Article Text

PDF
G12 Development and clinical acceptability of a pre-operative risk stratification tool of cerebellar mutism syndrome in children with posterior fossa tumour
  1. RA Dineen1,
  2. S Avula2,
  3. T Chambers1,
  4. M Dutta3,
  5. JF Liu4,
  6. D Soria5,6,
  7. P Quinlan5,6,
  8. D MacArthur7,
  9. S Howart7,
  10. S Harave2,
  11. C Ong2,
  12. C Mallucci8,
  13. R Kumar9,
  14. B Pizer10,
  15. DA Walker4
  1. 1Department of Radiology, University of Nottingham, Nottingham, UK
  2. 2Department of Radiology, Alder Hey Children’s NHS Foundation Trust, Liverpool, UK
  3. 3School of Medicine, University of Nottingham, Nottingham, UK
  4. 4Children’s Brain Tumour Research Centre, University of Nottingham, Nottingham, UK
  5. 5School of Computer Science, University of Nottingham, Nottingham, UK
  6. 6Advanced Data Analysis Centre, University of Nottingham, Nottingham, UK
  7. 7Department of Neurosurgery, Nottingham University NHS Trust, Nottingham, UK
  8. 8Department of Neurosurgery, Alder Hey Children’s NHS Foundation Trust, Liverpool, UK
  9. 9Department of Paediatric Neurology, Alder Hey Children’s NHS Foundation Trust, Liverpool, UK
  10. 10Derpartment of Paediatric Oncology, Alder Hey Children’s NHS Foundation Trust, Liverpool, UK

Abstract

Aims Despite identification of numerous pre-operative cerebellar mutism syndrome (CMS) clinical and radiological predictors, a unifying pre-operative risk stratification model for use during surgical consent is currently lacking. The aims of the project are (1) to develop a simple, easy to implemented risk scoring scheme to flag patients at higher risk of post-operative CMS; and (2) to assess its clinical acceptability amongst medical professionals.

Methods The combined cohort consists of 89 patients from two major treatment centres (age: 2–23yrs, gender 28M,61F, MRI pathology estimate 36 medulloblastoma, 40 pilocytic astrocytoma, 12 ependymoma, 1 non-committal); 26 (29%) of whom developed post-operative CMS. Post-operative CMS status was ascertained from clinical notes and pre-operative MRI scans, blinded to CMS status, underwent structured evaluation for 21 tightly-defined candidate imaging risk markers based on prior literature. All variables were first screened based upon results from univariate analysis and C4.5 decision tree. Stepwise logistic regression was then used to develop the optimal model, and multiple logistic regression coefficients for the predictors were converted into risk scores.

Results Univariate analysis identified five significant risks and C4.5 decision tree identified six predictors. The final model (Table 1) has an accuracy of 88.8% (79/89), with a sensitivity of 96.2% (25/26) and specificity of 85.7% (54/63). Using risk score cut-offs 203 and 238 permit discrimination into low (38/89, predicted probability < 3%), intermediate (17/89, predicted probability 3–52%) and high-risk (34/89, predicted probability ≥ 52%), respectively (Figure 1). Three illustrative cases from these categories will be used to collect clinicians’ opinion on surgical treatment decision and the acceptability of using this risk stratification for decision making and surgical consenting process. A web-based voting app will be used.

Abstract G12 Table 1

Variables in the risk prediction model and risk score

Abstract G12 Figure 1

Predicted post-operative CMS probability by risk score

Conclusions A risk stratification model for post-operative CMS could flag patients at increased risk pre-operatively and may influence strategies for surgical treatment of cerebellar tumours. Following future testing and prospective validation, this risk scoring scheme may be utilised during the surgical consenting process.

Statistics from Altmetric.com

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.