Article

MR and PET imaging may help predict onset of psychosis and guide treatment

Three out of 100 people in the US will experience a psychotic episode in their lifetime.1 In the UK, it’s estimated that 1% of the population will suffer an episode of the most common form of psychotic disorder: schizophrenia.2 A small proportion of patients with psychotic symptoms may have an underlying or secondary organic cause, such as a brain tumor or temporal lobe epilepsy. Therefore, a Magnetic Resonance Imaging (MRI) exam can help clinicians rule out an underlying condition and an Electroencephalogram (EEG) can detect spikes or sharp waves in, for example, the frontal or the temporal lobe in epilepsy patients.

How imaging exams may help our understanding

Many patients presenting with some degree of psychosis don’t automatically receive an MR exam, however. According to Matthew Kempton, PhD, Senior Lecturer and BRC Precision Psychiatry Fellow at King’s College London, psychiatrists will look for "red flags", such as visual hallucinations, which may be indicative of an underlying cause of the psychosis, before referring the patient for an MR exam.

Whilst brain scanning of psychosis patients is not routine clinical practice, there is an extensive neuroimaging research effort to better understand the underlying biology and help stratify patients at risk of psychosis, as well as predicting response to treatment.

Recent studies using MR and PET (Positron Emission Tomography) suggest that imaging may well have an important role in the prediction of psychosis onset and treatment management.3,4 This is particularly apposite for those patients not responding to traditional anti-psychotic medications.

Research studies have identified individuals who are at Clinical High Risk (CHR) of psychosis. These are people who either have a family history of psychosis, have subclinical symptoms or have experienced a brief episode of psychosis.

"We know that around 30 percent of these individuals will develop full-blown psychosis,” Dr. Kempton explains. “There is some evidence emerging that if we can treat these patients before they reach that threshold of psychosis then it may be possible to prevent it."

How imaging biomarkers may guide prediction

The challenge lies in identifying which individuals within this high risk group will go on to develop psychosis. Imaging biomarkers may help guide clinicians and researchers in such personalized prediction. Several studies have looked at volume changes in the hippocampus and anterior cingulate and lateral ventricles using structural MR imaging.5 Findings from diffusion tensor imaging suggest that fractional anisotropy reduces in CHR subjects who develop psychosis.

“There are two things we look at: prediction of clinical outcome from baseline imaging data when the patient first presents to us clinically,” says Dr. Kempton. “We can also look at the dynamic changes that occur in the brain when they transition to psychosis.”

It may also be possible to predict the transition to psychosis on a neurochemical basis.6 Researchers at King’s College are also currently looking to use this molecular information to predict treatment response by applying an established F-DOPA PET clinical protocol that has previously been used to aid diagnosis of Parkinson’s Disease.

How imaging may influence treatments

Similarly, imaging may also help predict whether a patient will respond to anti-psychotic medications.

“In PET imaging, there is some evidence that people with increased dopamine synthesis will respond to antipsychotic medication,” Dr. Kempton explains. “Conversely, some of the research that has been conducted at our Institute indicates that people with increased glutamate detected using an MR spectroscopy scan tend to not respond to medications that work on the dopamine system.”

MR spectroscopy has been used to measure glutamate levels in the anterior cingulate cortex (ACC) and the thalamus in minimally medicated and antipsychotic-naïve patients with a first episode. The results indicate a link of higher levels of ACC glutamate with a poor antipsychotic response.7

Yet, translating research findings into clinical practice remains difficult due to the small study sample sizes and lack of standardization in imaging protocols, sequences and systems. According to Dr. Kempton, there is a significant need for more robust, standardized MR sequences and large clinical data sets that are available to researchers.

There has been some success combining different study data sets and developing standardized protocols. The ENGIMA Consortium, led by Paul Thompson, MA, PhD, professor of neurology at the University of Southern California, has successfully pooled data to achieve the large sample sizes needed to examine and understand the subtle changes found in the MR images of mental health patients.

Dr. Kempton is involved in the EU-GEI (European Union Gene-Environment interaction) high risk study which links research centers that recruit individuals at Clinical High Risk of developing psychosis. This allows the centers to work together to collect much larger datasets than would otherwise be possible and supports international research across multiple scientific disciplines. The EU-GEI MR imaging protocols have been standardized across 11 different sites using the MR sequences recommended in ADNI-2 (Alzheimer’s Disease Neuroimaging Initiative).

How imaging may become more personalized

Yet, sometimes the biggest challenge is getting the most unwell patients to tolerate an imaging examination.

"That’s where advancements in MR imaging technology and sequences may help. More open or 'silent' MR systems and shorter, motion insensitive sequences which do not compromise image quality can make these exams more clinically viable," Dr. Kempton says. "Imagine if we could perform a short scan to look for increased dopamine or glutamate so we can better personalize their treatments."

There’s the potential for even greater implications.

“Some of the first PET scans linked personality with dopamine levels, where there was a link between high levels and psychosis,” adds Dr. Kempton.8 “No one has yet drilled down to link these behaviors, but in the near future it may well be possible.”

While deep learning algorithms may be the value-add that researchers are looking for, a key stepping stone is the development of a critical mass of expertly labelled imaging data—which is required for the effective training of these artificial intelligence networks and to allow them to generalize across different scanners and patient populations  

Combining imaging and other biomarkers with AI may increase sensitivity and specificity in the clinical assessment, help identify which patients should undergo more advanced imaging studies and help select candidates for clinical drug trials.

“Dopamine is a good example of where this could work,” explains Dr. Kempton. “If they have positive symptoms, then they may respond quickly to anti-psychotic meds. If we look at negative symptoms and the effects of emotion flattening, then we know these are typically the more difficult patients to treat.”

The EU-GEI study is now complete, with several papers being submitted to journals. According to Dr. Kempton, 65 of 344 people transitioned to psychosis. The study captured genetics, family history, longitudinal scans, blood based biomarkers and social/environmental factors in individuals at risk of psychosis and healthy controls. Other clinical studies are using medical imaging to evaluate people at high-risk for developing psychosis, such as PSYSCAN, NAPLS and PRONIA.  As findings are published, it is hoped the results can be linked and compared to better understand the brain and the relationship of structural and functional changes to the transition to psychosis.

References:

  1. Facts about psychosis. The National Institute of Mental Health. Available at: https://www.nimh.nih.gov/health/topics/schizophrenia/raise/fact-sheet-first-episode-psychosis.shtml
  2. Facts and Figures. Living with Schizophrenia. Available at: https://www.livingwithschizophreniauk.org/facts-and-figures/
  3. How can neuroimaging facilitate the diagnosis and stratification of patients with psychosis? Eur Neuropsychopharmacol. 2015 May; 25(5): 725–732. Last accessed October 22, 2019.
  4. Does neuroimaging have a role in predicting outcomes in psychosis? World Psychiatry. 2017 Jun; 16(2): 209–210. Last accessed October 22, 2019.
  5. Alterations in White Matter Evident Before the Onset of Psychosis. Schizophr Bull. 2012 Nov; 38(6):1170–1179. Last accessed October 22, 2019.
  6. Determinants of treatment response in first-episode psychosis: an (18)F-DOPA PET study. Mol Psychiatryhttps://www.ncbi.nlm.nih.gov/pubmed/29679071. Last accessed October 22, 2019.
  7. Response to initial antipsychotic treatment in first episode psychosis is related to anterior cingulate glutamate levels: a multicentre 1H-MRS study (OPTiMiSE). Mol Psychiatry. 2018 Nov;23(11):2145-2155. Last accessed October 22, 2019.
  8. Brain neuroreceptor density and personality traits: towards dimensional biomarkers for psychiatric disorders. Philos Trans R Soc Lond B Biol Sci. 2018 Apr 19; 373(1744):20170156. Last accessed October 22, 2019.