Abstract
Accurate differentiation of schizophrenia (SZ) and bipolar disorder (BD) is crucial for effective clinical management. However, current diagnostic methods, which rely heavily on subjective assessments, are prone to high rates of misdiagnosis. This study pioneers the investigation of natural variations in lithium (Li) isotopes as potential biomarkers for differentiating BD and SZ. We identified significant and distinct variations in the isotopic compositions of Li in serum (δ(7)Li(serum)) of SZ patients relative to BD patients and health controls. Furthermore, we established a machine learning model that achieved a remarkable 100% accuracy in distinguishing between SZ and BD patients based on δ(7)Li(serum) fingerprints and concentrations of biologically relevant elements (Ca, Mg, Zn, and Se) in serum. Our research reveals that δ(7)Li(serum) is notably lighter in both BD and SZ patients (approximately 11 and 5‰, respectively) compared to that of the ingested Li drugs and decreases over time, primarily due to renal excretion. Additionally, in induced pluripotent stem cell (iPSC) models, we observed substantially heavier intracellular δ(7)Li values (up to 10‰) compared to the culture medium (0‰), likely originating from specific intracellular biochemical processes associated with competitions between Li(+) and Mg(2+). These differences in intracellular processes may significantly contribute to the observed distinctions in δ(7)Li(serum) values between BD and SZ patients. Our findings demonstrate that the δ(7)Li(serum) fingerprints in homeostasis provide valuable insights into the differentiate biomarker and pathological mechanism research of mental diseases.