Deliverables from Metabolomics in Kidney Disease: Adenine, New Insights, and Implication for Clinical Decision-Making Free

A significant burden on healthcare systems worldwide is imposed by kidney diseases, which affect millions of individuals [1, 2]. These diseases encompass a wide spectrum of conditions, ranging from chronic kidney disease (CKD) [3], acute kidney injury (AKI) [4], and diabetic kidney disease (DKD) [5] as well as various glomerular and tubular disorders. Early diagnosis of these diseases and managing them effectively can significantly improve patient outcomes and reduce the socioeconomic impact of kidney disease. According to data from the National Kidney Foundation (NKF) in 2018, a total of 785,883 Americans were diagnosed with kidney failure, necessitating either dialysis or a kidney transplant for survival [6]. Among this group, 554,038 patients underwent dialysis treatment, while an additional 229,887 patients received kidney transplants. Unfortunately, individuals with kidney disease face a sobering reality as they are five to ten times more likely to succumb prematurely rather than experience dialysis or transplantation [7, 8]. In 2018, more than 100,000 people lost their lives to kidney disease despite ongoing efforts to address this critical health issue [3, 9, 10]. CKD has become a prominent and pressing issue in the 21st century, emerging as a leading cause of both mortality and suffering. This alarming trend can be attributed, at least in part, to the increase in risk factors such as obesity, hypertension, and diabetes mellitus (DM) [11].

While advancements have led to a decline in mortality rates for individuals with end-stage kidney disease (ESKD), the Global Burden of Disease (GBD) studies have revealed that CKD has now taken center stage as a major contributor to worldwide mortality [12, 13]. Consequently, it is crucial to identify, monitor, and treat CKD effectively. Identifying the disease in the early stages in transition from AKI to CKD, known as acute kidney disease (AKD), is critical as survivors of post-AKI incidents face an increased lifetime risk of mortality [14‒16]. One of the most common types of CKD is DKD, which results from hyperglycemia and leads to gradual changes in the function of the kidneys. The initial stages of DKD typically exhibit no noticeable clinical symptoms. It is often only when albuminuria is detected, often 10 years after the onset of diabetes, that the kidney damage is already advanced, resulting in a swift decline in kidney function toward ESKD [17]. However, if DKD is identified and addressed in its early stages, its progression can be halted or significantly delayed. A significant challenge with current diagnostic methods is their inability to detect ongoing disease during the silent phase of diabetic nephropathy (DN) as well as in overt disease which is not characterized by albuminuria. Developing innovative biomarkers capable of predicting progression and reflecting biologically significant pathways in DKD has the potential to improve the management of individuals with diabetes [5, 18]. A recent example is the identification of adenine as a key marker of kidney fibrosis for DKD [19].

Adenine, a pivotal purine nucleobase, plays a crucial role in cellular functions, including DNA/RNA synthesis, ATP-mediated energy processes, extracellular signaling, and enzyme cofactors. Under normal physiological concentrations, adenine undergoes absorption and metabolism via adenine phosphoribosyl transferase (APRT), resulting in the formation of AMP and various derivatives. The resultant metabolite xanthine is then converted to uric acid, which is excreted by the kidneys. However, when adenine accumulates with APRT deficiency, adenine follows an alternative pathway catalyzed by xanthine oxidase, leading to the formation of 2,8-dihydroxyadenine. Due to its limited solubility under normal urine pH conditions, 2,8-dihydroxyadenine may precipitate as crystals, presenting a potential risk of renal injury [20]. The kidneys play a pivotal role in maintaining purine homeostasis, and disruptions in this process may contribute to renal dysfunction [21]. Emerging evidence indicates that exposure to adenine also stimulates oxidative stress in the kidneys and may be responsible for cellular toxicity [22]. Exploring specific kidney diseases associated with adenine metabolism or its derivatives is of particular interest. Studies focused on adenine and its associated metabolic pathways may provide insights into conditions where adenine plays a role in the development of nephrolithiasis, glomerular disorders, or tubulointerstitial diseases [23]. Recent studies using untargeted metabolomics, targeted metabolomics, and spatial metabolomics have indicated that endogenous adenine plays a critical role in common kidney diseases, such as DKD. Apart from DN, adenine is also involved in familial kidney diseases such as APRT deficiency, leading to nephrolithiasis and kidney failure, tubulointerstitial diseases, and glomerular disorders. Plasma levels and RBC levels of adenine have been found to be increased with ESKD. Exogenous adenine is also one of the most widely accepted models of experimental CKD. According to our experimental results, the best reservoir for adenine is urine apart from the tissues, with low levels in blood [24]. Adenine itself does not exert a direct influence on the acid-base status of a patient. Nonetheless, the metabolism of adenine and associated compounds can impact the acid-base balance through its involvement in diverse metabolic processes such as purine metabolism, uric acid production, diabetic ketoacidosis, acid-base regulation, etc. [20]. Such investigations contribute to a more nuanced understanding of the intricate relationship between metabolites and kidney health.

This review offers a thorough analysis of a dynamic and swiftly evolving discipline that bridges metabolomics and nephrology. It encompasses discussions on the analytical methodologies employed in metabolomics, the spectrum of kidney diseases under scrutiny, the clinical utility of metabolomics in kidney disease contexts, the mechanistic insights gleaned through metabolomic approaches, as well as the challenges faced and potential future avenues in this exciting and fast-evolving field.

Kidney diseases encompass a spectrum of conditions impacting kidney structure and function [25]. CKD, a global health concern, exhibits varying stages based on estimated glomerular filtration rate (eGFR), ultimately progressing to ESKD [3, 12, 13]. Lack of awareness, as early CKD stages often remain asymptomatic, contributes to high morbidity and mortality. CKD also amplifies the mortality associated with other conditions like cardiovascular diseases, type 2 diabetes (T2DM), hypertension, and infections such as HIV, malaria, and COVID-19 [3, 11, 25, 26]. AKI, common in hospitalized patients, involves sudden kidney function decline due to infections, medications, or dehydration. DN stems from uncontrolled diabetes, leading to proteinuria and kidney function decline [4, 27, 28]. Polycystic kidney disease is a rare genetic condition forming fluid-filled cysts in the kidneys [29]. Nephrotic syndrome indicates kidney damage and features excessive protein in urine, swelling, and high cholesterol [30]. Hypertensive nephropathy results from uncontrolled high blood pressure, gradually impairing kidney function [31]. Awareness, early detection through health checkups, and lifestyle modifications play crucial roles in managing kidney diseases and improving overall well-being [11, 32‒34]. In many cases of kidney disease, there is often advanced disease which is difficult to arrest, and therefore, it is critical to identify the disease in its early, silent stages.

The diagnosis of kidney diseases relies on a range of diagnostic tools and markers, including serum creatinine and blood urea nitrogen measurements, as well as urinalysis, to assess kidney health and detect abnormalities [35, 36]. The glomerular filtration rate serves as a precise measurement of kidney function, aiding in accurate diagnosis and staging [18, 37‒39]. Various imaging techniques such as ultrasound, CT scan, MRI, and matrix-assisted laser desorption/ionization MS imaging (MALDI-MSI) are instrumental in visualizing kidney structure and identifying structural irregularities [18, 40, 41]. Treatment challenges for kidney diseases are diverse and depend on the underlying causes. Common treatment approaches involve managing blood pressure and blood sugar levels, dietary adjustments, and medications. In severe cases, dialysis or kidney transplantation may be necessary. Emerging therapies, including gene therapies and regenerative medicine, show promise but will require identifying patients at early stages of the disease. Early detection and preventive measures remain pivotal in effectively managing kidney diseases, emphasizing the importance of new biomarkers indicating relevant risk factors for precision intervention to improve outcomes [42].

Metabolomics plays a pivotal role in unraveling the intricate biochemical pathways associated with kidney diseases. Metabolomics can provide a comprehensive analysis of small molecules or metabolites within biological samples, such as urine and blood, to gain insights into the metabolic changes that occur during the development and progression of kidney diseases [25‒30, 43] and serve as biomarkers that are associated with different etiologies of kidney diseases. By analyzing the pattern of metabolites, researchers can gain insights into the altered biochemical pathways in the kidneys with insights often not possible from reliance on transcriptomics and proteomics alone. Metabolomics helps in understanding how metabolic processes are disrupted in kidney diseases, shedding light on the underlying metabolic pathways. Metabolomics allows for a personalized approach to treatment by identifying patient-specific metabolic profiles, which can be used to tailor treatment strategies and monitor responses to therapy, [28]. It can reveal how potential medications affect metabolic pathways and help in the design of more targeted and effective therapies. This information can guide treatment decisions and interventions. Metabolomics aids in understanding the impact of dietary choices on kidney health by informing dietary recommendations and interventions for individuals with kidney diseases. Metabolomics can also assist in assessing an individual’s risk of developing kidney diseases based on their metabolic profile and genetic predisposition using approaches such as metabolite-based genome-wide association study [44]. Early identification of at-risk individuals can lead to preventive measures [19]. Overall, metabolomics offers a comprehensive and powerful approach to unravel the biochemical and genetic basis of kidney diseases. As our understanding of metabolomics continues to grow, so does our ability to combat kidney diseases effectively.

Analytical techniques in metabolomics include nuclear magnetic resonance (NMR), GC, high-performance liquid chromatography, inductively coupled plasma atomic emission spectroscopy, inductively coupled plasma mass spectrometry, LC-MS, and GC-MS. Both NMR (¹H and ¹³C) and mass spectrometry (MS) (LC-MS and GC-MS) are essential techniques in metabolomics. However, NMR faces a significant drawback in sensitivity compared to MS, making it challenging to detect or gather information about low-abundance metabolites in biological fluids [45]. MS serves as a pivotal tool in metabolomics research, providing heightened sensitivity through both GC-MS and LC-MS. However, GC-MS has drawbacks, including the necessity for readily volatile samples and occasional chemical derivatization, which can take at least 1–2 days [46]. Additionally, it is limited to analyzing relatively small molecular masses (m/z, <800) and is unsuitable for medium polar compounds. In contrast, LC-MS offers distinct advantages, requiring minimal sample preparation and enabling analysis of diverse metabolite classes that cannot be separated or fragmented by GC-MS due to their higher molecular mass or polarity. Successful metabolomics often relies on LC-MS combined with soft ionization techniques like MALDI [47, 48] and electrospray ionization (ESI) [49, 50]. These interfaces coupled to mass analyzers play a vital role in MS, sorting ions by their mass-to-charge ratios (m/z). Types of mass spec separation and detection approaches include time of flight, triple quadrupole, ion trap, and orbitrap. Time of flight utilizes an electric field to accelerate ions from the source to the detector based on travel time, distinguishing ions by size. Orbitraps are widely used in proteomics and metabolomics across diverse fields. It uses electrostatic trapping and frequency detection to determine ion m/z. Ions from a sample are trapped in an electrostatic field, undergo harmonic oscillations, and their frequency directly correlates with m/z. This enables accurate mass determination. Orbitrap instruments offer high mass resolution, crucial for separating and identifying metabolites, reducing false positives in metabolomics. They are highly sensitive, detecting metabolites at low concentrations, crucial for studying diverse metabolites in biological samples. With a wide dynamic range, orbitraps simultaneously detect abundant and trace-level metabolites. Their rapid data acquisition makes them suitable for efficient high-throughput metabolomics studies [51]. LC-MS with multiple-reaction monitoring (MRM) is a key MS/MS technique for precise measurement of low-concentration metabolites in metabolomics [43, 45, 50, 52]. The typical workflow for the targeted and untargeted metabolomics approaches is shown in Figure 1 [53].

The metabolomics approach is an emerging and highly prominent technological frontier. This methodology entails meticulous scrutiny of metabolites, encompassing both endogenous and exogenous varieties (with a molecular mass <1,500 Da), derived from specific biological fluids and has emerged as a powerful approach to unraveling the complex metabolic alterations associated with kidney diseases [54]. Metabolomics can be categorized into two primary approaches: targeted and untargeted metabolomics. Targeted metabolomics focuses on the identification and quantification of a limited set of metabolites, using standards and calibration curves, associated with specific pathways or biochemical processes [52]. In contrast, untargeted metabolomics aims to capture as many diverse metabolite species as possible, followed by candidate annotation using various databases such as ChemSpider, METLIN, Human Metabolome Database (HMDB), MassBank, mzCloud, GNPS, Lipid Blast, and others. This approach proves invaluable for the relative quantification of metabolites and enables the prediction of metabolic changes within the studied groups [45]. Several analytical platforms are accessible for the study of metabolomics, with NMR spectroscopy (¹H and ¹³C NMR) and MS (LC-MS and GC-MS) emerging as the primary techniques for such investigations. Each of these techniques offers distinct advantages, particularly in terms of sensitivity, selectivity, and throughput, when addressing the complexities of various sample types such as urine, blood plasma/serum, tissue, cell extracts, or media [47].

In recent years, spatial metabolomics has received much recognition as a transformative approach for understanding complex diseases and interpreting heterogenous pathology in tissue section. MALDI-MSI technology has gained prominence as a cutting-edge analytical tool in the field of spatial metabolomics, enabling researchers to dissect the intricate molecular mechanisms involved in development of specific pathologic features in diseased kidneys [55]. This field is dedicated to the detection and interpretation of metabolites within their precise spatial context. Notably, the adoption of MALDI-MSI has played a pivotal and widely embraced role in transforming this approach [43]. The schematic approach of the MALDI-MSI is shown in Figure 2. Other platforms of spatial metabolomics include desorption ESI-MSI (DESI-MSI), nano-droplet extraction surface analysis (Nano-DESI); DESI-MSI enables direct surface analysis without extensive sample preparation. It utilizes charged solvent droplets to desorb and ionize molecules on the surface, with subsequent MS providing spatially resolved information on compound distribution. Nano-DESI, operating at micro and nanoscales, utilizes tiny droplets for high-resolution analysis of biological samples. Valuable in metabolomics, pharmacology, and environmental science, Nano-DESI allows detailed mapping of molecules in complex biological systems at the microscale [56].

Recent research endeavors have aimed to identify differentially expressed metabolites in AKD through an array of metabolomic methodologies and bio samples. A decade ago, a study utilized the LC-MS technique with Sprague-Dawley rats, subjecting them to bilateral renal artery clamping for 45 min, with or without L-carnitine pretreatment. Postischemic animals showed peaking serum concentrations of creatinine and blood urea nitrogen 24 h after reperfusion, which was reduced by L-carnitine. The authors attributed lipid metabolism changes as critical factors in ischemia/reperfusion kidney damage [57]. Another study by Wei et al. [58] employed GC-MS and LC-MS for a murine ischemic AKI metabolomics study, identifying 404 substances from tissue samples and 293 metabolites from plasma. Renal ischemia notably decreased plasma levels of several substances, including betaine, tyrosine, glutamine, and others, revealing significant alterations in glucose, lipid, and purine metabolism, as well as osmotic regulation and prostaglandin modulation. These findings highlighted the potential connection between hypoperfusion/ischemia and the key elements of ischemic AKI, including tubulopathy, inflammation, and (micro) vasculopathy [58]. Elmariah et al. [59] conducted metabolomic profiling using LC-MS on a cohort of 44 patients, including 22 with a history of CKD. Their study was cross-referenced with the Framingham Heart Study, encompassing 2,164 individuals. This investigation revealed alterations in plasma concentrations of certain metabolites, ultimately identifying S-5-adenosylhomocysteine as strongly correlated with serum creatinine and serving as a reliable marker for AKI. A metabolomics investigation utilizing LC-MS/MS was carried out on 42 kidney transplant patients, of whom 30 developed AKI [60]. Findings from this study indicated an adverse impact on tryptophan and arginine metabolism, with the concentration of these two amino acids being lower in transplant patients compared to that in healthy volunteers. Furthermore, in the comparison between AKI patients and patients without AKI, the levels of tryptophan and arginine were lower in AKI patients than in non-AKI patients. In 2021, a study utilized LC-MS/MS to investigate targeted metabolic profiling in cisplatin-induced AKI, focusing on tryptophan metabolism alterations [61]. Targeting 29 metabolites, the study found that indoxyl sulfate accumulated in a dose-dependent manner. Additional experiments with chlormethiazole, a CYP2E1 inhibitor, demonstrated the functional significance of indoxyl sulfate, showing that reducing hepatic indoxyl sulfate synthesis attenuated cisplatin-induced AKI.

Another recent study by Chen et al. [62] focused on 30 AKI patients and 20 healthy controls, proposing that 2-S-glutathionyl glutathione acetate, 5-l-glutamyl-taurine, and l-phosphoarginine were involved in altered cytochrome P450, arginine, and proline metabolism, serving as potential diagnostic markers for AKI. A separate urine metabolomic study conducted by Tian et al. [63] of in post-cardiac surgery subjects (n = 159) found that 55 patients experienced AKI, while the kidney functions of the remaining subjects were unaffected. The authors suggested that tyrosyl-gamma-glutamate, deoxycholic acid glycine conjugate, 5-acetylamino-6-amino-3-methyluracil, arginyl-arginine, and L-methionine could act as predictive factors for AKI in cardiac surgery patients [63]. To gain a deeper understanding of AKI, various experimental models have been tested, including rats, mice, murine, and pigs. In addition to these models, diverse metabolic analyses conducted on AKI patients revealed a metabolomic differentiation between patients with and without AKI. Notably, symmetric dimethylarginine (SDMA) and tryptophan, or the 5-HIAA/5-HT ratio, were identified as potential AKI risk predictors (AUC >0.9). These findings, along with various other studies [64‒69] exploring the role of metabolomics in AKI diagnosis and risk prediction, are summarized in Table 1.

DKD is a specific type of kidney disease that occurs in individuals with diabetes and is a leading cause of CKD. Managing diabetes effectively is essential in preventing or slowing the progression of both DKD and CKD. Metabolomics has proven to be an invaluable tool in comprehending the pathophysiology of CKD, which significantly contributes to morbidity and mortality [5, 19, 27, 47]. From a metabolomics perspective, DM is considered a quintessential metabolic disease due to the numerous metabolic variations linked with it. Recent studies have focused on identifying differentially expressed metabolites in DKD through various metabolomic approaches and biological samples, revealing several metabolites as potential biomarkers for predicting the onset of diabetes. Notably, separate studies by Wang et al. [70], Newgard et al. [71], and Guasch‐Ferre et al. [72] have indicated the involvement of isoleucine, leucine, and valine in the development of T2DM. Additionally, a comprehensive study by Merino et al. [73] involving 1,150 individuals with normal fasting glucose in a 20‐year follow‐up highlighted the correlation between increased plasma taurine concentration and decreased glycine with the onset of T2DM. A MS metabolomics investigation conducted a decade ago by Hirayama et al. [74] on 78 subjects at various stages of DKD revealed increased levels of aspartic acid, citrulline, SDMA, and kynurenine in the DKD group. Conversely, Zhang et al. [75] reported reduced plasma concentrations of several amino acids, including isoleucine, leucine, valine, and phenylalanine, in individuals with T2DM. The observed contradictions in amino acid metabolism across studies might be attributed to variations in drug treatments and interventions among individuals with T2DM.

Our group quantified 94 urine metabolites in cohorts of patients with DM and CKD (DM with DKD; n = 61), in patients with type 1 DM without CKD (n = 32), type 2 DM without CKD (n = 41), and in healthy controls (n = 23) by using a targeted GC-MS technique [27]. The study revealed that 13 metabolites which included 3-hydroxy isovalerate, aconitic acid, citric acid, 2-ethyl 3-OH propionate, glycolic acid, homovanillic acid, 3-hydroxy isobutyrate, 2-methyl acetoacetate, 3-methyl adipic acid, 3-methyl crotonyl glycine, 3-hydroxy propionate, tiglylglycine, and uracil were statistically related to eGFR or albuminuria (albumin-to-creatinine ratio), and these metabolites levels were consistently reduced in the urine of DM with CKD. These metabolites play key roles in processes such as the Krebs cycle and lipid synthesis [76, 77]. Notably, 12 of these metabolites remained significant in DM with CKD compared to DM without CKD. Additional studies by van der Kloet et al. [78], a FinnDiane study, corroborated these observations, indicating a decrease in hippuric acid levels in the cohort with progressive disease. Additionally, there was an elevation in S-(3-oxododecanoyl) cysteamine and acylcarnitines in this group.

The urine metabolomic analysis also suggested a decline in mitochondrial function, confirmed by urine exosome analysis and immunohistochemistry [27], indicating a reduction in mitochondrial content. This decrease in content could be associated with lower levels of the coactivator PGC1a, a crucial component in mitochondrial biogenesis. Corresponding studies corroborated these observations, highlighting alterations in specific metabolites during the progression of DKD [79, 80]. Overall, this research emphasizes the potential of urine metabolomics as a noninvasive technique for detecting kidney biomarkers and quantitating kidney health. These findings not only shed light on the diminished mitochondrial activity in DKD but also offer the possibility of discovering new therapeutic targets and innovative biomarkers for evaluating kidney function in diabetes. A recent investigation by Parmar et al. [47] explored the ADMA/SDMA ratio (ASR) in urine across various stages of diabetes, including normal glucose tolerance, newly diagnosed diabetes, macroalbuminuria (MAC), and diabetic microalbuminuria. The results indicated a decrease in the ASR in the microalbuminuria and MAC groups compared to the newly diagnosed diabetes and normal glucose tolerance subjects, suggesting its potential diagnostic value for the early detection of DN among Asian Indians. In conjunction with these studies, several other research works [81‒91] have identified different biomarkers using urine, plasma, and tissue, aiding in the early detection of DKD through diverse metabolic pathways using the metabolomics approach tabulated in Table 2.

Apart from DKD, a comprehensive analysis of CKD progression remains incomplete. Several studies have utilized metabolomics to uncover new insights for ESKD and advanced CKD. A recent study by Dahabiyeh et al. [92] aimed to gain fresh insights into perturbed biochemical reactions and identify distinct biomarkers between ESKD and CKD. The study, employing an isotope-labeled LC-MS approach on ESKD patients (n = 13) and CKD patients (n = 13), revealed significant alterations in 193 metabolites between ESRD and CKD. These metabolites were primarily involved in aminoacyl-tRNA biosynthesis, branched-chain amino acid biosynthesis, taurine metabolism, kynurenine, and tryptophan metabolism. Notably, the study highlighted the upregulation of three kynurenine derivatives, 2-aminobenzoic acid, xanthurenic acid, and hydroxy picolinic acid, in ESRD compared to CKD, owing to the significant decrease in the glomerular filtration rate with CKD progression to ESRD. Mitrofanova et al. [93] reviewed the lipid pathogenesis in CKD conditions, discussing the clinical and experimental evidence of pathogenic lipid droplet accumulation in kidney parenchyma and the molecular mechanisms by which cholesterol, fatty acids, triglycerides, and lipid droplet accumulation contribute to CKD progression. The authors also focused on current and emerging therapeutic strategies for treating and preventing lipid-induced nephrotoxicity. Numerous studies have been conducted employing metabolomics in models of CKD, identifying promising metabolites that could serve as biomarkers [94‒96]. Mor et al. [97] recently conducted a review on the kynurenine pathway in CKD. Primarily, the aromatic amino acid tryptophan undergoes extensive metabolism through various kynurenine-related pathways, resulting in the generation of numerous biologically active compounds such as 3-hydroxyanthranilic acid, 3-hydroxykynurenine, anthranilic acid, kynurenine, KYNA, kynurenic acid, quinolinic acid, tryptophan, and xanthurenic acid. These metabolites may contribute to the development of systemic disorders that accompany the progression of CKD [97]. A recent study led by Hu et al. [98] highlighted trimethylamine N-oxide (TMAO) as a promising marker for potentially decelerating the progression of CKD. The study suggested that controlling TMAO production through inhibition using 3,3-dimethyl-1-butanol or disruption of gut microbiota function with an antibiotic cocktail reduced renal injury in a murine CKD model. This study underscores the potential of metabolomics, suggesting TMAO production as a novel strategy to mitigate the progression of renal injury in CKD.

Patients with advanced kidney disease like CKD often require dialysis or kidney transplantation. Hemodialysis (HD) plays a pivotal role in renal replacement therapy for ESKD patients, and understanding the HD process, efficacy, and patterns through plasma metabolomics has significant potential. In a study by Ragi et al. [49], the behavior of uremic toxins, including indoxyl sulfate, p-cresol sulfate, phenyl sulfate, catechol sulfate, and guaiacol sulfate, along with creatinine and urea, was evaluated in the pre-dialysis (n = 90), post-dialysis (n = 90), and healthy control (n = 74) groups. The study revealed the dominance of p-cresol sulfate and indoxyl sulfate in the pre-dialysis group. Moreover, the study demonstrated that the concentrations of phenyl sulfate, catechol sulfate, and guaiacol sulfate were approximately 50% of that of indoxyl sulfate. The dialytic clearance of indoxyl sulfate and p-cresol sulfate was found to be less than 35%, while that of the other three sulfates was 50–58%. The concentration levels of these targeted uremic toxins could be used to assess kidney function and the efficacy of HD [49]. A 2023 study led by Pallerla et al. [50] evaluated the levels of amino acids and other related metabolites in ESKD patients using LC-MS/MS and GC-MS techniques. The study developed analytical methods for 29 targeted metabolites, revealing dialytic losses ranging from 52% to 45% for arginine, lysine, and histidine and 38–26% for glycine, cysteine, proline, alanine, threonine, glutamine, valine, and methionine. In contrast, the dialytic loss was low (≤12%) for aspartic acid, glutamic acid, asparagine, leucine, tyrosine, tryptophan, and isoleucine (Table 3) [50]. A recent untargeted metabolomics investigation focused on plasma samples obtained from individuals with a single kidney, non-dialysis CKD, ESKD, and a control group of healthy individuals [99]. The study revealed that more than 400 features in the plasma of non-dialysis CKD and ESKD patients were significantly different from those in the control group. In contrast, individuals with a single kidney exhibited alterations in less than 35 features. The metabolites N,N,N-trimethyl-L-alanyl-L-proline betaine (TMAP, AUROC = 0.815) and pyrocatechol sulfate (AUROC = 0.888) demonstrated superior accuracy compared to creatinine (AUROC = 0.745) in identifying individuals with a single kidney. Furthermore, when comparing pre- and post-dialysis samples, TMAP emerged as the most robust biomarker for dialytic clearance (AUROC = 0.993). The key finding of this study is the identification of TMAP as a potential novel plasma biomarker, indicating reduced kidney function in early CKD, ESKD, and hemodialytic clearance [99]. Moreover, the study highlighted pronounced metabolic differences between transplantation and dialysis therapy, emphasizing the differences of nocturnal intermittent peritoneal dialysis in dialytic metabolite clearance compared to conventional HD as nocturnal intermittent peritoneal dialysis relies predominantly on the passive diffusion of solutes, which may lead to reduced metabolite clearance. Consequently, employing multiple metabolites could offer a more comprehensive evaluation of dialysis, assisting clinicians in selecting an optimal modality for their patients.

Metabolomics has significantly contributed to our understanding of the dynamic changes in renal metabolism associated with kidney diseases, revealing shifts in metabolite profiles that can act as biomarkers for the early detection and monitoring of these conditions. However, the spatial resolution and capability of traditional metabolomic techniques to capture metabolite distribution within specific kidney regions and cell types are not possible. By identifying which structures or cell types within the kidney experience disturbances in metabolic pathways, we can target therapeutics more effectively and reduce off-target effects. To create a detailed spatial map of kidney metabolites, MALDI-MSI has emerged as a powerful tool, enabling researchers to uncover precise alterations in metabolites at a spatial level, identify potential therapeutic targets, and advance our comprehension of the pathophysiological processes involved in kidney diseases [100]. In recent studies [19] involving the Chronic Renal Insufficiency Cohort (CRIC; including n = 558 with normal albuminuria or microalbuminuria and n = 341 with MAC) [101], the Singapore Study of Macro-Angiopathy and Reactivity in Type 2 Diabetes (SMART2D; n = 309), and an American Indian cohort (n = 54), advanced Zip-Chip-MS and MALDI-MSI (spatial metabolomics) techniques were employed to analyze kidney tissues and urine samples. The study demonstrated that the urine adenine/creatinine ratio (UAdCR) measurement was closely linked to the progression of DKD in the American Indian, CRIC, and SMART2D cohorts with T2DM without MAC. Notably, the UAdCR emerged as an independent predictor of ESKD and all-cause mortality. The distribution (localization) of adenine in kidney tissue from healthy control subjects (shown in Fig. 3a) and diabetic kidney tissue (shown in Fig. 3b) was performed using MALDI-MSI, and the comparative analyses between these two study groups are shown in Figure 3c. These findings demonstrated that adenine is co-localized with regions of tubulointerstitial and vascular pathology. The study also demonstrated that inhibition of adenine is beneficial in the mouse model of DKD. Overall, the data indicate the promising clinical utility of the UAdCR as a potential precision biomarker for identifying high-risk patients, despite lack of proteinuria and potential targeted therapeutics for these high risk patients. Adenine has now been demonstrated to be increased in DKD in both patients with T1D and T2D, thus establishing itself as a major cause of kidney disease. The measurement of adenine in the urine appears to reflect kidney production of adenine and is therefore clinically relevant. The authors have found that high levels of urine adenine identify patients at high risk of kidney failure across many ethnicities and geographic regions and detection of adenine in kidney tissue is meaningful during clinical examinations, especially with the spatial metabolomics as it provides valuable insights into the localization and concentration of adenine within the kidney tissue. Investigating into the involvement of adenine in kidney diseases holds the potential to enhance our comprehension of renal pathophysiology. This exploration offers the opportunity to uncover molecular mechanisms, identify biomarkers, and introduce novel paths for therapeutic interventions. The outcomes of the research into adenine metabolism carry the promise of refining diagnostic precision, enhancing treatment effectiveness and ultimately improving overall patient outcomes within the field of kidney diseases.

A recent (spatial metabolomics) study using apolipoprotein E knockout mice, diabetes was induced through streptozotocin injections, while the control group received citrate injections, and both were subjected to a cholesterol-enriched diet. This model consistently shows elevated plasma cholesterol levels compared to wild-type mice (∼400 mg/dL vs. ∼80 mg/dL), regardless of diet. Interestingly, the cholesterol-enriched diet is not expected to significantly impact mouse cholesterol levels in their setup. However, authors studied whether elevated cholesterol may influence phosphatidylinositol (PI) metabolic changes in the proximal tubular PT_S3 segment, given the known interaction between PI lipids and cholesterol homeostasis. The crosstalk between PI lipid and cholesterol pathways, linked through acetyl-CoA, suggests potential effects on both cholesterol and PI lipid species due to high glucose levels in the diabetic setting. Metabolic histology analysis using MALDI-MSI highlighted the sensitivity of the PT_S3 segment to diabetes-induced changes, suggesting its role as an accelerating factor in DKD [102].

An important study by Wang et al. [103] pioneered an elegant application of MALDI-MSI for high-spatial resolution metabolomics coupled with isotope tracing, to address the intricate challenge of cell-to-cell heterogeneity during ischemia in complex tissues like the kidney. This innovative approach facilitates the precise mapping of cell-type-specific dynamic changes in central carbon metabolism, even within heterogeneous tissue architectures. Through strategic integration with multiplexed immunofluorescence staining, the authors successfully discerned metabolic changes in endothelial and proximal tubular cell types within a renal ischemia-reperfusion injury model. Notably, this methodology revealed region-specific metabolic perturbations during both the injury and recovery phases, uncovering unexpected anomalies in ostensibly normal cells. The authors aptly demonstrated the power of this approach in unraveling the complexities of kidney homeostasis, illustrating its broader potential for investigating tissue homeostasis across diverse organs [103].

While metabolomics plays a crucial role in understanding biological processes, its capacity to provide a comprehensive description of these processes is limited. The evolving field generates a wealth of data that must be integrated and analyzed alongside other omics data for complete interpretation. Combining metabolomic analysis with bioinformatics yields crucial insights into complex diseases. Common statistical analyses of metabolomics data include univariate methods like fold change analysis, t tests, and volcano plots for two-group data, one-way ANOVA, post hoc analysis, and correlation analysis for multigroup data. Multivariate approaches, including multiple variance analysis, multiple regression analysis, principal component analysis, partial least squares discriminant analysis, cluster analysis, and machine learning methods such as random forest and Support Vector Machine, allow the evaluation of input metabolomic datasets and identification of metabolites undergoing abnormal changes. Subsequent data mining methods help distinguish functionally relevant metabolites [104]. The selection of a suitable p value for ranking significantly expressed metabolites is crucial for determining a reliable threshold [105]. Metabolomics analyses rely on powerful software tools such as MetaboAnalyst 5.0, MS-DIAL 4.0, MetFlow, METLIN, MetaSpace, GraphPad, and Lilikoi 2.0, among others, which aid in establishing an effective metabolomics analysis pipeline for researchers [106].

Over the past decade, significant progress has been made in metabolomics, broadening the detection of various metabolites. However, when compared to transcriptomics and proteomics, which offer near-complete coverage, metabolomics currently captures only a small portion of these numerous metabolites. The chemical diversity of metabolites and their wide-ranging levels within cells add complexity to metabolite measurements. The current literature does not support the possibility of extracting and measuring all metabolites through a single extraction and analytical method. Therefore, researchers often employ various extraction techniques and combinations of analytical approaches to enhance metabolite coverage. Given the diversity in the goals and methodologies of metabolite analyses, it is crucial to establish guidelines for acquiring and reporting metabolite data as there are numerous potential sources of error and misinterpretation. Hence, there is a pressing need to develop guidelines to ensure the robustness of obtained and reported metabolite data. These guidelines should encompass sampling, extraction, storage, metabolite identification, handling large sample numbers, and recommendations for reporting metabolite identification methods and the levels of certainty in quantification. Notably, Fernie et al. [107] have made recommendations for reporting metabolite data, which they suggest should be a standard practice for future manuscript submissions. These recommendations cover both LC-MS and GC-MS data, enhancing the quality of metabolite data reporting and contributing to community efforts to improve metabolite annotation, ultimately benefiting the field of metabolomics.

Metabolomics stands at the forefront of precision medicine, merging omics and clinical data with epidemiological and environmental factors to enable precise diagnoses and personalized interventions tailored to each patient. In the context of kidney disease management, biomarkers are indispensable for improving survival rates and preventing the onset of kidney disease. Several large consortiums, including the Kidney Precision Medicine Project (KPMP) and the Human Biomolecular Atlas Program (HubMap), integrate spatial and bulk metabolomics as critical platforms in interrogating human tissue samples [108, 109]. However, a major hurdle for metabolite identification persists in the scarcity of high-quality spectral data across diverse instruments and collision energies. Identifying unidentified mass spectral features remains a major challenge in the field of metabolomics. Recent progress in instrument technology, experimental methodologies, and data analysis softwares such as PubChem, ChemSpider, the HMDB, BioCyc, and KEGG has alleviated many of the challenges associated with MS-based metabolomics. To enhance the confidence of compound identification from MS-based datasets, future strategies should focus on techniques that eliminate unwanted MS features, expand MS databases and data curation, automate metabolite identification, and standardize analytical methods across various platforms. Development in terms of instrumentation and data analysis programs is essential to tackle the intricate biological complexities. Anticipated advancements are expected in single-cell metabolomics and MSI to gain insights into the pathophysiology of kidney diseases. Recognizing the heterogeneous nature of the kidney tissue and the diverse subcellular processes within it, MS imaging has gained prominence for revealing the spatial distribution of specific metabolites in kidney tissues. Various imaging techniques, including MALDI, nanostructure imaging MS, DESI-MS, and secondary-ion MS, contribute to this field. Notably, MALDI-MSI has emerged as a powerful and sophisticated method with high sensitivity and throughput, enabling the localization of a broad spectrum of metabolites within specific regions of tissue samples [19, 55]. Recent groundbreaking discoveries using the MALDI-MSI approach [19, 110, 111] have accelerated the clinical application of metabolomics. The recent study to demonstrate that the UAdCR measurement closely correlates with DKD progression has provided compelling evidence to add the UAdCR to be a new independent predictor of ESKD and mortality [112], adding to conventional markers of albuminuria and baseline eGFR [19]. While MALDI-MSI has been instrumental in enhancing our understanding of various diseases, there remains a need to confirm the spatial confirmation of metabolites through further MALDI-MS/MS analysis. These evolving perspectives are poised to significantly contribute to the understanding and management of various diseases, including different types of kidney diseases.

The review article covers several of metabolomics and the accelerated advances in the field, especially in the last 2–3 years. Spatial metabolomics has provided dramatic insights by linking metabolites to disease features, and the kidney biopsy has been an excellent example of how to incorporate several omics datasets for clinical implications. The uAdCR shows promise as a new disease causing diagnostic marker. The identification and examination of adenine and related metabolites is expected to enhance our understanding of kidney disease pathophysiology and offers encouraging prospects for better prognosis and early diagnosis. Current limitations of bulk and spatial metabolomics include incomplete and nonoverlapping metabolite coverage across various methods, the necessity for structural confirmation, and the introduction of variations during sample preparation and data analysis. Advancements in high-resolution imaging and atmospheric pressure imaging are on the horizon. The anticipated use of MSI holds the promise of a deeper understanding of kidney pathology as it enables the visualization and localization of metabolites within specific regions of tissue samples.

K.S. is a member of the DSMB for Cara Therapeutics and has had research support from Boehringer-Ingelheim. He holds equity in SygnaMap.

This work was supported by NIDDK UO1, VA Merit, and Department of Defense grants.

N.R. contributed to the design, prepared the initial draft, and provided final approval for the published version. K.S. contributed to the conception, conducted critical revisions, and gave the final approval for the published version.