The title of a presentation made by Nikolaos Koutsouleris, Munich, Germany, states clearly the goal of his research efforts. Machine learning – a multivariate pattern recognition technique - allows researchers to make predictions based on the patterns seen in various conditions. There are 2 phases – train the machine to identify the same patterns consistently – and then validate the training to identify the patterns in ‘unknowns’.
To differentially diagnose patients with two neurological conditions, such as MDD and schizophrenia, and to allow brain morphological changes to be predictive of disease, various sociodemographic confounding factors must be filtered from the patient information. By removing the effects of age and gender in the training group, Dr Koutsouleris can now distinguish between the brains of those with schizophrenia and MDD with 76% accuracy. The MDD and schizophrenia classifier has been externally validated and initial results seem promising.
However, further developments are in the pipe-line. Other variables such as age at onset and brain age permit separation of each group of patients from normal controls. Furthermore, it may be possible to separate early- and late-onset MDD using this technique; schizophrenia is less separable when these factors are taken into consideration. By incorporating additional variables such as these into the multivariate analyses, even better differentiation of MDD from schizophrenia may be possible.
However, for classification models to be used in clinical care, they need to present changes across the lifetime of the condition. And for this, there still seems some way to go.
It’s well known that the life expectancy of patients with schizophrenia is 10-25 years lower than the general population. All-cause mortality is 2.4 times greater. What may not be fully appreciated is that patients experience a significant burden of somatic morbidity, some of which is driving early mortality. For example, the risk of metabolic syndrome and all its components is 2 to 3 times greater in patients with schizophrenia than in the general population.
Under-diagnosed and under-treated
In Finland, a country which has national guidelines outlining a programme of interventions that should occur throughout a patient’s lifetime, somatic conditions still remain under-diagnosed and under-treated, especially in older patients. This is all the more disappointing as the main causes of somatic illness are the usual suspects – smoking, poor diet, lack of exercise, alcohol abuse, medication side-effects – as well as others that often seem to be overlooked – poor healthcare services and neglect.
However, Finland does have a remarkable resource at its disposal. The North of Finland 1966 Birth Cohort (NFBC1966) is a cohort of patients whose health has been followed since birth with snapshots being taken a 1, 14, 31 and 46 years of age. Jussi Seppӓlӓ, Oulu, Finland, utilised NFBC1966 to evaluate the prevalence of physical illness in patients with schizophrenia.
Psychotic co-morbidities extensive
From a population of 10,933 people followed from the ages of 16 to 46 years of age, 228 had schizophrenia and 240 others had other psychotic illness. The remainder of the population acted as a control. He found that in patients with schizophrenia, blood, endocrine, nutrition and metabolic disorders were more common than in the general population. However, in those with other psychotic illness, the rates of comorbidities were higher still. Clearly, there remains much still to be done not only in Finland, a nation with guidelines in place, but worldwide to counter this problem.
In the past, it was thought exercise was good for patients with schizophrenia because it simply reduced mortality, Alkomiet Hasan, Munich, Germany recalled. Indeed, exercise is recommended to all of us for this same endpoint! However, if schizophrenia is considered as a neurodevelopmental disorder, by interfering with the disease processes while the brain is still developing, some level of remission may be achieved. In patients with schizophrenia, exercise may be that plasticity-enhancing intervention.
Disturbed neuronal plasticity is considered to be part of the pathophysiology of schizophrenia and has been linked to the different clinical features associated with the disease. Thus, multi-episode schizophrenia patients showed significantly reduced plasticity compared to first episode schizophrenia patients and healthy controls. This deficit in plasticity appears related to the disease course and also to disturbed information processing in schizophrenia patients.
It has been shown that exercise and Cognitive Remediation Therapy (CRT) in patients improved global functioning. Until long-term data are available to support exercise as a facilitator of plasticity for sure, doesn’t it make sense to exercise anyway? It cannot do any harm.
Use it or lose it!
The important thing is to exercise as soon as disease is suspected or someone is at high risk– because this is when the plasticity of the brain is at its most vulnerable. The period when plasticity can be induced is thought to diminish over the course of the disease.
Once again, the NFBC1966 has proved highly valuable. Hannu Koponen, Helsinki, Finland, explained how he used its data to investigate real-life use of anti-psychotic medication throughout a patient’s life time.
Permission from 43-year old patients with schizophrenia was sought to allow details of their medication to be examined. Of the 258 approached, 99 agreed to be included. Patients were grouped according to their daily antipsychotic use: 50%, 50-95% or >95% of the time, and use was assessed since the onset of schizophrenia.
Mixed responses – heterogeneous disease?
81% of patients used antipsychotics. However, during the first 2-5 years post-diagnosis, use of antipsychotics was <50%. 83% of patients had drug-free periods. Drug-free periods became rarer over the course of the disease and those who had no drug-free periods at all had better SOFAS scores than those who had.
Higher functioning correlates with lower antipsychotic use
Those using antipsychotics 50% of the time were typically male, highly-educated, married and had better PANSS scores and clinical outcomes than those using medicines 50-95% of the time. The latter group had the most psychotic episodes.
Patients using lower doses of antipsychotic drugs had better outcome measures (SOFAS, remissions, CGI) than those using higher doses.
Slow to go from low to no
Dr Koponen suggested the data support the heterogeneity of schizophrenia and the need identify those patients who don’t need antipsychotics or who could to use low doses initially and then discontinue them. However, he warned, changes to medication shouldn’t be too quick.