Assessment of transthyretin cut-off values for a better screening of malnutrition: Retrospective determination and prospective validation

The use of transthyretin (TTR, prealbumin) as a marker of malnutrition and the definition of associated cut-offs are a matter of debate. In order to clarify this issue, we performed a retrospective study and then a prospective validation one. In the first study, data from 23,617 consecutive patients from our University hospital were analysed. Using the 0.11 and 0.05 g/L cut-off values defined by the French Health Authority, only 3.13% and 0.49% appeared malnourished or severely malnourished indicating that these cut-off values are clearly inappropriate. In the prospective study, consecutive patients were stratified for normal (≥0.2 g/L) or low (<0.2 g/L) TTR, and normal (<15 mg/L) or high (≥15 mg/L) C-reactive protein, hence defining 4 groups (n = 50 to 57/group), and data were analysed according to nutritional status estimated from patient files. Receiver operating characteristic (ROC) curve of TTR level associated with malnutrition allowed setting cut-off values at 0.17 and 0.12 g/L for malnutrition and severe malnutrition respectively.