Discussion
Neuroblastic tumors are subjected to risk stratification for management purposes and have an overall poor prognosis despite the extensive multimodal therapies highlighting the need for discovery of alternative therapeutic targets. 4,5,18 One such target is the Anaplastic Lymphoma Kinase or ALK gene which has been the subject of intense research in neuroblastomas and reviewed by several authors recently7,19 This study was a small step to establish the expression of ALK protein in samples obtained by fine needle aspiration biopsy and evaluate its role in prognostication and risk stratification. We further evaluated ALK gene mutations in a subset of cases by NGS on cell scrapes.
ALK protein is a receptor tyrosine kinase that has an expression restricted to the developing central and peripheral nervous system and is critical for embryonal neuronogenesis in the sympathetic ganglia and tumors showing ALK overexpression referred to as ALKomas might benefit from directed therapies.20 ALK protein expression in tissues may be evaluated by immunohistochemistry. There are many antibodies commercially available and a comparative analysis of various antibody clones revealed the high sensitivity of D5F3 clone, justifying its choice in this study.13 In this study, ALK expression was evaluated on adequately cellular cell blocks from FNA samples and 65% (35/54) cases showed positivity. The heterogeneity of ALK expression in most cases necessitated H-scoring which allowed the selection of strongly positive cases even if heterogenous and excluded the weak positives. There are a few studies that have evaluated ALK protein in histopathology sections including tissue microarrays with a frequency of 41-66% which is in agreement with our observations. 11,13,14 Our results on FNA cell blocks are comparable to studies on conventional histopathology specimens and thus it may be concluded that cell blocks are an acceptable substitute for evaluating ALK expression by immunocytochemistry.
ALK gene mutations in neuroblastoma is well studied.19,21,22 There is however no report from India and from cytology samples. Tumour DNA was extracted from MGG stained representative and cellular smears, a novel aspect of the study. The use of smears for molecular studies including NGS has been highlighted previously.23 Out of 21 cases evaluated, 3 pathogenic/likely pathogenic mutations were observed, two in ALKgene and one in the IDH1 gene. Both ALK mutations resulted in F1174L, which is the most frequent oncogenic driver mutation seen in 12.4% cases of sporadic neuroblastomas.8,24 Bresler et al in their comprehensive genomic study of ALK mutations in 1596 samples revealed tyrosine kinase domain mutations in 8% cases involving 3 hotspots R1275Q, F1174 and F1245 accounting for 43%, 30% and 12% of cases respectively.21 The F1174 was associated with an aggressive tumor phenotype as was observed in our cases too.
An interesting aspect of neuroblastoma is the chromosome 2p, which is home to two important oncogenes, MYCN  at 2p24 and ALK  at 2p23.2. In our study, MYCN amplification by FISH did not correlate with ALK protein expression (p=0.35). However, immunohistochemistry for MYCN protein was significantly associated with ALK protein expression.14 A meta-analysis of neuroblastomas comparing mutations in the  ALK  gene mutation spectrum with MYCN  amplification status found ALKmutations occur in equal frequencies across all genomic subtypes, but the F1174L mutants were observed in a higher frequency inMYCN -amplified tumors, demonstrated higher transforming capacity resulting in more aggressive tumors.25 Interestingly the ALK F1178L knock-in mice (corresponding to F1174L in humans) resulted in enhanced proliferation and extended neurogenesis but no tumor formation in the absence ofMYCN  co-expression.26 In another study on a mouse model, a co-operative interaction ofALKF1174L/MYCN   led to aggressive and lethal tumors with activation of the PI3K/AKT/mTOR and MAPK pathways.27
Out of the 50 patients with complete follow-up in this study, 28 patients (56%) died due to disease progression and 22 patients (44%) were alive without disease. In spite of this, in this small cohort of 60 patients, INRG stratification into high, intermediate, and low-risk groups had a significant effect on outcome (p=0.02) and overall survival (p=0.0001), similar to previous reports.28,29 ALK expression did not correlate with the age of patient, morphological type, stage (stage 4 vs non-stage 4), MYCN  status, and INRG group. We chose a cut-off of an H-score of 100 for statistical analysis. Those less than 100 represented negative or with low levels of expression. ALK protein expression independently had no correlation with outcome and overall survival which is in concordance with a previous report by Lee et al11 and De Brouwer et al25. Hence, we evaluated the effect of various parameters in the ALK positive cases wherein morphology correlated well with outcome (p=0.01) and overall survival (p=0.02) with poorly differentiated having a better overall survival than undifferentiated neuroblastoma. Our pilot observations need confirmation in larger study cohorts. When risk stratification was applied only to ALK positive tumors, significant results towards overall survival were observed (p=0.03). Weiser et al found a significantly higher number ofALK  mutations in high-risk versus low-risk cases (p=0.01) but there were no significant results with age or histology similar to our study.30 On the other hand, Carpenter et al reported a correlation of increased ALK  mRNA with poor prognosis.19 In vitro studies of ALK signaling in cultured neuroblasts resulted in transient proliferation followed by differentiation and long-term survival of post-mitotic neurons.26 A novel IDH1 gene missense mutation was observed in a single case. IDH1 gene mutations are common in gliomas and other tumors of the CNS and hemato-lymphoid malignancies.31 Functional in vitro studies on cell lines or mouse models are required to elucidate its role in neuroblastoma.
In view of the complex interplay of so many variables for risk stratification in neuroblastoma, all parameters which significantly correlated with outcome such as age, stage, cytomorphology were evaluated along with ALK expression in a multivariate Cox proportional hazard model which gave significant results (p= 0.04). A similar analysis substituting ALK with MYCN  was performed which was not statistically significant. Our pilot study data suggests that while ALK is not an independent prognostic factor, that it may have a role along with other traditional prognostic factors with ALK negativity predicting a worse outcome (Hazard Ratio, 2.3). A subset of poorly differentiated neuroblastomas who are ALK positive have a good survival and would otherwise would have been categorized as having a poor prognosis based on traditional parameters. Our data is limited due to the small cohort size and a potential inclusion bias of advanced cases which are subjected to FNAB. However, these observations need validation on studies with larger number of cases with better statistical power.
On the other hand, ALK has emerged as a potential predictive therapeutic biomarker and an ideal candidate for targeted therapy in neuroblastoma due to the dismal prognostic scenario.32 Several ALK inhibitors are in various phases of drug development and scrutiny in neuroblastoma and have shown promising results. Apart from crizotinib, other ALK inhibitors like lorlatinib, brigatinib, ceritinib, alectinib, entrectinib are undergoing clinical trials along with a combination of TRKA/B/C, MDM2, and CDK4/6 inhibitors and these are reviewed by Pacenta and Macy.33 These trials offer a ray of hope in neuroblastoma, a tumor that remains a challenge for the pediatric oncologist. Hence detection of ALK protein expression levels in any sample of neuroblastoma assumes significance.