Association of Tumor Markers CA 15-3, CEA, and CA 125 with [¹⁸F]NaF PET Findings in Breast Cancer Patients

Arvin Naeimi^1,2Sara Harsini³Vilma Derbekyan⁴Gad Abikhzer⁴Marc Hickeson²Farzad Abbaspour^4,5*

• ¹Student Research Committee, School of Medicine, Guilan University of Medical Sciences, Rasht, Iran
• ²Department of Radiology, McGill University Health Center, McGill University, Montreal, Canada
• ³BC Cancer, Vancouver, Canada
• ⁴Division of Nuclear Medicine, Department of Radiology, McGill University Health Center, McGill University, Montreal, Canada
• ⁵Division of Nuclear Medicine, Department of Medicine, The Ottawa Hospital and University of Ottawa, ottawa, Canada

Introduction: Breast cancer often metastasizes to bone, and [¹⁸F]NaF PET is commonly used to detect skeletal involvement. This study examines the association of serum CA 15-3, CEA, and CA 125 with [¹⁸F]NaF PET findings in breast cancer to guide clinical decision-making. Methods: This retrospective study included 360 breast cancer patients who underwent [¹⁸F]NaF PET. Associations between serum tumor markers(CA 15-3, CEA, CA 125) and [¹⁸F]NaF PET findings and lesion count were analyzed. Optimal cut-off values for predicting [¹⁸F]NaF PET positivity were determined using ROC analysis. Multivariable logistic regression identified independent predictors. Results: Among 360 patients(mean age 61.1±13.9 years), median serum CA 15-3, CEA, and CA 125 levels were significantly elevated in patients with positive versus negative PET scans(all p<0.001). Marker levels revealed a dose–response relationship, rising with increasing numbers of skeletal lesions. In multivariable analysis, CA 15-3(OR 1.053, p=0.002) and CEA(OR 1.264, p=0.001) independently predicted PET positivity, whereas CA 125 showed a marginal trend(p=0.081). ROC analysis identified optimal cut-offs of 19.25 U/mL for CA 15-3(sensitivity 70.1%, specificity 90.4%, AUC 0.837) and 3.15 ng/mL for CEA(sensitivity 65.6%, specificity 85.2%, AUC 0.821). Combined model incorporating all three markers(probability cut-off 0.29) improved diagnostic performance(AUC 0.863; sensitivity 79.7%, specificity 92.3%). Invasive lobular histology and restaging indication were significant predictors of PET positivity. Conclusion: Elevated CA 15-3 and CEA independently predict [¹⁸F]NaF PET positivity in breast cancer. Optimal cut-offs were 19.25 U/mL for CA 15-3(sensitivity 70.1%, specificity 90.4%, LR+ 7.38, AUC 0.837) and 3.15 ng/mL for CEA(sensitivity 65.6%, specificity 85.2%, LR+ 4.43, AUC 0.821). The clinical utility of CA 15-3 and CEA lies in rule-in and risk-stratification strategies. Patients above these thresholds, particularly those with invasive lobular carcinoma undergoing restaging, may benefit from prioritized [¹⁸F]NaF PET evaluation or improved interpretation of equivocal PET findings. CA 15-3 threshold, lower than routine laboratory reference, may guide aggressive screening and prioritized [¹⁸F]NaF PET in patients with high clinical suspicion. Multivariable model combining CA 15-3, CEA, and CA 125(probability cut-off 0.29) improved diagnostic performance(sensitivity 79.7%, specificity 92.3%, AUC 0.863). Integrating CA 15-3 and CEA into clinical decision-making may enable a nuanced, risk-adapted approach, optimizing metastasis detection and resource allocation.