC). Qin et al. (46) performed a meta-analysis on the effectiveness of fMRI for identifying early CKD. It concluded that BOLD, DTI, and IVIM could distinguish the early CKD population from the general population and that DWI has an advantage in diagnosing all stages of CKD. Niu et al. (47) summarized studies using DWI to assess various stages of CKD and concluded that DWI could be used for early diagnosis and staging of CKD. DKD is a type of CKD that is heterogeneous from other types of CKD, so a meta-analysis is necessary to address the value of multiple fMRIs in assessing renal injury in DKD.

This study shows promising results for ASL in distinguishing DKD from healthy people and patients with early and moderately advanced DKD. This suggests that the RBF values are negatively correlated with the course of DKD in patients. However, it does not seem to distinguish diabetic patients from those with diabetic kidney disease. Such a result may be related to the number of included literature or the fact that high glucose toxicity at the beginning of diabetic disease already mediates multiple homologous pathways causing glomerular basement membrane thickening (48). Unfortunately, we could not search to find studies in this area. The R2* values in the meta-analysis of BOLD did not show its usefulness for assessing DKD. Previous studies have shown that BOLD is of limited importance in responding to long-term renal fibrosis (49). Michaely et al. (50) also confirmed that R2* did not differ significantly in different stages of CKD, including DKD. We make the speculation that arterio-venous oxygen shunting facilitates the dynamic regulation of renal oxygenation (51), making renal hypoxia less pronounced in the course of chronic kidney disease. Therefore, BOLD may help diagnose acute kidney injury but is insignificant in assessing chronic kidney disease.

The role of DTI-DWI in the assessment of DKD disease is satisfactory. We found a significant difference in renal cortical ADC values between DKD patients and healthy volunteers. However, the medullary ADC values cannot distinguish DKD from HV nor determine DKD in microalbuminuria from simple DM. We also found lower ADC values in early DKD compared to mid to late-stage DKD. And the FA values seemed to complement the ADC values because the FA parameters differed significantly between DKD and HV patients in the renal medulla. However, it cannot distinguish between the FA values in patients with DKD and simple DM or early DKD and intermediate to late DKD. The physiology of the kidney determines such a result, as the renal tubules and collecting ducts enter the renal pelvis in a radial pattern (52). The main difference between DTI and DWI is the introduction of three-dimensional space. DTI can respond to the axial or longitudinal diffusion rate of water molecules, so DTI is more significant in responding to the renal medulla. In contrast, DWI can be susceptible to blood perfusion (53). Compared to the renal medulla, the renal cortex requires more blood perfusion to perform the physiological function of filtering blood. Therefore, DWI and DTI must complement each other to assess the extent of renal parenchymal injury in DKD.

IVIM-DWI uses a bi-exponential model to enhance DWI’s perception of capillary perfusion and diffusion, making it more accurate in assessing renal function or microstructure (54). Compared to the D value, the f and D^* values are limited in determining DKD. We found significant differences in cortical, medullary, or parenchyma D values between the DKD and HV groups and differences in D values between the DKD and simple DM groups. In contrast, the f values were significantly different only in comparing the DKD and simple diabetes groups. No significant differences in the D^* values were found in several group comparisons. The results of this study are consistent with the study of Ren et al. (31). The difference in f-values between the DKD group and the simple DM group demonstrated their sensitivity in assessing pathological changes in DKD, which was also confirmed in the study by Qin et al. (46). The D* and f values reflect the microcirculatory perfusion and the D value demonstrates the movement of water molecules (55), so we speculate that such results are related to the included studies being all for early DKD. Although there is microcirculatory damage in early DKD, the reliable compensatory mechanism of the kidney allows for faster blood flow, evidenced by the altered hemodynamics in DKD, thus compensating for the lack of f and D* values. Schurek et al. (56) showed no effect of these pathological alterations on the diffusion of water molecules, which does not agree with our findings, as we found statistical differences in the distribution of water molecules between early DKD and healthy populations. Therefore, we would like more studies to focus on the differences in IVIM parameters between early DKD and mid to late-stage DKD as a way to verify our conjecture.

In the course of DKD, renal hemodynamic changes, hypoxia, and even glomerulosclerosis or interstitial fibrosis (4), can be reflected in functional magnetic resonance imaging parameters. With so many types of fMRI, the pathological changes they reflect have their preferences. ASL and BOLD were mainly used to assess renal perfusion; DWI, DTI, and IVIM were used to evaluate the diffusion of water molecules in the kidney. Therefore, Makvandi et al. ( 35) combined multiple fMRI analyses, which seems more accurate than a single fMRI for assessing pathological changes in DKD. Such an idea has also been promoted by Mehmet et al. (53). We believe combining multiple fMRI test parameters is necessary to assess the renal structure and function and establish uniform and standardized test criteria. In addition, uniform standards for parameter ranges need to be installed, as is the case for blood and urine biochemical markers. Only when these consistent criteria are established it will be possible to quantify the assessment of structural or functional changes in the kidney by fMRI.

In addition to the several types of fMRI included in this paper, there are several other functional magnetic resonance imaging. For example, magnetic resonance spectroscopy imaging (MRS-MRI) and magnetic susceptibility weighted imaging (SWI-MRI). MRS-MRI is mainly used to determine the concentration of compounds and metabolites in tissues, and is currently primarily used in brain tissue-related studies and has not been widely used in diabetic kidney disease (57); SWI-MRI is a further development of the BOLD-MRI technique, which uses the different magnetic susceptibility of tissues to perform imaging by identifying high concentrations of deoxygenated hemoglobin in the veins in contrast to surrounding tissues and is more suitable for the identification of hemorrhagic disorders. However, the use of SWI-DWI is controversial owing to the unique physiological characteristics of the renal body (58). Due to the limitations of the number of included studies, a meta-analysis of these types of fMRI could not be performed. More clinical studies are needed in the future to enrich our conclusions.

This study has the following shortcomings: 1. The included literature is only in English, which may have a particular bias; 2. Different measurement methods, MRI scanners, field strengths, and standards for determining ROIs heterogeneity cannot be ruled out; 3. All the included studies were not diagnosed by renal puncture, so there was a deviation in the diagnosis of DKD; 4. We did not collect data on how DKD compares to other CKD, which is essential for making a differential diagnosis clinically. To address these shortcomings, we hope they can be remedied in the subsequent study.

Functional MRI is not necessary for the diagnosis of DKD. Therefore, fMRI has not been widely promoted in the field of DKD, which may be due to its high price or the lack of uniform standards. Meanwhile, MRI requires more imaging time compared to X-rays. Any slight movement can have an impact on the quality of the image. It is also a problem for the same patient to have significantly different renal fMRI findings due to differences in diet or medications taken. These issues have been raised in the consensus (59, 60), but more clinical studies are needed to address them. Any new science and technology need to develop and grow with continuous exploration. Although functional magnetic resonance imaging is not mature, we believe that fMRI to evaluate DKD or other renal diseases can be widely respected and used.

In conclusion, ASL and DWI on parameters can better distinguish DKD patients and healthy people. The parameters RBF and ADC have certain values in distinguishing DKD staging. In the future, we hope that more investigators will see the value of fMRI in the assessment of DKD so that this contrast-free, noninvasive test will be more involved in clinical decision-making and prognosis determination.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

ZZ designed the meta-analysis. ZZ, YC, and SL selected the eligible articles. ZZ and SL abstracted the data. ZZ and JY analyzed the data. ZZ and YC wrote the paper. ZZ, YC, XZ, SL; JY interpreted the results; and all authors approved submitting the final manuscript. All authors contributed to the article and approved the submitted version.