GLIM criteria for definition of malnutrition in peritoneal dialysis: a new aspect of nutritional assessment

Abstract

Background: The aim of our study was to evaluate the effectiveness of Global Leadership Initiative on Malnutrition (GLIM) criteria in assessing malnutrition within the peritoneal dialysis (PD) population.Methods: We conducted a retrospective analysis involving 1057 PD patients across multiple institutions, characterized by an age of 56.1 ± 14.4 years, 464 (43.9%) female, and a median follow-up of 45 (25, 68) months. Malnutrition was diagnosed according to GLIM criteria. The endpoint event was overall mortality. The survival rate and hazard ratio (HR) of death between malnutrition and well-nourished were analyzed in all patients and various subgroups. Receiver operator characteristic curve and integrated discrimination improvement (IDI) were used to distinguish the efficacy of the nutritional tools prediction model.Results: According to the GLIM criteria, the prevalence of malnutrition among the study population was 34.9%. The adjusted HR of overall mortality was 2.91 (2.39 − 3.54, p < 0.001) for malnutrition versus well-nourished. In sensitivity analyses, the HR remained robust except the cardiovascular disease subgroup. The area under the curve of GLIM predicting 5-year mortality was 0.65 (0.62–0.68, p < 0.001). As a complex model for forecast the long-term mortality, the performance of adjusted factors combined with GLIM was poorer than combined malnutrition inflammation score (MIS) (IDI >0, p < 0.001), but fitter than combined geriatric nutritional risk index (GNRI) (IDI <0, p < 0.001).Conclusions: The GLIM criteria provide a viable tool for nutritional assessment in patients with PD, and malnutrition defined according to the GLIM can predict prognosis with an acceptable performance.

Keywords:

• GLIM
• malnutrition
• assessment
• peritoneal dialysis
• prognosis

1. Introduction

Malnutrition serves as a potent indicator for predicting significant morbidity and mortality among patients with chronic kidney diseases (CKD) stage 5 who are undergoing peritoneal dialysis (PD) [1,2]. Compared to hemodialysis, PD is associated with increased challenges such as loss of protein through peritoneal effluent, metabolic irregularities due to chronic inflammation, and gastrointestinal dysfunction, all of which contribute to the pronounced risk of protein-energy wasting (PEW) in this patient group [3]. The incidence of PEW among PD patients ranges from 18% to 54%, highlighting its prevalence and established link with increased mortality rates in end-stage renal disease [4]. The Kidney Disease Outcomes Quality Initiative (KDOQI) 2020 clinical practice guideline for nutrition underscore the critical need for proactive assessment and clinical intervention regarding the nutritional status of patients undergoing PD [5].

Currently, most of the common nutritional assessment tools or laboratory biomarkers such as Subjective Global Assessment (SGA), Malnutrition Inflammation Score (MIS), Geriatric Nutritional Risk Index (GNRI), albumin, and normalized protein equivalent of total nitrogen appearance (nPNA) have been demonstrated to be prognostic factors for dialysis mortality [6–8]. Nevertheless, there are individual pros-and-cons in these indicators for screening and planning malnutrition risk. For instance, serum albumin level can be influenced by variables such as dietary protein intake (DPI) and the presence of inflammation. MIS is relatively complex, with a wide range of related items, including subjective evaluations, physical and laboratory indicators, and is usually time-consuming [9,10]. GNRI only includes objective indexes. Therefore, there is no unified consensus on nutrition assessment tools.

In 2018, the committee of the European Society for Clinical Nutrition and Metabolism (ESPEN) proposed an updated standardized global consensus definition of malnutrition in adults – the Global leadership initiative on malnutrition (GLIM), which contains phenotypic and etiologic diagnostic criteria [11]. As an evidence-based nutritional scoring system, the GLIM has been criteria have been validated for assessing the poor prognosis with a range of chronic conditions in both prospective and retrospective cohort studies, as well as in clinical trials [12–14]. Interestingly, its applicability has recently been identified in patients undergoing hemodialysis [15]; however, its validation within PD clinical settings remains pending. Therefore, we analyzed the utilization and validation of GLIM and its predictive value for mortality in PD individuals using a multi-center database, and compared it with several common existing malnutrition assessment approaches.

2. Methods

2.1. Data sources

This study was a retrospective cohort analysis, incorporating multiple Peritoneal Dialysis (PD) registries across China, and predominantly included subjects from the Chinese Han demographic. Employing a standardized data design format and entry protocol, the consortium of database endeavored to elucidate the associations between various clinical factors and poor outcomes such as malnutrition, infection, major adverse cardiovascular events (MACE), and death in the dialysis population. Specifically, this investigation concentrated on evaluating the nutritional status, wherein 1,057 adult individuals were meticulously selected from an initial screening pool of 1433. Over an 11-year span from March 2012 to March 2023, the study cohort comprised patients who had been receiving peritoneal dialysis therapy for a minimum duration of three months. Cases with malignant tumors, acute infections and metabolic disorders (defined as disorders with abnormal metabolism of endocrine hormones such as hyperthyroidism, hypothyroidism, and adrenocortical dysfunction, which usually result in abnormal metabolism or distribution of body protein muscle and fat.) were excluded. Cases missing essential outcomes and nutrition-related data were also excluded. Details of the recruitment process were shown in Figure 1. The protocol was authorized by the local medical research ethics committee, and the informed consent was waived due to the noninvasive and anonymous.

Figure 1. Flow chart of study subjects enrollment.

2.2. Study design

Nutritional status was assessed using GLIM at 2 to 4 weeks after the initiation of PD (the first follow-up after new-onset dialysis), a stage of effective assessment early in the dialysis transition period. For classification purposes, we divided all subjects into malnutrition and well-nourished groups according to GLIM and verified the difference in all-cause mortality. The consistency between GLIM and other nutritional assessment methods in PD (also divided into two categories based on the purpose of analysis) and the predictive efficacy of poor prognosis were analyzed.

2.3. Nutritional status assessment

2.3.1. GLIM

The GLIM was performed in two steps: Firstly, screening a compromised nutritional status. We used several variables of PEW criteria (serum biochemical measures, body weight, muscle mass, and protein intake) [3,16]. Patients at risk were identified if no less one of the following standards was met: Serum albumin <38 g/L, serum prealbumin level <0.3 g/L, serum total cholesterol level < 2.6 mmol/L, body mass index (BMI) <23 kg/m², weight loss >5% in the past 3 months, and nPNA < 1 g/kg/day. In the second step, phenotypic and etiologic criteria for malnutrition were confirmed and graded. The original GLIM criteria included three phenotypic criteria (non-subjective weight loss, low BMI and reduced muscle mass) and two etiologic criteria (reduced intake or digestive malabsorption, inflammation or disease burden). Malnutrition was defined as meeting at least one of the phenotypic and one of the etiologic criteria [11].

Owing to the retrospective nature of this research, accurate measurement of reduced muscle mass was not comprehensively performed. The three phenotypic criteria we used included: a) an unintentional weight loss >5% for 6 months or >10% for beyond 6 months; b) a low BMI (Asian criteria, a BMI of <18.5 kg/m² in younger individuals and <20 kg/m² in people older than 70 years); c) reduced muscle mass [mid-arm muscle circumference calculated (MAMC)<90% of median value]. The dialysis population have a broad background of high catabolic states, intake disorders, micro-inflammation, and uremic burden. We accepted that all participants met the definition of the etiological criteria [15].

2.3.2. SGA

The SGA is a nutritional risk screening tool recommended by the KDOQI guidelines, which includes recent changes in body weight and nutrient intake, gastrointestinal symptoms, edema, degree of fat and muscle wasting, and functional activity. 7-point SGA, a seven points evaluation system, was used in our database. According to the scoring results, they were defined as well-nourished (6–7 points), mild to moderate malnutrition (4–5 points), and severe malnutrition (1–3 points) [6]. In this study, we summarized all cases with less than 6 points as malnutrition.

2.3.3. MIS

MIS was proposed with a revision based on SGA. The content was comprised of patient’s medical history and subjective evaluation of SGA, and BMI, albumin and transferrin laboratory biochemical indicators were added. We identified MIS >8 as malnutrition [7].

2.3.4. GNRI

The formula was as follows: GNRI = 1.489 × serum albumin (g/L) + [41.7 × body weight/ideal body weight]. Ideal body mass was reckoned as BMI =22 kg/m². In this study, GNRI ≤96 was set as malnutrition [8].

2.4. Demographic, clinical and outcome data

Demographic and general data including age, gender, smoking history, diabetes, cardiovascular disease (coronary heart disease, severe arrhythmia, cerebrovascular disease, peripheral arterial disease, congenital heart disease,etc.), pharmacohistory and blood pressure. Anthropometric data was collected at dry weight after the dialysate was released such as: BMI, triceps skinfold thickness (TSF) and mid-arm muscle circumference (MAC). MAMC was estimated by the formula: MAMC (cm) = MAC (cm) − 0.314 × TSF (mm) [16]. Other fasting laboratory data: hemoglobin, serum prealbumin, serum albumin, blood lipids, alanine transaminase, urea, serum creatinine, serum calcium, phosphate, parathyroid hormone (PTH), residual renal function (RRF), CRP. Related indicators of protein nutrition intake was nPNA. Dialysis-related data: peritoneal transport characteristics index (D/Pcr), totals dialysis adequacy index (Kt/Vurea) per week. RRF, D/Pcr and Kt/Vurea were performed by the PD Adquest 2.0 software. The primary follow-up outcome was all-cause mortality.

2.5. Statistical analysis

All continuous variables in this study were tested for normal distribution (Kolmogorov-Smirnov test, p < 0.05 is considered skewed). The standard deviation of the mean was used to indicate normal distribution and the median (interquartile range) was used for skewed distribution. Numbers (frequency) were used to present all counting data. Fisher exact test, t test and Wilcoxon U test were applied for comparison between groups as appropriate. The consistency of different nutrition assessment tools was tested by the kappa test. In addition, we applied Kaplan-Meier curves and log-Rank tests to assess discrepancy in overall survival between subgroups with malnutrition and with well-nourished. Cox regression was adopted to analyze the hazard ratios of malnutrition and total mortality, adjusting for potential covariates. Subgroup interactive analysis was performed to verify robustness in diverse categories. ROC curves, net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were performed to estimate the predictive efficacy of existing evaluation tools. Statistical analysis was implemented with SPSS 25.0 and R software 4.3.1.

3. Results

3.1. Participants characteristics

At baseline, for the 1057 participants, the mean age was 56.1 ± 14.4 years, 464 (43.9%) were female, 276 (26.1%) combined cardiovascular disease, 206 (19.5%) had diabetes, BMI was 22.0 ± 3.4 kg/m², and albumin was 35.0 ± 5.2 g/L. After the first step of screening, 742 (70.2%) were screened as being at risk of malnutrition. Ultimately, 369 (34.9%) patients were identified as malnourished, and 688 (65.1%) were considered to be well-nourished according to GLIM criteria. Compared to well-nourished patients, malnourished patients were older, had a higher prevalence of diabetes and cardiovascular diseases, more severe edema, higher CRP levels, but lower body mass index, serum albumin, prealbumin, transferrin, and nPNA levels. Unsurprisingly, patients classified as malnourished according to the GLIM criteria had higher MIS scores and lower SGA and GNRI scores, indicating poorer nutritional status. Differences in the characteristics of other clinical elements between the categories were shown in Table 1.

Table 1. Baseline characteristics of categories according to GLIM in all participants.

	All patients (n = 1057)	Well-nourished (n = 688)	Malnourished (n = 369)	p-vaule
Age (years)	56.1 ± 14.4	55.0 ± 14.8	58.2 ± 15.1	<0.001
Female, n (%)	464 (43.9)	306 (44.5)	158 (42.8)	0.649
Smoking, n (%)	201 (19.0)	136 (19.8)	65 (17.6)	0.412
Diabetes, n (%)	206 (19.5)	114 (16.6)	92 (24.9)	0.001
Cardiovascular disease, n (%)	276 (26.1)	129 (18.8)	147 (39.8)	<0.001
Systolic pressure (mmHg)	129 ± 38	128 ± 39	130 ± 33	0.403
Diastolic pressure (mmHg)	82 ± 26	80 ± 28	83 ± 22	0.075
Medicines
Antianaemic, n (%)	402 (38.0)	253 (36.8)	149 (40.4)	0.259
Statins, n (%)	280 (26.5)	172 (25.0)	108 (29.3)	0.144
Hypotensive agent, n (%)	873 (82.6)	562 (81.7)	311 (84.3)	0.308
Body mass index (kg/m^2)	22.0 ± 3.4	22.5 ± 4.1	21.6 ± 3.3	<0.001
Triceps skinfold thickness (mm)	15.1 ± 5.8	15.8 ± 6.1	13.9 ± 5.3	<0.001
Mid arm circumference (cm)	28.6 ± 4.2	29.3 ± 4.6	28.4 ± 4.0	0.002
MAMC (cm)	23.9 ± 3.1	24.4 ± 3.3	23.7 ± 3.1	<0.001
Edema				<0.001
None, n (%)	635 (60.1)	495 (71.9)	140 (37.9)
Mild, n (%)	331 (31.3)	164 (23.8)	167 (45.3)
Moderate-severe, n (%)	91 (8.6)	29 (4.2)	62 (16.8)
Hemoglobin (g/L)	86.6 ± 14.2	88.1 ± 15.1	85.7 ± 13.6	<0.001
Alanine aminotransferase (U/L)	22.1 ± 7.6	22.3 ± 8.0	21.9 ± 7.4	0.427
Serum albumin (g/L)	35.0 ± 5.2	35.5 ± 5.3	33.9 ± 5.0	<0.001
Prealbumin level (g/L)	0.3 ± 0.1	0.4 ± 0.1	0.2 ± 0.1	<0.001
Transferrin (g/L)	2.1 ± 0.7	2.2 ± 0.8	1.9 ± 0.6	<0.001
Total cholesterol (mmol/L)	5.42 ± 1.18	5.43 ± 1.21	5.36 ± 1.07	0.348
High-density lipoprotein (mmol/L)	1.24 ± 0.41	1.30 ± 0.45	1.28 ± 0.34	0.455
Low-density lipoprotein (mmol/L)	2.93 ± 1.02	3.01 ± 1.04	2.96 ± 0.99	0.449
Creatinine (umol/L)	769 ± 236	776 ± 248	758 ± 205	0.233
Calcium (mmol/L)	2.01 ± 0.27	1.99 ± 0.27	1.98 ± 0.26	0.561
Phosphate (mmol/L)	1.82 ± 0.56	1.81 ± 0.56	1.83 ± 0.57	0.582
Parathyroid hormone (pg/ml)	200 (101, 326)	195 (99, 327)	214 (111, 336)	0.496
C-reactive protein (mg/L)	0.7 (0.5, 1.2)	0.5 (0.3, 1.0)	1.0 (0.7, 1.6)	<0.001
Residual renal function (ml/min/1.73 m^2)	5.0 (4.1, 8.1)	5.2 (4.2, 8.5)	4.9 (4.1, 7.4)	0.208
D/Pcr	0.66 ± 0.21	0.65 ± 0.23	0.64 ± 0.20	0.481
Kt/Vurea,weekly totals	1.92 ± 0.65	1.93 ± 0.67	1.88 ± 0.52	0.213
nPNA (g/kg/day)	0.9 ± 0.3	1.0 ± 0.3	0.8 ± 0.2	<0.001
SGA	5.7 ± 0.9	6.1 ± 1.0	4.3 ± 0.8	<0.001
MIS	8.5 ± 0.8	7.3 ± 0.8	10.2 ± 0.9	<0.001
GNRI	93.7 ± 10.1	94.7 ± 9.9	91.9 ± 9.7	<0.001

MAMC: mid-arm muscle circumference calculated; D/Pcr: peritoneal transport characteristics index; Kt/Vurea: dialysis adequacy index; nPNA: normalized protein equivalent of total nitrogen appearance; SGA: subjective global assessment; MIS: malnutrition inflammation score; GNRI: geriatric nutritional risk index.

3.2. Diagnostic consistency between GLIM and other nutritional tools

Malnutrition was diagnosed using nutritional tools in 391 (37.0%) SGA, 309 (29.2%) MIS, and 468 (44.3%) GNRI patients. 116 (11.0%) patients were diagnosed simultaneously as malnourished patients by all four composite nutritional indices, while 699 (66.1%) patients were identified by only one of the four tools. On the other hand, 47 (4.4%) malnourished patients were identified by GLIM alone. Further application of the kappa test was utilized to precisely measure the concordance of diagnoses, with GLIM and MIS exhibiting the highest consistency (kappa = 0.412, p < 0.001), followed by SGA (kappa = 0.308, p < 0.001), and GNRI showing the lowest (kappa = 0.339, p < 0.001). The kappa values of other mutual diagnoses are shown in Figure 2.

Figure 2. Conformance between GLIM and other nutrition assessment tools. a) Number and percentage of malnutrition diagnosed by various nutritional assessment tools. b) Intersection of the number of people diagnosed as positive for malnutrition various nutritional assessment tools. c) Concordance index (kappa) of various nutritional assessment tools for the diagnosis of malnutrition. GLIM: Global Leadership Initiative on Malnutrition; SGA: subjective global assessment; MIS: malnutrition inflammation score; GNRI: geriatric nutritional risk index.

3.3. GLIM and mortality

During a median follow-up of 45 (25, 68) months, 441 (41.7%) all-cause deaths were observed. All deaths were categorized by specific causes as follows: 255 (57.8%) deaths from cardiovascular diseases, 124 (28.1%) deaths from infections and 62 (14.1%) from other causes. The analysis of the Kaplan-Meier survival curves revealed that the patients identified as malnourished through various scoring methodologies demonstrably exhibited inferior prognoses compared to their well-nourished counterparts, Figure 3. The Cox regression analysis results indicated a markedly elevated incidence of all-cause mortality among individuals diagnosed with malnutrition, a finding consistent across all applied criteria of assessment. After adjusted factors (age, sex, diseases, medicines, hemoglobin, lipids, calcium, phosphate, parathyroid hormone, RRF, D/Pcr, C-reactive protein and Kt/Vurea) in the ultimate model, the HR of all-cause mortality in GLIM was 2.91 [95% CI (confidence interval): 2.39 − 3.54, p < 0.001], whereas MIS had the highest HR of 3.93 (3.23 − 4.80, p < 0.001), Table 2.

Figure 3. Survival curves of GLIM and other nutrition measurement tools. GLIM: global leadership initiative on malnutrition; SGA: subjective global assessment; MIS: malnutrition inflammation score; GNRI: geriatric nutritional risk index.

Table 2. Hazard ratios for malnutrition diagnosed by GLIM and other nutrition measurement tools with overall mortality.

	Unadjusted			Model 1			Model 2
	HR	95% CI	p-value	HR	95% CI	p-value	HR	95% CI	p-value
GLIM	2.96	2.44 − 3.57	<0.001	2.94	2.42 − 3.56	<0.001	2.91	2.39 − 3.54	<0.001
SGA	2.26	1.87 − 2.72	<0.001	2.12	1.75 − 2.56	<0.001	2.14	1.77 − 2.59	<0.001
MIS	4.09	3.37 − 4.95	<0.001	4.03	3.31 − 4.92	<0.001	3.93	3.23 − 4.80	<0.001
GNRI	1.93	1.59 − 2.33	<0.001	1.84	1.52 − 2.22	<0.001	1.77	1.46 − 2.15	<0.001

Model 1: adjusted for age, sex, diabetes, cardiovascular disease and medicines; Model 2: adjusted for Model 1+ lipids, calcium, phosphate, parathyroid hormone, residual renal function, D/Pcr: C-reactive protein and Kt/Vurea; HR: adjusted hazard ratios; CI: confidence interval; GLIM: global leadership initiative on malnutrition; SGA: subjective global assessment; MIS: malnutrition inflammation score; GNRI: geriatric nutritional risk index.

Additionally, subgroup analyses were used to analyze the association between malnutrition diagnosed by GLIM and mortality among different categories of population. As shown in Figure 4, the p for interaction of subgroup interaction for age, gender, diabetes, CRP, and Kt/Vurea was >0.05, suggesting that the association between GLIM and death is independent of age, gender, diabetes, inflammation, and dialysis adequacy. However, we noted that GLIM exhibited a potential trend toward a higher risk of death in cardiovascular disease subgroup (p for interaction =0.021).

Figure 4. Adjusted hazard ratios of all-cause mortality for malnutrition diagnosed by GLIM in different subgroups Kt/Vurea: dialysis adequacy index; HR: hazard ratio.

3.4. Potency of GLIM and other nutrition measurement tools on mortality prediction

We analyzed the effectiveness of GLIM and other nutritional tools on 5-year mortality prediction using ROC curves. As shown in Figure 5, the sensitivity of GLIM was 55.0%, specificity was74.8%, and area under curve (AUC) was 0.65 (0.62 − 0.68), MIS had the leading comprehensive performance with a sensitivity of 54.8%, a specificity of 88.9%, and an AUC of 0.71 (0.69 − 0.75). In addition, NRI and IDI were used to distinguish the superiority of prediction models incorporated into nutritional markers. The results suggested that compared with the adjusted factors + GLIM model, the adjusted factors + MIS model had improved diagnostic performance for long-term overall survival (NRI and IDI >0, p < 0.001). In contrast, adjusted factors + SGA was homologous (p > 0.05), and adjusted factors + GNRI showed decline (NRI and IDI <0, p < 0.001), as shown in Table 3.

Figure 5. The ROC curves of GLIM and other nutrition measurement tools on mortality prediction. AUC: area under curve; GLIM: global leadership initiative on malnutrition; SGA: subjective global assessment; MIS: malnutrition inflammation score; GNRI, g:riatric nutritional risk index.

Table 3. Comparison of predictive performance in long-term survival of models matching GLIM and other nutrition measurement tools.

	NRI			IDI
Model	Estimate	95% CI	p-value	Estimate	95% CI	p-value
Adjusted factors + GLIM	Ref. Ref. Ref.			Ref. Ref. Ref.
Adjusted factor + SGA	−0.33	−0.39 − 0.27	0.211	−0.05	−0.09 − 0.01	0.082
Adjusted factors + MIS	0.24	0.03 − 0.40	0.018	0.06	0.01 − 0.12	0.040
Adjusted factors + GNRI	−0.31	−0.39 to −0.23	0.002	−0.05	−0.08 to −0.01	<0.001

NRI: net reclassification improvement; IDI: integrated discrimination improvement; GLIM: global leadership initiative on malnutrition; SGA: subjective global assessment; MIS: malnutrition inflammation score; GNRI: geriatric nutritional risk index; CI: confidence interval.

4. Discussion

Improving the nutritional status of patients with Protein-Energy Wasting (PEW) undergoing Peritoneal Dialysis (PD) continues to be a significant challenge. Existing assessment tools in the dialysis field range from simple to complex systems, however, they often lack ease of use and consistency in their conclusions. [17]. Designing more practical and feasible assessment tools is still a hot topic in the field of CKD nutrition. In this study, without fully considering the individual characteristics of dialysis patients, we assessed the effectiveness of GLIM and its correlation with mortality risk, aiming to align it with previous evaluation benchmarks. Our finding of a GLIM prevalence of malnutrition of 34.9% among patients undergoing PD is similar to that of Avesani et al. in the two hemodialysis cohorts (38.8% and 47.9%, respectively) [14]. The overall mortality rate was significantly higher in the malnourished group than in the well-nourished, even after adjusting for a multiple potential confounding factors. Moreover, in the sensitivity analysis, we found a clearly enhanced risk of mortality within the subgroup of patients with cardiovascular disease who were classified as malnourished according to GLIM. Finally, we examined the predictive value of multiple tools. For a 5-year survival rate, GLIM showed decent sensitivity, specificity and predictive efficiency. The predictive capacity of the multivariate model including GLIM was weaker than MIS, but fitter than GNRI. In brief, this study obtained available information on nutritional status in PD from a novel aspect.

The initial step of GLIM involves screening for the risk of undernutrition. However, within the unique population undergoing dialysis treatment with sugar-sweetened peritoneal dialysate, there exists significant biological individual variability. The prevalence of malnutrition or PEW can vary widely, influenced by underlying diseases and non-nutritional factors. In fact, the positive rate of malnutrition was less than 50% for any of the nutritional tools presented in this study. However, we used diverse nodal values for screening in the first step, which improved enrollment in the second step by 70.2%. In this regard, we believe that it is inappropriate to use an isolated scale that is one-size-fits-all, and some cases may be missed. Given that patients with PD are at broad risk for malnutrition, expanded screening is worthwhile in these people.

Regarding the phenotypes mentioned in GLIM, there is no concurrence on how to optimally assess and determine lost muscle mass (ranked third among phenotypic criteria). GLIM recommends diverse methods to measure reduced muscle mass, such as dual energy x-ray absorptiometry (DEXA), bioelectrical impedance analysis (BIA), computerized tomography (CT), magnetic resonance imaging, or ultrasound. Other modalities, for instance, physical examination or anthropometric measurements are also deemed alternatives. Since composition analysis is not universally applied in dialysis, in this study, manual measurement of muscle mass dimensions were used as an alternative. Indeed, many studies reports have used disparate criteria to evaluate reduced muscle mass [18,19]. Therefore, in light of the discrepancies observed within the data, further research is imperative to investigate muscle wasting across diverse conditions.

The value of predicting poor prognosis by previous nutritional status assessment instruments has been verified to be feasible in PD. Parameters in the SGA scale were considered to be benchmarks for nutritional assessment systems [20]. Prospective cohort studies have confirmed that the modified MIS scale based on the SGA is a powerful predictor of fatal and nonfatal cardiovascular events in dialysis patients [21]. The Geriatric Nutritional Risk Index (GNRI), derived solely from objective parameters such as BMI and serum albumin levels, has been substantiated as a straightforward and efficacious nutritional assessment tool within the dialysis cohort. [22]. Our consistency analysis data demonstrated that GLIM showed varying degrees of correlation with these evaluation methods (Kappa, p < 0.05), suggesting that these evaluation systems had statistical overlap in the selection of benchmark parameters, such as body weight and BMI, which were the top two factors in the ESPEN voting for phenotypic criteria. These shared components keep them close to each other. Moreover, the AUC illustrated that GLIM was generally effective in predicting 5-year mortality. Further IDI analysis showed that the combined prediction model of GLIM and traditional clinical factors for total mortality was inferior to MIS (IDI >0, p = 0.041) equivalent to SGA (p > 0.05), and slightly better than that based on GNRI (IDI <0, p < 0.001). We hypothesized that the predictive power of GLIM might be related to the integration of factors such as body weight, BMI, muscle mass, protein-energy intake, or inflammation, as these factors are mostly markers of all-cause mortality in dialysis patients. It merits mention that serum albumin, a robust prognostic indicator among dialysis patients, is not encompassed within the direct delineation of the GLIM criteria, thereby engendering a deficiency in certain essential evaluative dimensions. This could account for its relatively lower predictive capacity compared to the MIS, which integrates additional components such as albumin and transferrin, thereby constituting a more comprehensive tool.

Recently, the KDOQI-CKD Nutrition Guidelines have delineated that biomarkers, including nPNA, serum albumin, or prealbumin, may serve as adjunctive instruments for evaluating the nutritional status in adults undergoing PD. Concurrently, a diligent assessment of patients’ protein and energy consumption is instrumental in enhancing the efficacy of medical nutrition therapy [23]. Unfortunately, the pathogenesis of PEW in patients with PD is complex, and there is no single tool that can faultlessly assess malnutrition in patients with PD. It remains imperative to advocate for the development of holistic, bidirectional tools for intervention and feedback in the management of malnutrition. Although GLIM was originally designed for general use in differentiated clinical practice, the most optimized and modified version of the GLIM will need to be explored in individuals undergoing PD, and further studies are required to determine whether the findings are consistent with other regional or ethnic populations.

This work has some strengths and limitations that need to be stated. To the best of our knowledge, this represents the inaugural application of the GLIM criteria to investigate malnutrition within the PD cohort, thereby offering valuable reference points for future studies. The utilization of cohort datasets featuring standardized evaluation benchmarks across multiple centers may enhance the generalizability of the findings. Concurrently, the execution of multiple subgroup analyses may further underscore the robustness of the screening methodologies across diverse cohorts. However, as a retrospective study, selection bias represents an unavoidable limitation inherent to our study. Another limitation is that we did not use the body composition assessment to evaluation muscle mass loss in precise detail in GLIM. Furthermore, we lack complete and dynamic GLIM data, and cannot exclude that certain patients have ameliorated or deteriorated nutritional status. The significance of evaluating nutritional status cannot be ignored, since some scholars have found that the variability of nutrition is also related to the prognosis of dialysis [24]. GLIM continues to stand as a novel and pragmatic instrument. Our insights could bear implications for clinical practice and decision-making among nephrologists or dietitians, thereby necessitating additional prospective investigations for validation.

5. Conclusion

In conclusion, in the peritoneal dialysis population, we found that malnutrition as defined using GLIM criteria predicts worse outcomes, which is manifested in multiple subgroups. Its overall performance in the predictive value of mortality is inferior to the MIS, comparable to the SGA, and slightly fitter than the GNRI.

Ethical approval and consent to participate

This study was conducted in compliance with the ethical standards of the Declaration of Helsinki and authorized by the Medical Research Ethics Committee of the Jiujiang First People’s Hospital(jjsdyrmyy-yxyj-2021-107). And the Medical Research Ethics Committee of Jiujiang First People’s Hospital waived the informed consent procedure due to the noninvasive and anonymous.

Acknowledgement

The authors of this study would like to thank the patients who contributed to the data in the article.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

All available data have been presented in the paper.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.