Peripheral inflammatory and metabolic markers as potential biomarkers in treatment-resistant schizophrenia: Insights from a Qatari Cohort
Mohamed Adil Shah Khoodoruth, Tarteel Hussain, Sami Ouanes, Nuzhah Widaad Chut-Kai Khoodoruth, Adel Hmissi, Samuel L Lachica, Mustafa Nissar Bankur, Abdul Waheed Khan, Mohamad Samir Makki, Yasser Saeed Khan, James Currie, Majid Alabdullah, Farhan Mohammad
Abstract
• Schizophrenia is diagnosed clinically without established biomarkers. Identifying biomarkers could help identify TRS early. • CBC inflammatory markers differed between TRS and NTRS patients. • There are platelet-related differences between schizophrenia and controls. • Monocyte/Lymphocyte Ratio is significantly associated with PANSS positive scores. • The random forest machine learning model distinguishes TRS from NTRS with 88.41 % accuracy. Schizophrenia presents significant diagnostic and treatment challenges, particularly in distinguishing between treatment-resistant (TRS) and non-treatment-resistant schizophrenia (NTRS). This cross-sectional study analyzed routine laboratory parameters as potential biomarkers to differentiate TRS, NTRS, and healthy individuals within a Qatari cohort. The study included 31 TRS and 38 NTRS patients diagnosed with schizophrenia, alongside 30 control subjects from the Qatar Biobank. Key measurements included complete blood count, lipid panel, HbA1c, and ferritin levels. Our findings indicated elevated body mass index (BMI) and triglyceride (TG) levels in both patient groups compared to controls. The NTRS group also showed higher HbA1c levels. Variations in inflammatory markers were noted, with the NTRS group exhibiting a higher platelet/lymphocyte ratio (PLR). Multivariate analysis highlighted significant differences in platelet count, mean platelet volume (MPV), TG, HbA1c, BMI, neutrophil/lymphocyte ratio (NLR), monocyte/lymphocyte ratio (MLR), and ferritin among the groups. Linear regression analysis revealed that MLR and clozapine treatment were significantly correlated with the severity of schizophrenia symptoms. The Random Forest model, a supervised machine learning algorithm, efficiently differentiated between cases and controls and between TRS and NTRS, with accuracies of 86.87 % and 88.41 %, respectively. However, removing PANSS scores notably decreased the model's diagnostic effectiveness. These results suggest that accessible peripheral laboratory parameters can serve as useful biomarkers for schizophrenia, potentially aiding in the early identification of TRS, enhancing personalized treatment strategies, and contributing to precision psychiatry. Future longitudinal studies are necessary to confirm these findings and further explore the role of inflammation in schizophrenia pathophysiology and treatment response.