Question 20

College answer

What was required to score well in this question was an understanding of positive and negative predictive value of these imaging modalities, and how they need to be used and interpreted in conjunction with other prognostication tests. Many candidates mentioned a multimodal prognostic model, however, they failed to elaborate on how they are utilised in a multimodal approach.

Discussion

If "what was required to score well in this question was an understanding of positive and negative predictive value", then perhaps the words "positive and negative predictive value"  should have appeared in the stem, otherwise it would seem unfair to expect the exam candidates to guess their way to the optimum marks. Still, one may be able to argue that the stem is about diagnostic tests, and is phrased in a way that leads naturally to the discussion of their validity, of which NPV and PPV are an important part. 

a)

CT features of hypoxic ischaemic encephalopathy:

• Specific findings
  • Loss of grey-white matter differentiation
  • Reversal of grey-white differentiation
  • "White cerebellum" sign
  • Effacement of CSF structures (ventricles, sulci, gyri)
  • Tonsillar herniation
  • Pseudosubarachnoid (a late and severe sign)
  • Cortical laminar necrosis
• Generalised features:
  • Bilateral diffuse lesions
  • Predominantly grey matter damage
  • Distribution along watershed areas

MRI features of hypoxic ischaemic encephalopathy:

• Early increased signal intensity on DWI
• Diffusion restriction on ADC
• Later, increased signal intensity on T2/FLAIR
• Very late, cortical laminar necrosis on T1

b)

• Predictive value of CT 
  • Early (within 12-24 hrs) CT may not demonstrate any findings.
  • Loss of grey white differentiation at 24-48 hrs has false positive rate of around 8% (qualitative) but close to 0% with quantitative measurements of the grey-white ratio (threshold values vary in the literature
  • Sensitivity for predicting poor neurological outcome remains poor.
  •  Thus, CT has good positive predictive value, but poor negative predictive value (i.e. a normal CT does not rule out a severe HIE)
• Predictive value of MRI
  • A diffuse pattern of restricted diffusion on MRI (DWI with ADC analysis) at days 2 to 7 post ROSC has a very high positive predictive value (0.95, sensitivity 63%, specificity 96%).
  • Absence of DWI lesions (or an isolated lesion) has a sensitivity of 94% and a specificity of 92% for good 6-month functional outcome (Rajajee et al, 2023). 
  • Therefore MRI has good positive and negative predictive value, with a false positive rate quoted as  0-5%.
• Integration into a multimodal prognostic model:
  • The stand-alone predictive value of imaging in neuroprognostication is low.
  • Prognostication algorithms such as the ERC/ESICM (2020) or NCS (2023) incorporate imaging as "useful but less robust predictors", to be used together with other predictors.
  • When strongly suggestive CT or MRI are combined with one other clinical predictor (eg. poor motor score or absent pupillary reflex), the false positive rate drops to 0% (Youn et al, 2022).