Introduction

Globally, chronic kidney disease (CKD) is a significant clinical and public health issue [1]. Chronic kidney disease is a significant risk factor not only for end-stage renal disease (ESRD), but also for cardiovascular disease [2] and higher death rates [3]. Because there is no effective therapy for ESRD, identifying the early risk factors for CKD is crucial [4].

One of the major risk factors for CKD is obesity [5] and obesity-related metabolic disorders such as diabetes, hypertension, and metabolic syndrome [6]. However, there is disagreement regarding the contribution of obesity to renal insufficiency [7], and it remains unclear whether metabolic abnormalities associated with obesity precede obesity itself. The majority of studies indicate that the glomerular filtration rate (GFR) is decreased in obese individuals [8, 9], although not all studies support this finding [10, 11]. There is a possibility that these discrepancies due to the heterogeneity of obesity phenotypes [12]. In fact, much of the increased risk for CKD among individuals with obesity is thought to stem from underlying cardiometabolic abnormalities associated with excess adiposity [13]. Nevertheless, not every obese individual will experience the metabolic problems associated with obesity. The variability in obesity outcomes is produced when obesity is combined with distinct metabolic characteristics. Consequently, six subtypes of the population can be identified: metabolically healthy with normal weight (MHNW), metabolically healthy overweight (MHOW), metabolically healthy obese (MHO), metabolically unhealthy with normal weight (MUNW), metabolically unhealthy overweight (MUOW), and metabolically unhealthy obese (MUO) [14].

Currently, various studies yielded contradictory effects of obesity-metabolic subphenotypes on the risk of developing CKD. Additionally, data from these isolated investigations might not provide sufficient evidence to support a potential differential risk for CKD. Therefore, it is essential to synthesize the findings of these related investigations. Here, we conducted a systematic review and meta-analysis of observational studies to determine the associations of different obesity phenotypes with the risk of CKD.

Methods

Search strategy

We conducted a systematic review of observational studies examining the associations between various obesity phenotypes and the risk of chronic kidney disease (CKD). Our literature search was exhaustive, covering databases such as PubMed, Scopus, EMBASE, Web of Science, and Google Scholar up to February 2024, without restrictions on date or language. Additionally, we performed an in-depth bibliographic analysis of relevant publications to identify further studies.

The search terms utilized included “kidney failure”, “chronic renal insufficiency”, “chronic kidney disease”, “CKD”, “chronic renal disease”, “obesity”, “body mass index”, “overweight”, “normal weight”, “metabolically healthy”, and “metabolically unhealthy” (see Supplementary 1). Only articles published in English were included in the final selection. The study selection process is illustrated in Fig. 1. The systematic review was in line with the guidelines outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement [15]. Pre-registration code of the systematic review and meta-analysis protocols in PROSPERO is 564,164.

Fig. 1
figure 1

PRISMA flow diagram for selection process of the studies

Full size image

Eligibility criteria

Studies were considered suitable for meta-analysis if they met the following criteria: (1) utilized a cross-sectional or cohort design; (2) examined participants with both metabolically healthy and unhealthy obesity phenotypes, with kidney dysfunction as the outcome; (3) categorized subjects based on body mass index (BMI); (4) provided hazard ratios (HRs) or odds ratios (ORs) for kidney dysfunction incidents; and (5) disclosed the criteria for categorizing subjects as metabolically healthy or unhealthy. Clinical trials, literature reviews, animal studies, genetic research, and studies that did not report renal dysfunction occurrences in the subgroups were excluded from evaluation.

Study selection

After removing duplicates, two authors (BA and FA) independently evaluated the titles and abstracts obtained from the initial search. Both authors (BA and FA) assessed the full-text articles to ensure they met the eligible inclusion and exclusion criteria. Any disagreements were reevaluated by a third author (ARA).

Data extraction and quality assessment

We extracted information about the first author, year of publication, country, study design, population, the number and age of participants, follow-up duration, definitions of obesity phenotypes, definition of kidney dysfunction, and confounding factors that were adjusted for the analysis. Only the fully adjusted effect estimates reported in the studies were extracted and subsequently included in the meta-analysis.

In this systematic review and meta-analysis, the quality of nonrandomized studies was assessed using the Newcastle–Ottawa Scale (NOS) [16]. The NOS evaluates studies based on eight items divided into three categories: selection, comparability, and exposure. Each item offers several response options, with the highest quality studies receiving a maximum of one star per item. An exception is made for the comparability category, where up to two stars can be awarded. Consequently, the NOS operates as a nine-star rating system [17]. Table 1 details the quality assessment and data extraction processes utilized in this study.

Table 1 Characteristics of the studies included in the meta-analysis
Full size table

Statistical analysis

All statistical analyses were performed using Stata, version 17.0 (Stata Crop, College Station, TX, USA). The effect size was pooled using the fixed method meta-analysis (inverse variance). If significant heterogeneity was present, the random-method (DerSimonian–Laird) was employed to pool effect sizes. The reference category was considered an MHNW group in all studies.

The I2 index and Cochrane’s Q test were used to assess the heterogeneity between studies. The I2 interpretation is as follow: low if I2 < 30%, moderate if I2 = 30–75%, and high if I2 > 75%. Subgroup analyses were performed according to study design, and definition of kidney disease. Meta-regression was employed to investigate the potential confounding effects of variables, such as follow-up duration, age and sex on the association. Publication bias was investigated using Begg’s rank correlation and Egger’s linear regression.

Results

Search results

Following a preliminary search using relevant keywords, we identified 4528 full-text articles. After eliminating duplicates and applying our inclusion and exclusion criteria through a two-phase review process (title and abstract review), we included 13 suitable observational studies (four cross-sectional and nine cohort studies) in our evaluation. Figure 1 provides an overview of the main study findings and details the procedure used for selecting relevant studies. This meta-analysis included 13 articles and 492,829 participants in total [18,19,20,21,22,23,24,25,26,27,28,29,30] (Table 1).

Study characteristics

The various features of the included studies are compiled in Table 1. The sample sizes varied from 2324 to 319,674 participants. Research was conducted in China [25, 30], Taiwan [21], Korea [20, 22, 24, 26, 27], Iran [18, 28], USA [19, 23], and Japan [29]. The follow-up period for cohort studies ranged from 3.2 to 14 years. BMI was employed in all trials to diagnose overweight and obesity. Both sexes were included in every study.

Metabolic health was defined using the National Cholesterol Education Program’s Adult Treatment Panel (NCEP ATPIII) criteria [19,20,21,22,23,24,25, 27, 28] and the International Diabetes Federation (IDF) criteria [18, 29, 30]. One study [26] defined metabolic health as having a HOMA-IR < 2.5 and the absence of any metabolic syndrome components. The results of the included studies were adjusted for common confounders, including age, sex, physical activity, smoking status, alcohol use, socioeconomic factors, liver enzymes, inflammatory factors, lipid profile, and glucose profile. Comprehensive details on the studies included in the meta-analysis are provided in Table 1.

Metabolically unhealthy normal-weight (MUNW) phenotype and KD risk

We identified 12 relevant studies (encompassing 13 effect sizes) examining the relationship between the metabolically unhealthy normal weight (MUNW) phenotype and the risk of chronic kidney disease (CKD), totaling 430,580 participants. These studies reported various effect measures, such as odds ratios (OR) and hazard ratios (HR). Our systematic literature search yielded an overall effect estimate (OR/HR) of 1.58 with a corresponding 95% confidence interval (CI) of 1.43–1.76, regardless of the specific effect measure used. This indicates that individuals with the MUNW phenotype have a 58% higher risk of CKD compared to those with the metabolically healthy normal weight (MHNW) phenotype. A considerable variability was observed (I2 = 51.33%, P = 0.02). Detailed information can be found in Table 2 and Fig. 2. There was no evidence for small-study bias on the basis of the visual inspection of the funnel plot, Egger’s regression test, and Begg’s test in all studies together (Egger’s test P = 0.65; Begg’s test P = 0.42), studies reporting OR (Egger’s test P = 0.99; Begg’s test P = 1.00), and HR (Egger’s test P = 0.83; Begg’s test P = 1.00) for the observations concerning MUNW phenotype and KD risk.

Table 2 Overall and subgroup estimates of meta-analysis for the risk of kidney dysfunction among different phenotypes of obesity compared to metabolically healthy normal weight phenotype
Full size table
Fig. 2
figure 2

Overall effect estimate of the association between metabolically unhealthy normal-weight (MUNW) phenotype and risk of kidney dysfunction

Full size image

We employed a stratified strategy, recognizing the distinct characteristics of these effect measures and their potential influence on interpreting associations. To enhance the precision of our findings, we pooled studies separately based on the type of effect measure. In subsequent analyses, we distinguished between studies reporting odds ratios (OR) and those reporting hazard ratios (HR) to address variations in research design and methodology.

The pooled analysis of studies reporting ORs demonstrated a strong positive correlation between the MUNW phenotype and the risk of kidney disease (KD), with a combined effect estimate of 1.66 and a 95% confidence interval (CI) of 1.35–2.03 (Table 2). This finding substantiates that individuals with the MUNW phenotype have a 66% higher risk of KD compared to those with the metabolically healthy normal weight (MHNW) phenotype.

On the other hand, in the subgroup of studies reporting HR, the overall effect estimate (HR) was 1.54 with a 95%CI (1.35–1.75) (Table 2).

Additional subgroup analyses based on research design and the definition of KD further substantiated the significant relationship between the metabolically unhealthy normal weight (MUNW) phenotype and the development of KD within each subgroup (Table 2). These analyses consistently indicated a substantial association between MUNW and increased KD risk, underscoring the robustness of the observed link across various study designs and KD definitions.

Metabolically healthy overweight and metabolically unhealthy overweight phenotypes and KD risk

Three prospective cohort studies [21, 26, 28], encompassing a total of 82,904 participants, investigated the relationship between the metabolically healthy overweight (MHOW) phenotype and the risk of KD. Additionally, two studies [21, 28] involving 20,655 participants examined the risk associated with the metabolically unhealthy overweight (MUHOW) phenotype. Individuals with the MHOW phenotype exhibited a non-significant higher risk of KD events compared to those with the metabolically healthy normal weight (MHNW) phenotype, with an effect size (ES) of 1.34 (95% CI = 0.85–2.12) (Table 2; Fig. 3). There was considerable between-study heterogeneity (I2 = 78.85%, P < 0.01). Visual inspection of the funnel plot and results from Begg’s test indicated no signs of publication bias (Begg’s test P = 1.00).

Fig. 3
figure 3

Overall effect estimate of the association between metabolically healthy overweight (MHOW) phenotype and risk of kidney dysfunction

Full size image

When comparing the risk of KD among MHOW against MHNW phenotype, Egger’s test revealed a substantial publication bias (Egger’s test P = 0.03). Moreover, Table 2 indicates that there was no significant connection between MUOW persons and the risk of CKD (ES = 1.63, 95% CI = 0.97–2.73), evidence of considerable heterogeneity (I2 = 89.47%, P < 0.01), and no indication of publication bias (Egger’s test P = 0.33) when compared to the MHNW phenotype.

Metabolically healthy obese and metabolically unhealthy obese phenotypes and KD risk

The study of KD risk in connection to MHO phenotype comprised a total of 13 observational studies [18,19,20,21,22,23,24,25,26,27,28,29,30] with 492,829 individuals. These studies provided a variety of effect measures, such as odds ratios (OR) and hazard ratios (HR). Furthermore, 12 studies with 430,580 people examined the relationship between the MUHO phenotype and incident KD risk. Upon aggregating all studies, irrespective of the impact measure utilized, the findings indicated that the MHO phenotype was linked to a 20.0% higher risk of CKD in contrast to the MHNW phenotype (ES = 1.20, 95% CI = 1.06–1.34) (Table 2; Fig. 4).

Fig. 4
figure 4

Overall effect estimate of the association between metabolically healthy obese (MHO) phenotype and risk of kidney dysfunction

Full size image

The effect sizes for these associations did not show any significant evidence of heterogeneity between studies (I2 = 34.38%, P = 0.10), and no evidence of bias was found in the combined studies (Egger’s test P = 0.67; Begg’s test P = 0.58), in the studies that reported OR (Egger’s test P = 0.98; Begg’s test P = 0.71), or in the studies that reported HR (Egger’s test P = 0.58; Begg’s test P = 0.70). Comparatively, the MUO participants’ comparable pooled ES was 1.90 (95% CI = 1.63–2.20) (Table 2; Fig. 5). When the degree of the connection was assessed, there was, however, considerable variability in the individual estimates (I2 = 76.82%, P < 0.001), and there was no indication of publication bias across all studies combined (Egger’s test P = 0.31; Begg’s test P = 0.95), studies reporting OR (Egger’s test P = 0.42; Begg’s test P = 1.00), HR (Egger’s test P = 0.91; Begg’s test P = 1.00).

Fig. 5
figure 5

Overall effect estimate of the association between metabolically unhealthy obese (MUO) phenotype and risk of kidney dysfunction

Full size image

A substantial positive association between the MUNW phenotype and KD risk was found in the pooled analysis of studies reporting odds ratios (OR), with the combined effect estimate showing an effect size (ES) of 1.05 (95% CI = 0.86–1.30) for the MUNW phenotype. Additionally, for the metabolically healthy obese (MHO) and metabolically unhealthy obese (MUO) phenotypes, the ES was 2.03 (95% CI = 1.51–2.72) and 1.26 (95% CI = 1.11–1.44) for MHO, and 1.81 (95% CI = 1.50–2.20) for MUO, respectively (Table 2).

The subset of studies reporting hazard ratios (HR) indicated an overall effect estimate (HR) for MHO and MUO as 1.26 (95% CI = 1.11–1.44) and 1.81 (95% CI = 1.50–2.20), respectively (Table 2). No indication of publication bias was found according to Egger’s test (for MHO, P = 0.58; for MUO, P = 0.91) and Begg’s test (for MHO, P = 0.70; for MUO, P = 1.00).

Subgroup analysis

The pooled analysis of metabolic obesity phenotypes and the development of kidney disease (KD) for studies reporting odds ratios (OR) and hazard ratios (HR) revealed significant statistical heterogeneity. Consequently, subgroup analyses were conducted to better understand these relationships.

Subgroup analyses of studies reporting OR indicated a significant relationship between the MUNW phenotype and KD development. This relationship was observed in the cross-sectional subgroup and both subgroups defined by different KD criteria (Table 3). However, no significant relationship was found in other subgroups (Table 3).

Table 3 Subgroup estimates of studies reporting OR for the risk of kidney dysfunction among different phenotypes of obesity compared to metabolically healthy normal weight phenotype
Full size table

For studies reporting HR, subgroup analyses revealed a significant relationship between MUNW and KD development within each subgroup, divided by study design and KD definition (Table 4). These findings suggest that the association between MUNW and KD is robust across various study designs and definitions of KD. Subgroup analyses of studies reporting OR found no significant association between the MHO phenotype and the development of KD (Table 3). However, subgroup analyses of studies reporting HR revealed a significant correlation between the MHO phenotype and KD development within each subgroup, classified by study design and KD definition (Table 4). These findings highlight the importance of considering study design and KD definitions when assessing the relationship between MHO and KD. In subgroup analyses of studies reporting HR, a significant association between the MUO phenotype and the development of kidney disease (KD) was evident within each subgroup categorized by study design, particularly when proteinuria was not included in the definition of KD (Table 4). Similarly, subgroup analyses of studies reporting OR indicated a significant association between the MUO phenotype and KD development. This was observed in the cross-sectional subgroup and both subgroups defined by different KD criteria (Table 3). These results suggest a robust relationship between the MUO phenotype and increased risk of KD across various study designs and definitions.

Table 4 Subgroup estimates of studies reporting HR for the risk of kidney dysfunction among different phenotypes of obesity compared to metabolically healthy normal weight phenotype
Full size table

Meta-regression

Meta-regression analysis revealed that age significantly influenced the association between the MUNW phenotype and the risk of KD (Table 5). This effect was significant only in studies reporting odds ratios (OR) (coefficient = 0.051, SE = 0.015, P value <0.01).

Table 5 Meta-regression for the effect of age, follow-up duration, and sex
Full size table

Furthermore, the risk of incident KD in the comparison of MUO versus MHNW was significantly modified by sex in the overall analysis of OR/HR and specifically in studies reporting OR. However, in other instances, meta-regression did not detect any significant link between obesity phenotypes and incident KD in relation to age, follow-up duration, and percentage of female participants. These findings highlight the importance of considering demographic factors such as age and sex when evaluating the risk of KD associated with different obesity phenotypes.

Discussion

We examined the combined impact of BMI and metabolic health status on the risk of CKD in the current meta-analysis of 13 observational studies involving 492,829 participants. The findings showed that, in contrast to MHNW subjects, overweight and obese individuals had a significantly higher risk for CKD even in the absence of overt metabolic abnormalities. Furthermore, this meta-analysis demonstrated that overweight and obese individuals are at higher risk for CKD regardless of their metabolic status, refuting the idea that being overweight and obese without metabolic abnormalities is a benign condition. Various prospective longitudinal studies reported inconsistent effects of obesity-metabolic subphenotypes on the risk of CKD [31,32,33].

A meta-analysis of nine prospective cohort studies [34] indicated that individuals with metabolic abnormalities are more likely to develop CKD even if they maintain a normal weight. Obese and overweight individuals who are otherwise healthy were also found to be at increased risk of CKD. Another meta-analysis found that a higher body mass index (BMI) is associated with a progressively greater risk of renal disease progression [35].

Additionally, research examining the relationship between renal disease and metabolic syndrome [36] showed that the risk increases proportionally with the number of metabolic syndrome components present. Specifically, the odds ratios (ORs) for CKD increased from 1.39 to 1.96 as the number of components increased from two to five, with rising increments of 0.03, 0.24, and 0.3.

The inconsistent outcomes observed across various studies may be attributed to different classifications of CKD, as many studies did not include proteinuria and kidney structural abnormalities as markers. None of the aforementioned studies applied structural changes in the kidney when defining CKD, and this might have resulted in them missing the early stage of CKD. Another explanation may be related to the different metabolic status of the obesity phenotypes and the lack of consensus for the definition of MHO [21].

Independent of metabolic risk factors, several common biological states link obesity to kidney dysfunction. These include hemodynamic changes, oxidative stress, hormonal effects, and activation of the renin–angiotensin–aldosterone system [37,38,39]. Adipose tissue acts as an active endocrine organ, and dysregulation in the production of adipokines and cytokines derived from adipose tissue, such as leptin, adiponectin, tumor necrosis factor-α, interleukin-6, and plasminogen activator inhibitor-1, may also contribute to the pathogenesis of chronic kidney disease (CKD) in overweight and obese individuals [39,40,41]. Low mitochondrial number and activity in adipose tissue have been suggested as underlying factors in metabolic syndrome (MetS). Recent studies have shown that mitochondrial dysfunction is involved in obesity-related metabolic diseases, evidenced by the downregulation of mitochondrial biogenesis, oxidative metabolic pathways, and oxidative phosphorylation (OXPHOS) proteins [42]. These findings indicate that impaired mitochondrial biogenesis in adipocytes is a potential mechanism contributing to energy imbalance and metabolic homeostasis disturbances in MetS.

It is unclear what processes determine a person’s metabolic state when they have the same body mass index. The primary issue put forth is the distribution pattern of fat, with excess visceral fat having a greater negative impact on metabolic health than excess subcutaneous fat [43]. Furthermore, recent research indicates that behavioral and contextual variables [46], genetic and epigenetic programming [44, 45], and ethnicity may also be relevant factors.

Individuals with a MUNW phenotype exhibit metabolic problems despite having a normal BMI. These disturbances are characterized by low levels of physical activity, low lean body mass, low resting metabolic rate, low insulin sensitivity, and a high body fat percentage, especially visceral fat [47].

The primary strength of this study lies in its relatively large pooled sample size, which enables robust estimates of the relationship between the six BMI–metabolic categories and related diseases, providing precision that individual studies might lack. Additionally, the inclusion and separate analysis of both prospective cohort studies and cross-sectional studies in this meta-analysis offer a comprehensive understanding of the association between obese-metabolic subphenotypes and chronic kidney disease (CKD) risk.

Given the association between the MUNW phenotype and increased risk of CKD and other diseases [48, 49], early identification of MUNW individuals is crucial. These individuals often elude screening because they are not perceived as high risk, yet early detection is important for predicting and preventing renal insufficiency.

However, several limitations of our study should be acknowledged. First, the period of exposure to the present metabolic phenotype and longitudinal changes in metabolic status and BMI were not recorded in the research, which might partially alter the predicted risk. Research has indicated that metabolic wellness in obese individuals may be a temporary state that is becoming more prevalent [49]. Therefore, it is more probable that changes in weight and metabolism will have a greater influence on CKD risk than cross-sectional weight status. Second, none of the included studies provided information on the relative changes in GFR and proteinuria across the follow-up periods. Third, there was a noticeable amount of variability in the analysis between the various phenotypic categories. Another key limitation of our study is the inability to differentiate between individuals with and without diabetes mellitus and arterial hypertension in the included studies. The metabolic health definitions based on the NCEP ATPIII and IDF criteria encompass both groups, preventing us from performing subgroup analyses by diabetes and hypertension status. This limitation may affect our interpretation of the relationship between metabolic health and kidney disease outcomes. Future research should explicitly account for both diabetes and hypertension status to better understand their independent impacts on metabolic health and kidney health.

Conclusion

All metabolically abnormal individuals are at a higher risk for chronic kidney disease (CKD) compared with corresponding healthy individuals, and our current meta-analysis of observational studies supports the idea of heterogeneity in metabolic status among individuals within similar BMI categories. Therefore, assessing both BMI and metabolic status simultaneously is crucial for estimating CKD risk in high-risk populations. Furthermore, effective measures should be implemented to prevent and control obesity and metabolic abnormalities. With the exception of MUOW individuals, significantly elevated risks for incident and prevalent CKD were observed in other obesity-metabolic groups.