OBJECTIVE: Metabolic disorders are common in cancer patients. This study aimed to classify the metabolic disorder status of cancer patients using hematological indicators and to examine the association between disorder types and prognosis. METHODS: A cohort of 6307 patients from INSCOC was classified into three clusters via K-means clustering based on hematological indicators. Logistic regression and Cox models assessed each cluster's impact on adverse outcomes. RESULTS: A total of 6,307 participants were included in the study, K-means clustering divided the population into three groups, Cluster 1 (Normal Group, NG), Cluster 2 (Mild Disorder Group, MDG) and Cluster 3 (Severe Disorder Group, SDG). Compared to NG, MDG (OR = 2.268
95% CI: 1.967-2.616) and SDG (OR = 4.317
95% CI: 2.441-7.634) had significantly higher risks of sarcopenia. MDG (OR = 1.943
95% CI: 1.717-2.198) was associated with a higher risk of moderate malnutrition, and both MDG (OR = 3.786
95% CI: 3.282-4.368) and SDG (OR = 14.501
95% CI: 6.847-30.709) were identified as risk factors for severe malnutrition (p <
0.05). Cox regression analysis indicated that MDG and SDG were independent risk factors for all-cause mortality (MDG: HR = 1.460, 95% CI: 1.341-1.590
SDG: HR = 2.257, 95% CI: 1.622-3.140) and cancer-specific mortality (MDG: HR = 1.192, 95% CI: 1.039-1.367
SDG: HR = 2.068, 95% CI: 1.825-2.343) (p <
0.05). CONCLUSION: K-means clustering effectively categorized metabolic disorder subgroups, supporting targeted interventions and demonstrating a significant link between disorder severity and adverse outcomes.