A data-driven framework for predicting PEG-defined MacroB12 in hypercobalaminemia

MacroB12 interference presents a significant challenge in the diagnosis of Vitamin B12 status, potentially masking true deficiency. To establish robust predictors and quantify the utility of pre-polyethene glycol (PEG) B12 levels, we conducted a retrospective analysis of 875 individuals with hypercobalaminemia (>1000 pg/mL), using multiple imputation to handle missing data. Multivariable regression modelling revealed that PEG-defined MacroB12 positivity (<30% PEG recovery) was independently associated with a profile suggestive of autoimmunity, including older age (adjusted odds ratio [aOR] 1.03 per year, P<0.001) and a Rheumatologic/Autoimmune diagnosis (aOR 2.96, P=0.01). We also identified a counterintuitive haematological signature, with higher haemoglobin (aOR 1.14, P=0.02) and Mean Corpuscular Volume (aOR 1.04, P=0.01) predicting MacroB12. Receiver Operating Characteristic (ROC) analysis of pre-PEG B12 concentration yielded moderate discriminatory power (Area Under the Curve [AUC] = 0.744). However, a combined multivariable model integrating clinical features significantly enhanced prediction (AUC = 0.802; P=0.002). The optimal threshold of 1584.0 pg/mL (sensitivity 71.3%, specificity 69.7%) serves to stratify clinical suspicion, but its performance confirms that it cannot replace definitive confirmatory testing. Our findings define a clinical and laboratory phenotype for suspected MacroB12 but highlight that confirmatory PEG precipitation remains essential for accurate diagnosis, particularly in older patients and those with autoimmune disease. These findings were derived using the Abbott Alinity i platform; validation on alternative immunoassay systems is required before broader clinical adoption.