Gene expression of CHD5, an ATP-dependent chromatin remodeling enzyme, has been reported to be restricted essentially to the nervous system [8, 10]. We describe for the first time that CHD5 is a neuron specific protein in normal neural tissue, with variable immunostaining intensity and intracellular localization among the neuron types of the cerebral cortex. Recent evidences suggest that the diverse neuron cell classes derive from distinct embryonal germinal zones and are characterized by specific cell signaling systems that regulate neural stem cells throughout the developing brain [13–15]. Thus, neuronal cells adopt a brain layer fate determined by their molecular profiles . While we did not observe a layer specific distribution of CHD5 in the cerebral cortex, we did note an association of CHD5 expression with neurons with distinct morphological, physiological and neurochemical features.
In normal neural tissue, glial cells appeared consistently devoid of CHD5 expression. In human glial tumors, chromosome arm 1p allelic loss is a frequent genetic abnormality, especially in oligodendrogliomas (70-85%) and astrocytomas (20-30%) . Recently, low levels of CHD5 expression have been reported in gliomas with 1p deletion, whereas nondeleted tumors displayed expression levels comparable to normal brain . Thus, deletion of CHD5 has been proposed as an initiating event in gliomas . Our findings, however, suggest that the role of CHD5 as a tumor suppressor in glial tumors needs further investigation.
NTs are embryonal cancers that are assumed to originate from primitive sympathetic neuroblast aggregates located in neural crest derived sympathetic nervous system. We observed how primitive neuroblast aggregates found in fetal adrenal gland specimens generally lack CHD5 expression. Interestingly, only a few cells were found with a variable degree of nuclear reactivity in larger aggregates. To date, the fate of these immature neuroblastic aggregates remains unsolved, and spontaneous involution and cell maturation have been proposed . The immunoreactivity observed in a small proportion of neuroblasts within these islets could suggest the establishment of CHD5 expression prior to their disappearance; however, no evident differentiating features were observed in these immunopositive cells that suggested the activation of the maturation process.
In NTs, CHD5 is essentially expressed in the nucleus of differentiating neuroblastic cells and ganglion cells, and absent in the Schwannian stromal component. However, the most intense immunoreactivity was observed in stage 4s NB, a rare subgroup of histologically undifferentiated, highly proliferative, metastatic tumors with a high incidence of spontaneous regression, affecting young infants. Accurate distinction of spontaneously regressing infant NB from high risk infant stage 4 can be difficult, but critical for therapeutic decisions. In our hands, the intensely positive CHD5 nuclear staining enabled a clear distinction of stage 4s NB from stage 4 NB, which was consistently immunonegative. These results are consistent with our previous gene expression profiling study, where similar differential CHD5 expression profiles were observed amongst infants with disseminated NB subgroups . Thus, CHD5 immunohistochemical staining may be clinically useful for a more accurate characterization of disseminated infant NB.
In NB, CHD5 nuclear staining was strongly associated with established favorable prognostic variables like low clinical stage, age at diagnosis <12 months and favorable histology. Our findings suggest that CHD5 protein expression may accurately define NB risk groups and may, therefore, be a prognostic marker. Evidence is provided by the statistically significant association found between high CHD5 immunoreactivity and favorable OS and EFS. These results are consistent with recent studies reporting a strong association of CHD5 mRNA levels with patient outcome in NB [5, 10]. Furthermore, Cox multivariate analyses suggest that the prognostic value of CHD5 protein expression is independent of other clinical and biological variables currently used in risk stratification of NB patients and could therefore represent an immunohistochemical marker of prognosis in NB.
Currently, risk stratification of NB patients is performed by combining different markers with strong prognostic impact, including patients' age at diagnosis, tumor stage, genomic amplification of the oncogene MYCN, copy number alterations of chromosomal regions 1p, 11q and 17q, tumor DNA content [1, 19] and Shimada histological score . However, despite elaborate risk stratification strategies, outcome prediction in neuroblastoma is still deficient. In recent years, to improve risk assessment additional prognostic indicators such as gene-expression signatures [21–23], combined genomic and molecular signatures  or expression levels of single candidate genes, e.g., Trk (NTRK) family of neurotrophin receptors [25, 26], FYN , PRAME  and ZNF423 , have been associated with NB clinical behavior. Expression of the Trk family receptors has been the most extensively characterized marker in NB and has been found to be consistently correlated with the biology and clinical behavior of NB. Based on our results, there is an apparent similarity between the expression patterns of CHD5 and TRKA in NB and their patterns of association with NB disease outcome. TRKA expression has been reported to be high in biologically favorable NB tumors and inversely associated with MYCN amplification . The prognostic value of the immunohistochemical detection of TrkA has also been examined and reported to be high, especially in combination with Ha-Ras expression pattern [31, 32]. Further IHC studies have correlated the lack of TrkA expression with metastatic malignant NB . However, in the latter study, 34% of the patients with stage 4 NB displayed TrkA expression, a subset of which died of aggressive metastatic disease despite TrkA expression [33, 34]. In our study, the majority of stage 4 NB either lacked CHD5 immunoreactivity (83%) or exhibited weak nuclear staining (13%), a high risk phenotype according to our scoring system. Only one stage 4 tumor was found to be clearly immunoreactive for CHD5; at the time of analysis the patient is alive, 29 months from diagnosis. These observations further confirm CHD5 as a powerful prognostic marker that could complement other known markers such as age at diagnosis, stage, MYCN status, cellular DNA content, 1p deletion and tumor histology. However, the potential clinical use of this marker must be tested in larger, prospective cohorts.
It is known that tumor histology and gene expression can change with treatment as a result of important changes in cellular processes, e.g., induced tumor differentiation, DNA repair, apoptosis and tissue necrosis. Undifferentiated NB occasionally exhibit neuroblastic maturation in response to chemotherapy. Assessment of CHD5 gene and protein expression in NB post-therapy specimens revealed that tumors with evident neuroblastic maturation showed both CHD5 gene and protein reactivation. Notably, none of these tumors harbored 1p deletion. Conversely, in tumors where minimal or no morphological changes were observed in the post-treatment specimens, low CHD5 expression persisted. These observations suggest the existence of a subset of tumors within high risk NB where CHD5 expression can be reactivated from the silenced state by standard chemotherapy. Remarkably, when post-therapy reactivation was observed, CHD5 expression was largely associated with disease response to cytotoxic induction therapy and subsequently with longer patient OS. All 12 patients included in the study received the same treatment, nevertheless some tumors failed to respond. At present, treatment response in NB is routinely evaluated by monitoring urine levels of catecholamine and its metabolites (VMA/HVA ratio) and by estimating the decrease in the size of measurable lesions with conventional imaging modalities, such as computed tomography (CT) or magnetic resonance imaging (MRI). At the time of second-look surgery, the degree of induced tumor cell differentiation and the extent of necrosis can also be useful to estimate treatment response. However, no biological markers for tumor chemotherapy responsiveness have been reported in NB. The use of such biomarkers would make chemotherapy more effective for individual patients by allowing timely changes of therapy in the case of nonresponding tumors. Furthermore, markers reflecting tumor response can function as surrogates of long-term outcome. Taking into account the small cohort of cases that may have led to an overestimation of the data, our findings would suggest that restoration of CHD5 expression could be a surrogate marker of treatment response that can be clinically useful to identify patients that do not benefit from conventional treatment. These results warrant further investigation in a larger cohort of uniformly treated patients.
In summary, we report that the differential expression of the neuron-specific protein CHD5 accurately defines NB risk groups and may represent a marker of outcome in neuroblastoma that can be tested by conventional immunohistochemistry. In high risk NB patients, re-establishment of CHD5 expression following chemotherapy should be tested prospectively as a surrogate marker of treatment response.