A [18F]FDG PET based nomogram to predict cancer-associated cachexia and survival outcome: A multi-center study

Objectives

Cancer patients with cachexia face poor prognosis and shortened survival. Early diagnosis and accurate prognosis prediction remain challenging. This multi-center study aims to develop and externally validate a nomogram integrating [¹⁸F]fluoro-2-deoxy-D-glucose ([¹⁸F]FDG) PET findings and routine clinical biochemistry tests for predicting cancer-associated cachexia, while also assessing its potential prognostic value.

Research Methods & Procedures

A retrospective analysis of 658 cancer patients (390 in the development cohort, 268 in the validation cohort) utilized [¹⁸F]FDG PET/CT data from two centers. Logistic regression identified organ-specific standardized uptake values (SUVs) and clinical variables associated with cancer-associated cachexia. Diagnostic accuracy, discriminative ability, and clinical effectiveness were assessed using area under the curve (AUC), calibration curve, and decision curve. Nomogram predictability for overall survival was evaluated through Cox regression and Kaplan–Meier curves.

Results

The combined nomogram incorporating age (odds ratio [OR] = 1.893; P = 0.012), hemoglobin (OR = 2.591; P < 0.001), maximum SUV of the liver (OR = 3.646; P < 0.001), and minimum SUV of the subcutaneous fat (OR = 5.060; P < 0.001) achieved good performance in predicting cancer-associated cachexia (AUC = 0.807/0.726, development/validation). Calibration and decision curve analyses confirmed its clinical effectiveness. Kaplan–Meier curves analysis showed that overall survival can be categorized using the combined nomogram (P < 0.001).

Conclusion

Combining radiological information from clinical standard [¹⁸F]FDG PET data from cancer patients with biochemical results in their routine clinical blood tests through a well-constructed nomogram enables predicting cachexia and its effect on the prognosis of cancer patients.