Familial aggregation and prediction models of patients with early-onset and adult-onset schizophrenia and their nonpsychotic relatives using neurodevelopmental markers

  • 蔡 依凝

Student thesis: Master's Thesis


Background The neurodevelopmental hypothesis proposes that schizophrenia is originated from abnormal brain development Minor physical anomalies (MPAs) and neurological soft signs (NSSs) are suggested as biomarkers associated with disruptions of fetal development Schizophrenia patients have been reported to have more deficits in neurodevelopmental markers than healthy controls but family studies have produced different results Moreover the onset age of schizophrenia is related to the severity of the subsequent symptoms and thus it may be possible to estimate the predictive abilities of neurodevelopmental markers from early-onset and adult-onset schizophrenia We aimed to examine the familial aggregation of neurodevelopmental markers in early-onset and adult-onset schizophrenia families Methods We developed a modified physical measurement scale composed of both qualitative and quantitative items and used the Neurological Evaluation Scale (NES) to assess NSSs Participants included 182 schizophrenia patients 147 unaffected first-degree relatives of schizophrenia patients and 241 healthy controls First we estimated the predictive abilities of neurodevelopmental markers between early-onset schizophrenia (EOS) (onset age <20) and adult-onset schizophrenia (AOS) (onset age ? 20) using two data mining algorithms (artificial neural networks and decision trees) and a commonly used statistical method (logistic regression) we also used a 10-fold cross-validation method to measure the unbiased estimate of the prediction models Second we assessed the magnitude of familial aggregation for neurodevelopmental markers for EOS and AOS families using the relative recurrence-risk ratio Results The results of artificial neural networks were significantly more accurate than the other two methods Therefore we only showed the accuracy of artificial neural networks For the measurement of qualitative MPAs the accuracies for EOS and AOS were 78% and 73% respectively For qualitatively and quantitative measurements of MPAs (combined MPAs) the accuracies for EOS and AOS were 91% and 81% respectively The recurrence risk ratio for the total score of qualitative MPAs (Cut-Off Score MPAs ? 10) in EOS families was 4 16 (95%CI: 2 05-8 43) and in AOS families was 2 56 (95%CI: 1 06-6 19) The combined MPAs (Cut-Off Score MPAs ? 19) had a higher recurrence risk ratio than qualitative MPAs The recurrence risk ratio was 9 27 (95%CI: 3 88-22 16) in EOS families and 2 47 (95%CI: 0 61-9 97) in AOS families For the measurement of NSSs the accuracies of EOS and AOS were 85% and 78% respectively For sensory integration and motor coordination subscale the recurrence risk of EOS families was greater than that of AOS families in all NSS cut-off scores For example for the cut-off point of ? 1 in the Sensory Integration subscale the results showed that risk ratios were 2 63 (95%CI: 1 36-5 09) in EOS families and 2 15 (95%CI: 1 19-3 89) in AOS families For the cut-off point of ? 1 in the motor coordination subscale the results showed that risk ratios were 24 54 (95%CI: 7 81-77 05) in EOS families and 19 91 (95%CI: 6 49-61 10) in AOS families However no significant differences were found between EOS and AOS families for the sequencing of complex motor acts subscale and the others subscale Conclusion To our knowledge this study was the first to examine the association between familial aggregation and the age of onset among schizophrenia patients and their relatives in terms of neurodevelopmental markers These findings provide support for the potential of neurodevelopmental markers as a vulnerability indicator to schizophrenia Both MPAs and NSSs had higher predictive abilities for EOS than for AOS Therefore neurodevelopmental markers may have more accuracy in distinguishing EOS patients from healthy controls Furthermore evidence suggests that EOS families might have higher familial aggregation than AOS families in terms of MPAs and NSS Hence the findings of this study support the neurodevelopmental hypothesis that EOS might have greater familial and genetic risks
Date of Award2015 Jul 23
Original languageEnglish
SupervisorSheng-Hsiang Lin (Supervisor)

Cite this