A large-scale analysis published in Cardiovascular Diabetology has identified a novel composite biomarker, ln[ALP × sCr], that independently predicts long-term mortality in patients with type 2 diabetes. It is a natural logarithm of the product of serum alkaline phosphatase (ALP) and serum creatinine (sCr)

Data were analysed from 4,839 adults with type 2 diabetes drawn from the US National Health and Nutrition Examination Survey (NHANES) between 1999 and 2014. Using deep learning for feature selection, serum alkaline phosphatase (ALP) and serum creatinine (sCr) were identified as key predictors. When combined into the logarithmic index ln[ALP × sCr], the biomarker showed a strong association with mortality outcomes.

During a median follow-up of 11.4 years, higher quartiles of ln[ALP × sCr] were linked to significantly increased all-cause mortality. Death rate increased from 20.6 to 66.8 deaths per 1,000 person-years between the lowest and highest quartiles. Fully adjusted analyses showed the index remained a robust predictor of both cardiovascular and diabetes-related deaths, with the latter risk more than doubling in the highest quartile.

Mediation analysis suggested that vitamin D deficiency partially reduced the association between ln[ALP × sCr] and mortality, highlighting a possible protective pathway. The biomarker also had a correlation with higher C-reactive protein, lower albumin, and greater insulin resistance, pointing to links with systemic inflammation and metabolic dysfunction.

Because ALP and creatinine are already routinely measured in clinical practice, the study suggests that ln[ALP × sCr] could serve as a cost-effective tool for early risk stratification in type 2 diabetes. Further validation in diverse populations is needed before clinical adoption.