In 2002, Pomeroy et al., (2002) \citet{Pomeroy2002} demonstrated that CNS tumours (in particular MB) can be differentiated using their gene expression profiles at diagnosis. Groups have since have conducted similar studies and produce results generally in agreement \cite{Thompson2006,Northcott2009,Northcott2011,Cho2011,Fattet2009,Al-Halabi2011}, leading to the World Health Organisation (WHO) updating their CNS tumour classification system to include molecular profiling \citet{Louis2016}. The current consensus is that there are four molecular subgroups; Wingless (WNT), Sonic Hedgehog (SHH), Group 3 and Group 4. WNT and SHH are named due to their main signalling pathways, however less is understood surrounding the development of the tumour so more generic names are given \citet{DeSouza2014}.
Literature search and scope of review
A systematic literature search of PubMed,
[Insert search terms for genetics and genomics].
Apparent Diffusion Coefficient
Prior to the concept of radiomics, magnetic resonance imaging (MRI) studies investigating the nature posterior-fossa tumours using single quantifiable features. The primary aim of these studies was to distinguish tumour types by hypothesising that their singular extracted values would differ significantly enough to allow accurate classification. An accurate diagnosis of tumour type is important for correct treatment to be given, and allowing this to be performed non-invasively is beneficial.
By far the most investigated metric is apparent diffusion coefficient (ADC), acquired using diffusion weighted imaging (DWI). DWI is a non-contrast MR technique which provides spatial information on the Brownian motion of water molecules in tissue, including intra-, trans- and extra-celluar diffusion as well as perfusion \cite{Ahlawat2018}. Acquiring fully quantitative value of diffusion requires long scan times, so ADC is used as a semi-quantitative marker for diffusion which can be acquired in a much shorter time.
The assessment of ADC in paediatric MB has been thoroughly investigated \cite{Gauvain2001,Yamasaki_2010,Rumboldt2006,Yamashita2009,Jaremko2010,Pillai2011,Gimi2012,Bull2012,Yeom2013,Pierce2014,Assis2015,Zitouni2017,Koral2008}, with the earliest study being conducted in 2001 \cite{Gauvain2001}. The ADC within the MB is commonly compared with other posterior fossa tumours, notably with pilocytic astrosytoma (PA) and ependymoma (EP) \cite{Gauvain2001,Yamasaki2005,Rumboldt2006,Yamashita2009,Jaremko2010,Gimi2012,Bull2012,Pierce2014,Assis2015,Zitouni2017} .
The consensus of the literature is that MB possess, on average, a significantly reduced mean ADC when compared to PA and EP (Figure \ref{440437}), with PA also possessing a significantly higher ADC than EP. However, in some cases no significant difference can be seen in average ADC MB and EP \cite{Bull2012,Jaremko2010}. In addition, some papers have instead considered the minimum ADC as a key feature of tumour types \cite{Yamashita2009,Jaremko2010,Yeom2013,Pierce2014}, which have reported similar trends between MB, EP and PA.
These results demonstrate that ADC has the potential to be a useful marker for differentiate tumour type. Numerous studies have attempted to classify MB from other tumour types by calculating an upper ADC threshold, whereby a tumour contain a mean ADC lower than this threshold is classified as MB. Summarised below are these upper thresholds for MB reported in the literature.
Yamasaki et al., (2005) \cite{Yamasaki2005} reported that a threshold of 1x10-3 mm2/s allows can distinguish MB (N=9) from EP (N=6) with a sensitivity and specificity of 100%.
Similarly, Gimi et al., (2012) \cite{Gimi2012} reported a threshold of 0.909x10-3mm2/s can distinguish MB (N=73) from EP (N=30) with a sensitivity of 79% and specificity of 93%.
Pierce et al., (2014)
A threshold of 0.9 x10-3 mm2/s was proposed by Rumboldt et al., (2006) \cite{Rumboldt2006} to identify MB (N=8) with a sensitivity of 87.5% and a specificity of 100% from both EP (N=5) and PA (N=17).
Assis et al., (2015) \cite{Assis2015} agress with this threshold, suggesting that an average ADC of lower than this suggests a diagnosis of MB (N=8).
A lower threshold of 0.8x10-3mm2/s reported by Jaremko et al., (2010) \cite{Jaremko2010} allowed MB (N=7) to be distinguished from PA (N=12) with a sensitivity of 83% and specificity of 92%.\cite{Pierce2014a}
These cutoffs are in good agreement and yield good sensitivity and specificity for MB with both EP and PA.
A meta-analysis of the reported values (taking into account the sample sizes and including each study only once) shows that MB typically have an ADC of 0.70±0.16 x10-3mm2/s, whereas PA and EP have ADCs of 1.62±0.36 and 1.07±0.23 x10-3mm2/s. A total of 261 MB, 78 EP and 196 PA were included in the meta-analysis. Using these means and standard deviations to produce a normal distribution, mean ADC values were simulated for each tumour type, where the number of samples for each tumour match the proportions of tumours types from the meta-analysis. Using ROC analysis (Figure \ref{436425}), optimal thresholds can be found which minimises the sum of square differences between 100% sensitivity and specificity between different tumour types. When considering upper threshold for distinguishing MB from EP, it can be found that <0.872x10-3 mm2/s provides a sensitivity/specificity of 85/82%. Likewise, the upper threshold for distinguishing MB from PA can be found to be <1.00x10-3 mm2/s, providing a sensitivity/specificity of 97/95%. Finally, MB can be distinguished from non-MB (PA and EP) using a threshold of <0.946x10-3 mm2/s with a sensitivity/specificity of 93/90%.
These results suggest that average ADC is a useful measure in distinguishing MB from PA, however the mean ADC lacks the necessary sensitivity and specificity when attempting to distinguish