Predicting clinical gains and side effects of stimulant medication in paediatric ADHD

First of all Geir, can you tell us a little about yourself?

Image

I graduated as a psychologist from the University of Oslo when I was 25. I moved to Halden, where I still live, close to the Swedish border, and worked as a school psychologist for many years before I moved to child psychiatry. The focus of my professional interest has always been cognitive and behavioral disorders in children. In year 2000 I started as a part time consultant at the Norwegian Resource Centre for ADHD, Tourette syndrome and narcolepsy (NK), and met highly qualified and inspiring professionals from all Nordic countries, including members of GNC. This Nordic group had developed the broad spectrum 5-15 questionnaire. I was invited to be a member of this group, and we still work to improve this important, non-profit instrument.

Your research, how did that start?

-The Nordic group also inspired the researcher inside, and I started to work on my PhD 12 years ago. My primary motivation was to do a study on EEG biofeedback (neurofeedback) in ADHD. For several reasons my focus has changed towards EEG based methods as supplementary diagnostic tools in ADHD, and how such methods can help us predict responses to stimulant medication in ADHD.

-During my PhD years I was a part time student at the Norwegian University of Science and Technology (NTNU) in Trondheim. One of my advisers was Professor Juri Kropotov from St. Petersburg. He introduced the EEG/ERP (event-related potentials) methods to NTNU, applying the WinEEG program developed by him and his colleagues at the Institute of the Human Brain in St. Petersburg; the site where Pavlov did his famous dog experiments many years ago.

Can you, in few words, describe these EEG/ERP methods?

-EEG (electroencephalography) is used for registration of electrical brain activity, most often applied by neurologists looking for signs of brain pathology like epilepsy. Our use of EEG is different. We look for patterns that are associated with mental states, like focused attention, or diagnostic categories. The EEG signals are electronically processed and compared with a database, a method called QEEG (quantitative EEG). WinEEG is a program that we use for registration of EEG, for computations, and for comparisons with the database. In our paradigm we register EEG 3 min. eyes closed, 3 min. eyes opened, and during a 20 minutes attention task (a so called cued go/no-go task). ERPs (Event related potentials) are electrical signals extracted from the EEG during the task, and reflect stages of information processing in the brain, from early registration of stimuli to identification of target pictures, preparing a response, responding (“go”) or stopping a prepared response (“no-go”).

Your clinical experience and your research; what is the connection?

-In our neuropsychiatric clinic we had, for many years, used tests of attention, fine motor coordination and short-term memory/learning, on and off medication as supplements to rating scales in evaluating the quality of medication response in ADHD. From the literature and clinical practice we know that ratings from parents, teachers and the patients themselves can differ considerably, and as many as 30% of patients with ADHD do not have a clear positive effect of stimulants, usually methylphenidate (MPH). Acute side effects (insomnia, loss of appetite, negative emotions etc.) are manageable in most cases, but they are quite common, requiring a close follow-up.

-In our neurofeedback (NF) study we also looked for treatment induced changes in attention tests, EEG parameters and ERPs for those treated with NF, and for the matched control group treated with stimulants. We found no significant changes in the NF group. The medication responders (REs) improved their attention scores, and increased the amplitude of the ERP component P3 no-go, which is thought to reflect allocation of attention resources and/or inhibition. This change in P3 no-go was not seen in non-REs, or those treated with NF. 
Excess slow wave activity (theta) in EEG characterises 30-50% of ADHD patients (Kropotov 2016, Ogrim 2012), and most NF protocols for ADHD aim to reduce this activity. In addition, some studies report that excess theta is a predictor of positive response to stimulants. With this backdrop we implemented a clinical research project replacing our old test procedures with WinEEG methodology hypothesising that by combining attention scores and data from Quantitative EEG and ERPs we would be able to better predict medication responses.

Our first publication focused on predictors of acute side effects (SEs) based on one test completed without medication. The SE group had a faster test reaction time, had stronger ERPs reflecting emotional activation, and attention scores closer to normal values compared with the no-SE group.

-In a second study we found that responders (REs) had smaller P3 no-go amplitudes, more theta activity, and more deviant attention scores than the non-REs. The non-REs seemed to be more deviant in a posterior ERP component, probably reflecting problems in higher perceptual processes.

-The next study consisted of patients who had accepted two tests before the four-week medication trial; the last test on a single dose of stimulant medication. We focused on single-dose changes in attention scores and ERPs as predictors of clinical gains in everyday life. The single–dose test improved our predictions compared with predictions based on one test.

-In our last study, recently accepted for publication in Clinical EEG and Neuroscience, we combine all variables that significantly discriminate between REs and non-REs; SEs and no-SEs into two indexes – one for prediction of clinical gains and one for prediction of side effects. The effect size for clinical gains was d = 1.86 and for SEs: d = 1.08.

You have referred to “the innovation project”, can you tell us about that?

-Based on our data it seems that clinical gains and risk of acute side effects can be predicted with a clinically meaningful accuracy. We argued that this method should be offered to all patients in collaborating clinics for a one-year project period. We applied for and received innovation grants from Helse Sørøst (the largest of four health regions in Norway). Patients (8-17 years) diagnosed with ADHD and a decision to implement a medication trial are offered two tests, the last one on a single dose of stimulants, before the onset of a standard four weeks medication trial. Based on these two tests the prescribing doctor/local outpatient clinic and the parents receive a report estimating the probability of clinical gains and risk of side effects. We want to emphasise that there is uncertainty involved in all predictions, and that a standard trial should be completed. When we predict limited gains or risk of side effects, we advise the doctor to follow the patient extra closely. If negative predictions are confirmed, we suggest shifting to a medicine that is not based on MPH, or to rely completely on non-pharmacological treatments.

-A second group is also invited into the project; patients with uncertain effects of ongoing medication despite standard accommodations. Results from the comparison of the two tests, on and off medication, are reported to the doctor and parents/patients to help them decide whether to continue, change or stop the medication.

-Parents and teachers complete a rating scale before the medication trial, and after 2 months of treatment. The outpatient clinics/prescribing doctors/parents are asked to evaluate the procedure for each patient. Was the quality of the medication trial improved? Was the time needed to find the best treatment reduced? Were the predictions regarding clinical gains and side effects confirmed? These data will be used in the evaluation of the project.

Do you see future implications of this project?

-As a researcher I am of course interested in the accuracy of our predictors applied to a new sample. We also hope that the evaluations from clinicians, parents and patients turn out positive; i.e. improved quality of medication trials and reduced time to reach the best treatment. Each test takes about an hour. Time for scoring and report writing is estimated to take two hours. To spend four extra hours to increase the quality and reduce trial and error time is clinically meaningful in our opinion.

-With more data our predictors should be more reliable. In the future we hope to be able to say, supported by data, that MPH is / is not the right medication for this patient. One step further for non-REs to MPH: (S)he probably is a good candidate for treatment with medication X.

You are also involved in a biomarker project. How is this linked to the Innovation project?

-I am a supervisor and co-worker in the PhD project of my colleague and GNC member, neuropsychologist Linda Häger. The project is called “EEG based supplementary biomarkers for paediatric ADHD and autism spectrum disorders”. This is partly another story, but a future scenario is that the WinEEG test procedure can be offered as a supplement for making diagnoses. For patients diagnosed with ADHD and a decision to medicate, this test could be the first of two as described above. A specialised clinic/unit will be needed with qualified personnel.

What are the implications for future research?

-For the rest of my career as a researcher I think I will focus on EEG/ERP based methods in relation to ESSENCE.