METS-IR tracks cardiovascular death risk in peritoneal dialysis, but only modestly
In patients receiving peritoneal dialysis, a higher metabolic score for insulin resistance was linked to a greater risk of cardiovascular death, adding another layer of risk stratification to a population already burdened by competing threats from renal failure and vascular disease. The signal was consistent across standard survival modeling and competing-risk analysis, which strengthens the internal coherence of the finding even as the overall predictive performance remained limited.
The study examined 553 adults who began peritoneal dialysis between 2003 and 2017 at Changhua Christian Hospital in Taiwan. Participants were grouped into tertiles of the METS-IR index, a composite marker built from fasting glucose, triglycerides, high-density lipoprotein cholesterol, and body mass index. During follow-up, cardiovascular mortality rose stepwise across those tertiles: 16 deaths in the lowest group, 32 in the middle group, and 41 in the highest group. The gradient was sharp enough to suggest that insulin resistance biology is not merely background noise in dialysis care; it appears to map onto hard cardiovascular outcomes in a clinically meaningful way.
One number stands out immediately. The highest METS-IR tertile carried nearly double the adjusted hazard of cardiovascular death compared with the lowest tertile.
Cardiometabolic burden appears to matter in dialysis patients
Peritoneal dialysis patients live at the intersection of two high-risk states: kidney failure and metabolic dysfunction. Insulin resistance can accelerate atherosclerosis, worsen endothelial dysfunction, and contribute to myocardial injury through chronic disturbances in lipid handling, glucose metabolism, and vascular inflammation. In a dialysis cohort, those mechanisms may be amplified by the metabolic stress of uremia, fluid shifts, and the high prevalence of comorbidity.
The authors used Cox proportional hazards models and Fine-Gray competing-risk methods to account for the reality that not all deaths are cardiovascular and that some patients leave the risk set through transplantation or transfer to hemodialysis. After adjustment for medication use, demographic factors, comorbidities, peritoneal dialysis-related data, and laboratory values, tertile 2 was associated with a hazard ratio of 1.90 for cardiovascular mortality, and tertile 3 with a hazard ratio of 1.91, both compared with tertile 1. The close similarity of those adjusted estimates suggests the risk increase was not confined to only the very highest stratum; even intermediate elevations in METS-IR carried adverse prognostic weight.
The finding is biologically plausible. METS-IR blends glycemic exposure, triglyceride burden, HDL-C, and adiposity into one marker, and each of those components reflects the insulin-resistant state in a slightly different way. In a dialysis population, where conventional nutritional markers can perform poorly as prognostic tools, a score that captures metabolic stress may be more informative than a single laboratory value.
Predictive value was real, but discrimination stayed modest
The exploratory cutoff for cardiovascular mortality was 33.3, with 75.3% sensitivity and 48.3% specificity. The area under the receiver operating characteristic curve was 0.617, which indicates only modest discrimination. In practical terms, METS-IR can help identify a higher-risk group, but it cannot stand alone as a precise decision rule for individual bedside prediction.
That limitation is important because risk tools in dialysis often fail when asked to do too much. Patients with end-stage renal disease accumulate multiple overlapping causes of death, and cardiovascular mortality is only one part of a complex competing-risk environment. A marker may be statistically significant and still be too blunt for single-patient forecasting. METS-IR fits that pattern: useful as a screening signal, insufficient as a standalone prognostic engine.
The study design also shapes interpretation. This was a retrospective observational analysis from a single center, which makes the findings hypothesis-generating rather than definitive. Still, the absence of loss to follow-up is a strength, and the use of anonymized electronic medical records allowed longitudinal assessment across a long enrollment period. The cohort was followed until death or administrative censoring, including transition to hemodialysis, kidney transplantation, or the study endpoint.
Why insulin resistance may be a vascular problem in dialysis
Insulin resistance is often treated as a metabolic marker, but its vascular consequences are the reason it matters here. In a dialysis patient, persistent insulin resistance can worsen dyslipidemia, impair nitric oxide signaling, and promote coronary atherosclerosis. Over time, those pathways raise the likelihood that a patient with chronic kidney disease will die of cardiovascular causes rather than the renal disease itself. This is one reason the signal seen in METS-IR deserves attention: the score likely captures a cluster of injuries, not just a laboratory abnormality.
The study also speaks to an ongoing problem in nephrology practice. Common nutritional indicators such as albumin, potassium, and body mass index have not consistently predicted shortened survival in peritoneal dialysis patients. A composite metabolic score may offer a more stable lens through which to view risk, especially when clinicians are trying to distinguish between patients who look similar on routine dialysis labs but differ in cardiovascular vulnerability.
Still, the prognostic gap between statistical association and actionable precision remains wide. METS-IR identified risk, but its modest AUC means it should be treated as one piece of a broader assessment rather than a replacement for clinical judgment, cardiac evaluation, or management of traditional cardiovascular risk factors.
What this Taiwanese cohort adds to dialysis risk stratification
The data support a simple conclusion: higher METS-IR values are associated with greater cardiovascular mortality in peritoneal dialysis patients. That association persisted after adjustment and was reproduced by competing-risk modeling, which makes the result more robust than a single crude comparison. Even so, the score’s predictive performance was limited, and its best role at present is likely as an adjunctive marker rather than a standalone tool.
The open questions are straightforward and clinically important. This cohort was drawn from one hospital in Taiwan, and external validation in other dialysis populations is still needed. The study also does not establish whether interventions that reduce METS-IR would lower cardiovascular mortality, or whether the score is simply a marker of accumulated cardiometabolic injury. Until those questions are answered, METS-IR should be viewed as a risk flag with real but incomplete clinical utility, not as a treatment target in itself.