P13.04 Transcriptional profiling differentiates spinal cord ependymal tumours from other glial tumours in children

BACKGROUND

Spinal cord tumours in children include ependymal and other tumours of glial origin, all of them displaying GFAP marker. Some tumours may present difficulties in histopathological diagnosis e.g. due to intratumour heterogeneity. Therefore, molecular approach should be examined for its diagnostic usefulness. We investigated a series of originally diagnosed spinal cord ependymal tumours, using transcriptional profiling, to confirm if molecular classification is compatible with histopathological diagnosis.

MATERIAL AND METHODS

Overall, 9 patients diagnosed between 2001 and 2014 in The Children’s Memorial Health Institute (CMHI) in Warsaw, Poland, were included in the analysis. 4 patients were diagnosed with GI myxopapillary ependymomas, 3 patients with GII and 2 patients with GIII grade ependymal tumours. Total RNA was extracted from FFPE tumour samples using RNeasy kits (Qiagen). Transcriptional profiling was performed using NanoString nCounter system analysis. Tumours were analysed according to the manufacturer’s procedures for hybridization, detection and scanning. Raw counts for each gene underwent technical and biological normalization using nSolver software. A custom NanoString CodeSet consisted of 8 marker genes: GFAP as a glial marker, CAPS and FAM81B as the markers for ependymal tumours and OLIG1, OLIG2, PMP2, KLRC3 and C1orf61 as the markers for other gliomas. Marker genes were identified by re-analysis of publically available microarrays data from childhood brain tumour samples. Histopathological reassessment of tumour preparations was performed by two experienced neuropathologists.

RESULTS

Hierarchical clustering of 9 samples revealed that two tumours clustered separately and displayed low expression of ependymal markers but high expression of all other glioma markers. Subsequent histopathological re-analysis of preparation of the later 2 tumours showed that one of them displayed features of pilocytic astrocytoma with loose, microcytic and fibrillary areas accompanied by advanced myxoid degenerative changes mimicking the picture of myxopapillary ependymoma. The second tumour revealed heterogeneous morphology with areas of diffuse low grade astrocytoma, pilocytic astrocytoma and fragments displaying ependymal features, including angiocentric pattern with ependymoma-like perivascular arrangement of neoplastic cells.

CONCLUSION

The results of analysis indicate that RNA expression profiling from FFPE tumour samples using NanoString nCounter system may be a useful approach in differentiation of ependymal tumours from other glial tumours with misleading morphology located in the spinal cord in children. Funded by the National Science Centre, Poland (2016/21/B/NZ2/01785 and 2016/23/B/NZ2/03064).

Difficulties in recognition of pyruvate dehydrogenase complex deficiency on the basis of clinical and biochemical features. The role of next-generation sequencing

Pyruvate dehydrogenase complex (PDHc) defect is a well-known cause of mitochondrial disorders (MD) with at least six responsible genes (PDHA1, PDHB, DLAT, DLD, PDHX, PDP1). The aim of this work was to assess the diagnostic value of biochemical methods in recognition of PDHc defect in Polish patients with suspicion of MD.

In the first step, Western blot of the E1α subunit was performed on 86 archive muscle bioptates with suspicion of MD. In the second step, Sanger PDHA1 sequencing was performed in 21 cases with low E1α expression. In the third step, 7 patients with negative results of PDHA1 sequencing were subjected to whole-exome sequencing (WES). This protocol revealed 4 patients with PDHA1 and one with DLD mutations. Four additional probands were diagnosed outside the protocol (WES or Sanger sequencing).

The molecular characterization of PDHc defect was conducted in a total of 9 probands: 5 according to and 4 off the protocol. Additionally, two affected relatives were recognized by a family study. Altogether we identified seven different PDHA1 changes, including two novel variants [c.464T > C (p.Met155Thr) and c.856_859dupACTT (p.Arg288Leufs*10)] and one DLD variant.

The lactate response to glucose load in the PDHA1 subset was compared to a subset of non PDHc-related MD. Opposite responses were observed, with an increase of 23% and decrease of 27%, respectively.

The results show that determining lactate response to glucose load and muscle E1α expression may contribute to distinguishing PDHc-related and other MD, however, WES is becoming the method of choice for MD diagnostics.