Discussion
Eberhart and colleagues first described this new histologic category of infant brain tumors and called it embryonal tumors with abundant neuropil and true rosettes (ETANTR).1 Over the years, this was described with various names/acronyms including embryonal tumors with multilayered/ ependymoblastic rosettes, MEPL, EBL and supratentorial Primitive Neuroectodermal Tumors (sPNET).11 Until 2010, when Paulus and Kleihues proposed “ETMR” as a unifying entity to these embryonal tumors12 displaying multilayered rosettes as a common feature. In 2009, Pfister and colleagues discovered the novel oncogenic microRNA locus (C19MC) as the diagnostic genetic hallmark of this tumor.5 These studies led to the designation of ETMR-C19MC altered as a distinct new entity in the 2016 WHO CNS neoplasms.[13] C19MC amplification is present in about 90% of all ETMR regardless of histology.9,14 Transcriptomic profiling of C19MC-altered tumors revealed enrichment of these tumors for early neural and pluripotency genes including LIN 28/LIN 28B, reflecting their primitive nature. In 2013, Spence et al interrogated a large cohort of 450 paediatric tumors to assess the diagnostic specificity of C19MC amplification and LIN 28A.8 Their study demonstrated that C19MC alteration is seen in a subset of CNS-PNETs with high LIN 28 expression, however a small subset of other tumors such as rhabdoid brain tumors, malignant gliomas, atypical teratoid rhabdoid tumor (ATRT), germ cells tumors also express LIN 28A. While LIN 28A immunopositivity is not specific for the diagnosis of ETMR, we excluded our cohort for ATRT, germ cell tumors, high grade glioma by performing INI-1, Oct 3/4 and GFAP in all cases. LIN 28A expression was consistently expressed in all cases in our cohort. Additionally, to substantiate the above findings, we also performed LIN 28A immunohistochemistry in a case of ATRT (with INI-1 loss) featuring multilayered true rosettes. LIN 28A expression was seen within these rosettes as has been reported earlier in ATRTs.15Although LIN 28A is a surrogate marker for ETMR, its expression needs to be interpreted in context of histology. ATRT may often contain multilayered rosettes and presents a diagnostic challenge; INI-1 is extremely helpful in reliably distinguishing the two. All the tumors in our cohort readily expressed INI-1. Recently, using DNA methylation profiling and transcriptomic analysis Lambo et al,9have demonstrated that approximately 5% of ETMR negative for C19MC amplification contain biallelic DICER mutations (ETMR,DICER- altered) and these mutations are mutually exclusive to C19MC amplification.10 It is interesting to note how mutually exclusive DICER mutations and C19MC amplification leads to activation of oncogenic pathways. A common link is LIN 28A/Let 7 signalling which brings about reprogramming of stem cells. High expression of LIN 28A leads to depletion of Let-7 miRNA and subsequent upregulation of Let-7 targets.16 ETMR have low expression of Let-7 miRNA and typically have high expression of LIN 28A.
The tumor predominantly arises in the supratentorial region in 70% of cases 3,7,17 as was also observed in our cohort. Tumors in infratentorial region comprise 30% 3,17, 18with rare reports of these tumors arising in the spine.17,19 A trend for infratentorial location of C19MC- altered ETMR has been observed. ETMR has traditionally been described as a tumor of infants and young children, three patients in our cohort were more than 18 years of age.
ETMR encompass a wide spectrum of histology from tumors with abundant neuropil (ETANTR) with tumors with low neuropil content (EBL, MEPL). Multilayered rosettes are a distinctive feature, recognition of which amidst sheets of acellular neuropil (Figure 3B) or within cellular islands is critical. These are best recognized at low magnification, Ki67 is helpful in highlighting these mitotically active, undifferentiated cells with high N/C ratio surrounding central lumen (Figure 3C). In a single case, a continuum of morphological features was observed with tumor cells exhibiting ependymoblastic rosettes along with multilayered rosettes. MEPL comprising papillary and tubular structures reminiscent of embryonic neural tube was identified in the spinal tumor of adolescence girl. Another case of undifferentiated embryonal tumor with occasional rosettes was encountered with cells strongly expressing LIN 28A and retained nuclear expression for INI-1 (thereby excluding ATRT). ETMR are known to show neuronal and glial differentiation post-therapy.20,21 The tumor recurrence after 9 months of primary diagnosis in case #8 demonstrated areas of glial and rhabdoid differentiation. Interestingly, recurrent tumors on second surgery are known to transition from one histological variant to another.7
On imaging, these tumors usually appear large and well demarcated. Though 75% of them are supratentorial, posterior fossa lesions are well described. Large size of the lesions causes significant mass effect on the adjoining brain parenchyma, which is more marked in the infratentorial compartment, often with compression on peduncles. These solid masses are usually hypointense on T1-weighted and hyerintense on T2-weighted MR images showing diffusion restriction and mild patchy contrast enhancement. There is usually surrounding oedema and significant mass effect. The cerebral blood flow has been reported to be low on arterial spin labelling (ASL) sequence. However, the cerebral blood volume (CBV) on dynamic susceptibility contrast imaging is usually high. These features help differentiating ETMR from other childhood tumors. Pilocytic astrocytoma shows strong contrast enhancement with no diffusion restriction. Medulloblastomas and ATRT usually enhance well and have high cerebral blood flow using ASL. Furthermore, ATRT’s usually demonstrate central necrotic areas. Ependymomas often have increased T2 signal like ETMR, but have a variable diffusion restriction pattern.22
Complete resection undoubtedly translates into better survival; however certain location particularly brain stem and posterior fossa structures may not be amenable to complete resection. Possible benefits of second resections on survival following application of chemotherapy are also on record.21 Consensus guidelines for ideal chemotherapy regimen and radiotherapy dose- volume (CSI versus focal radiotherapy) remains poorly defined. Treatment strategies are based on results in CNS PNETs and other high-risk CNS-embryonal tumors. Various combinations of chemotherapeutic drugs that have been used so far include lomustine, vincristine, cyclophosphamide, etoposide and carboplatin.19
ETMR are aggressive pediatric brain tumors with universally poor prognosis and a median overall survival (OS) of only 12 months or less in most of the studies.7,8 The 5-year survival rate is 0-30%.14 The overall median survival observed in our cohort was 6.4 months and was worst for ETANTR (1.3 months) followed by EBL pattern (9.2 months); however, this did not differ significantly between the three histological patterns, similar to observations by Korshunov et al.7 As these tumors have recently been classified, there is limited data to guide prognostication and treatment.23 Cumulative data suggest C19MC patients who received chemotherapy with radiotherapy had significantly longer survival compared to untreated patients (median survival of 13 months versus 0.6 months, respectively).8 Studies have indicated that higher overall survival is seen in completely resected tumors with patients treated with radiotherapy and high-dose chemotherapy with autologous stem cell rescue.17,24
In conclusion, our study emphasizes the histological diversity inherent to ETMR. LIN 28A while being less specific, is a sensitive immunohistochemical marker. Prognosis of ETMR remains dismal with median survival of 6.4 months and this does not differ significantly amongst the three histological patterns.