Impact of 9p deletion and p16, Cyclin D1, and Myc hyperexpression on the outcome of anaplastic oligodendrogliomas

Objective To study the presence of 9p deletion and p16, cyclin D1 and Myc expression and their respective diagnostic and prognostic interest in oligodendrogliomas. Methods We analyzed a retrospective series of 40 consecutive anaplastic oligodendrogliomas (OIII) from a single institution and compared them to a control series of 10 low grade oligodendrogliomas (OII). Automated FISH analysis of chromosome 9p status and immunohistochemistry for p16, cyclin D1 and Myc was performed for all cases and correlated with clinical and histological data, event free survival (EFS) and overall survival (OS). Results Chromosome 9p deletion was observed in 55% of OIII (22/40) but not in OII. Deletion was highly correlated to EFS (median = 29 versus 53 months, p<0.0001) and OS (median = 48 versus 83 months, p<0.0001) in both the total cohort and the OIII population. In 9p non-deleted oligodendrogliomas, p16 hyperexpression correlated with a shorter OS (p = 0.02 in OII and p = 0.0001 in OIII) whereas lack of p16 expression was correlated to a shorter EFS and OS in 9p deleted OIII (p = 0.001 and p = 0.0002 respectively). Expression of Cyclin D1 was significantly higher in OIII (median expression 45% versus 14% for OII, p = 0.0006) and was correlated with MIB-1 expression (p<0.0001), vascular proliferation (p = 0.002), tumor necrosis (p = 0.04) and a shorter EFS in the total cohort (p = 0.05). Hyperexpression of Myc was correlated to grade (median expression 27% in OII versus 35% in OIII, p = 0.03), and to a shorter EFS in 9p non-deleted OIII (p = 0.01). Conclusion Chromosome 9p deletion identifies a subset of OIII with significantly worse prognosis. The combination of 9p status and p16 expression level identifies two distinct OIII populations with divergent prognosis. Hyperexpression of Bcl1 and Myc appears highly linked to anaplasia but the prognostic value is unclear and should be investigated further.


Introduction
The major aim of this study was to continue our previous work [1] on the prognostic value of chromosome 9p status in anaplastic oligodendrogliomas (OIII) and to confirm the reliability of the FISH technique using a standard FISH platform, an easily available commercial probe and an automated software analysis package with a previously established algorithm [2].
Since 2016, the WHO defines oligodendrogliomas (OGs) by the molecular genetic features of 1p/19q whole arm codeletion and IDH1/2 mutation [3,4]. These tumors are sensitive to chemotherapy given alone or after radiotherapy, with a global favorable outcome [5,6]. Additional genetic aberrations have been associated with higher grade OGs, in particular 9p loss, 9q loss, 10q loss, 11q gain, whole chromosome 7 gain and whole chromosome 4 loss [7,8]. Recent studies underlined the prognostic value of 9p deletion in OGs, which appears linked to two of the major histologic criteria of anaplasia traditionally used to define OIII, namely microvascular proliferation (MVP) and tumor necrosis [8][9][10] and may provide a genetic explanation for tumor progression in these cases [9].
In our previous study we showed the feasibility and reliability of an automated FISH technique for the study of chromosome 9p status in oligodendroglial tumors [1] but our conclusions were limited by the small size of the OG cohort. In the present study we wanted to confirm our previous findings on a larger cohort of well-defined IDH mutated and 1p/19q codeleted OG. At the same time we also wished to evaluate protein p16 (CDKN2A) expression in this cohort as a possible diagnostic and /or prognostic marker since the CDKN2A gene is located on 9p21. Finally, we studied the diagnostic and prognostic value of two additional proteins which have recently been implicated as markers of anaplasia and short outcome in OG [11,12] and which are also linked to p16: Cyclin-D1 (CCND1) which dimerizes with CDK4, the main target of p16 [13] and Myc (c-Myc) which impacts a wide number of cellular processes and may influence p16 via overexpression of HGMA2 and downregulation of CDKN2A [14].

Ethics statement
The local Institutional Care and Use Committee (IACUC) (ethics committee) of the Centre Hospitalier Universitaire de Québec was consulted and approved this study (notice 2017-3456): S1 File. Tumor samples were collected and anonymized by the Pathology Service of the Centre Hospitalier Universitaire de Québec (Hôpital de l'Enfant-Jésus, Quebec City, Canada).

FISH technique
FISH analysis of 9p status was performed using the Vysis CDKN2A/CEP 9 FISH Probe Kit (LSI 9p21 CDKN2A/9p11-q11 CEP 9; Abbott Molecular Inc., Abbott Park, Illinois, USA). Briefly, 5-μm-thick formalin-fixed, paraffin-embedded sections were deparaffinized, treated with saline sodium citrate and digested in pepsin solution. The probe mixture was prepared according to the manufacturer's instructions and an appropriate volume was added to each slide. Target DNA and probes were codenatured at 74˚C for 5 minutes and incubated at 37˚C overnight in a humidified hybridization chamber (Thermo-Brite, Abbott Molecular Inc.). Post-hybridization washes were performed in NP40 0.3%/2×SSC (pH 7.0) at 75˚C for 2 minutes. Finally, the slides were air dried and counterstained with DAPI (40,6-diamidino-2-phenylindole) diluted in Vectashield (Vector, Burlingame, CA, USA). Signal acquisition was performed for each slide over 12 more representative areas. These areas were automatically captured at x400 using a Metasystem station (Zeiss MetaSystems, Thornwood, NY) equipped with a Zeiss Axioplan fluorescent microscope. The acquired images were then used as the basis for automated analysis performed by the Metafer 4 software package (Metasystems).

FISH interpretation
9p status FISH analysis was automatically performed on all tumor cells identified by the Metafer 4 software using our internal algorithm [1,2]. This sampled an average of 891 cells (min: 90 -max: 1377 -median: 918) for chromosome 9p in the whole series.
The deletion status cut-off was calculated on a series of 5 normal autopsy brains using mean+2SD and was set at 20%. A tumor was classified as deleted if the percentage of deleted nuclei exceeded the deletion status cut-off of 20%. Otherwise a tumor was classified as nondeleted.

Statistical analyses
Statistical analyses was performed using the R statistical environment (http://www.Rproject.org/). Chi-square tests were used for group comparisons between clinical, histological and molecular status data. The strength of the linear relationship between quantitative variables was assessed using correlation analysis. Survival curves were obtained according to the Kaplan-Meier method and compared using the log-rank test. Event Free Survival (EFS) was defined as the time from diagnosis to the first recurrence. Overall Survival (OS) was defined as the time from diagnosis to death or last follow-up. The cohort was followed from 1998/01/01 to 2017/06/01. The following variables were queried for prognostic significance across the whole group: histological grade, recurrence (yes or no), age at diagnosis (cut off = 50 years), sex, extent of surgery (biopsy versus surgery), postoperative treatment (radio/chemotherapy versus none), microvascular proliferation (MVP) (present/absent), necrosis (present/absent), calcifications (present/absent), number of mitoses (cut off = 5 mitosis /10 HPF, median), INA expression (cut off = 10%), MIB1, p16, Cyclin D1 and Myc labelling index and 9p deletion. Variables that were significant in univariate analysis were evaluated in a multivariate cox regression model. Values 0.05 were considered statistically significant.

Histological data
The cohort was categorized into 3 subgroups according to WHO 2016 grading and chromosome 9p deletion status: OII, OIII -9p wild type (OIIIw) and OIII with 9p deletion (OIIId). No significant difference was observed between the sex, the localization and the extent of surgery distribution according to the 3 subgroups ( Table 1). The mean and median age at diagnosis for the OIIIw subgroup was significantly younger than the OIIId subgroup (p = 0.03). A significant proportion of OII patients was not treated by radiotherapy and/or chemotherapy compared to those of OIIIw and OIIId subgroups (p = 0.008).
MVP, high mitotic index (!5/10 HPF) and necrosis were linked to OIII subgroups (p = 0.0006, p<0.0001 and p<0.0001 respectively) as defined by the WHO 2016 classification without significant difference between OIIIw and OIIId subgroups.
Calcifications was not correlated to any subgroup.

Immunohistochemistry
For p16, Cyclin D1 and Myc, a strong nuclear staining was observed in some tumor cells in all cases, which allowed for the easy calculation of positive cell percentage for each case (Fig 1). MIB1 expression was present in both OII and OIII (mean = 9% and >23% respectively). Overexpression of MIB1 was highly correlated to the histological grade (Fig 2) but not to chromosome 9 deletion (p<0.0001 and p = 0.09 respectively): Table 1.
INA expression was present in the majority of the cohort (76%) with no significant correlation to the grade or the chromosome 9 deletion (Table 1).  IDH1 R132H expression which is correlated to IDH1 gene mutation was present in the majority of OII (9/10 = 90%) and OIII (38/40 = 95%). The three IDH1 R132H negative cases were mutated for IDH2 on complementary molecular analysis ( Table 1).
Loss of ATRX protein expression which is correlated to ATRX gene mutation [16] was not observed in any of our cases ( Table 1).

Molecular data
Automated analysis of 9p was easily performed in all the cases without need for a manual control. None of the OII cases presented a 9p deletion. In the OIII cohort, 22 cases were deleted for 9p (22/40 = 55%). The correlation of 9p deletion to the histological grade appeared very strong in univariate and multivariate analysis (p<0.0001). 9p deletion was also strongly correlated with the presence of tumor necrosis (p = 0.003) but not to the presence of MVP (Table 2).

Correlation of clinical, histological and molecular data with EFS and OS
In our series, histological grading showed no correlation with EFS or OS (Table 3) despite a higher median EFS and OS in OII compared to OIII (62 months versus 33 months and 110 months versus 53 months respectively). Chromosome arm 9p deletion appeared strongly linked to a poor survival in the total population for both EFS (median = 29 versus 53 months, p<0.0001) and OS (median = 48 versus 83 months, p<0.0001). This poor prognosis remained statistically significant in the OIII cohort. The OIIId subgroup had a shorter EFS (median = 29 versus 46 months, p = 0.0004) and OS (median = 48 versus 61 months, p = 0.001) than the OIIIw (Table 3) or OII subgroups (Fig 3a  and 3b). EFS median was 29 months for OIIId, 46 months for OIIIw and 62 months for O2 respectively (p = 0.0001). OS median was 48 months for OIIId, 61 months for OIIIw and 109 months for O2 respectively (p = 0.002).
Glioma recurrence was not observed in the OII subgroup. In the OIII subgroup recurrence was highly correlated with a shorter EFS (median = 16 versus 29 months, p = 0.002) and OS (median = 40 versus 61 months, p<0.0001), particularly in OIIId subgroup (p = 0.001 for EFS and p = 0.0001 for OS): Table 3.
Age at diagnosis over 50 years and sex were not correlated with shorter EFS or OS in our cohort. Surgical resection was correlated with a better OS than biopsy alone in the OII subgroup (median = 111 versus 78 months, p = 0.003). Frontal localization was correlated to a longer EFS (median = 47 versus 32 months, p = 0.004) and OS (median = 66 versus 54 months, p = 0.05) in the overall cohort.
The MIB1 labelling index presented a similar correlation with EFS and OS to the mitotic index.  Table 3.
INA overexpression showed no prognostic significance in our series (Table 3).
In the cohort as a whole p16 overexpression (cut off = 12%) was not correlated with EFS or OS (Table 3). In the OII cohort p16 overexpression was significantly correlated with a shorter OS (median = 84 versus 152 months, p = 0.02). There was a trend towards shorter EFS without statistical significance (median = 56 versus 108 months, p = 0.2). More detailed statistical analysis of the OIII cohort revealed two distinct p16 expression profiles which varied in impact according to chromosome 9p status. In the OIIIw subgroup a very high p16 overexpression (p16 > 30%, which corresponds to the highest quartile of this cohort) correlated with shorter EFS (median = 29 versus 57 months, p = 0.006) and OS (median = 32 versus 66 months, p = 0.0001) : Fig 3c and 3d respectively. At the opposite extreme in the OIIId subgroup, the absence of expression of p16 (p<7%) correlated with shorter EFS (median = 5 versus 32 months, p = 0.001) and OS (median = 28 versus 54 months, p = 0.0002) : Fig 3e and 3f respectively.
Cyclin D1 overexpression (cut off = 37%) was correlated with shorter EFS in the cohort as a whole (median = 35 versus 44 months, p = 0.05). Overexpression also showed a trend towards shorter OS without statistical significance (median = 61 versus 66 months, p = 0.09). Within the OIII series, Cyclin D1 overexpression was not correlated with EFS or OS (Table 3) for either OIIIw or OIIId.
A prognostic impact of Myc expression was only observed in the OIIIw subgroup where overexpression of Myc (cut off = 35%) was correlated with shorter EFS (median = 39 versus 66 months, p = 0.01).
In a multivariate analysis of clinical, histological and molecular data, only chromosome 9p deletion and tumor recurrence remained predictive of poor prognosis associating with both shorter EFS (p = 0.0001 and p = 0.001 respectively) and OS (p = 0.0005 and p<0.0001 respectively).

Discussion
As per the WHO 2016 classification of tumors of the central nervous system [3,4], our series of OG showed the double molecular signature of IDH mutation and 1p/19q codeletion. As 9p deletion and p16, Cyclin D1, and Myc hyperexpression in anaplastic oligodendrogliomas expected, the majority of our 50 cases showed a classical oligodendroglial morphology with monomorphic cells, uniform round nuclei and perinuclear haloes and classical honeycomb architecture. Only 8 cases (16% of the cohort), previously classified as astrocytomas or mixed oligoastrocytomas, were been reclassified into OG as a result of their molecular signature. This degree of reclassification is in concordance with the literature [17].
In our series, men were affected more frequently than women with a ratio of 1.3:1 and the median age at diagnosis was higher in OIIIs than OIIs (47 versus 41 years) similar to previous literature [18]. As expected, the majority of OGs were located in the frontal lobe (nearly 80%) then, in order of decreasing frequency, in the temporal, parietal and occipital lobes [4,19]. Frontal lobe localization was correlated to a longer EFS and OS [19]. There was no difference between the OII and OIII subgroups regarding other parameters except for postoperative treatment, with a higher percentage of untreated patients in the OII cohort than in the OIII cohort (40% versus 15%). Excluding tumor localization, these clinical parameters did not correlate with prognosis in this study, similar to previous literature [8].
According to the WHO 2016 criteria, high mitotic index, presence of MVP and tumor necrosis distinguish OIII from OII. In our series necrosis and high mitotic and MIB1 proliferative index had a strong negative predictive value, similar to previous literature [20]. A high proliferative index correlated with shorter EFS and OS within the OIII cohort, similar to previous literature [21]. Intratumoral calcifications were observed in approximatively half of our cases (46%) without discrimination between OII and OIII, similar to previous literature [4,20]. These calcifications correlated with a better OS in univariate analysis, as reported in few studies [22].
In addition to the canonical double molecular signature of OG, all cases in our series revealed other molecular characteristics typical of this tumor. All case showed retained nuclear expression of ATRX [23] and lacked p53 staining, consistent with the mutual exclusivity of TP53 mutation and 1p/19q deletion [24]. The majority of these cases (76%) also expressed INA which was expected since this protein is well known to be correlated to 1p/19q codeletion in the literature [8,15] and suggests a capacity for neuronal differentiation [25]. Unlike other studies in the literature [15,26], INA expression in our series was not correlated with EFS and/ or OS. This discordance may be explained by our smaller cohort size if not by the fact that these earlier studies predated the WHO 2016 classification and thus classified some 1p/19q non codeleted gliomas as OGs.
As expected, the FISH technique was easy to perform on all cases of our series and the automated analysis was easily and rapidly done on each case (less than 30 min / case for 12 images acquisition and analysis) without need for manual control. Chromosome arm 9p deletion is a common progression-associated alteration in several studies, found in up to 30 to 60% of high grade and/or recurrent OGs [1,9,27,28], although not all series show such frequent alteration [24]. Our present series shows no evidence of 9p deletion in the OII cohort, but presence of 9p deletion in the majority (55%) of our OIII cases, with a strong correlation between 9p deletion and histological grading shown by univariate and multivariate analysis. Previous studies have linked 9p deletion to necrosis and/or MVP in OGs [8][9][10] and 9p allelic loss has been linked to unfavourable outcome in a large clinical series [28]. Our data confirm the correlation of 9p deletion with necrosis but not with MVP and confirm a strong correlation with shorter EFS and OS by univariate and multivariate analysis. Unfavourable outcome secondary to 9p deletion has been attributed to the loss of the CDKN2A locus on chromosome 9p21 which includes the genes for p16 INK4A and p14 ARF . Protein p16, also called Cyclin-Dependent Kinase Inhibitor 2A (CDKN2A), is an inhibitor of the cyclin dependent kinases CDK4 and CDK6 which phosphorylate the retinoblastoma protein pRB (the key protein control of the cell cycle restriction point in G1 phase) and allow the progression from G1 phase to S phase [29]. By inhibiting CDK4 and CDK6, p16 impedes cell cycle progression from G1 to S phase and acts as a tumor suppressor that has been implicated in the prevention of cancers, notably gliomas [30,31]. Very few recent studies in the literature have explored p16 expression by immunohistochemistry in OGs. Studies predating the WHO 2016 molecular classification reported decreased p16 expression in a significant proportion of OGs (36 to 74%) independent of the histological grading [32][33][34]. This lack of expression correlated with poor survival [32,33] and was also associated with a higher proliferative activity [32]. Our results are only similar to these studies for the 9p deleted OIIId cohort in which 32% of our cases lacked p16 expression, with shorter EFS and OS by univariate analysis and with a significant association to a higher MIB1 proliferative index. Four other OIIIs in our series (10%) lacked p16 expression but were not deleted for 9p, which is also in concordance with published literature [9,33] and implies alternative genetic changes at the CDKN2A locus such as CDKN2A somatic mutation [9] or epigenetic changes such as hypermethylation [35,36]. Contrary to published literature we found no lack of p16 expression in the OII cohort and a correlation between p16 overexpression and a shorter OS in both OII and OIIIw, revealing two distinct unfavorable prognostic profiles for p16 expression according to the presence or absence of 9p deletion. This unexpected finding may be due to a statistical bias secondary to the small size of our cohort, but our series offers the advantage of being well characterized and homogeneous for IDH status and 1p 19q codeletion, which is not the case for other discordant studies in the literature. We observed two distinct outcomes for OIII according to the presence or absence of 9p deletion: the population without 9p deletion having a longer EFS and OS similar to that observed for OII and the population with 9p deletion presenting a significantly shorter OS and EFS consistent with a multistep model of oncogenesis in which multiple rounds of cell division and mutation take place, leading to acquisition of new genetic abnormalities which, together with clonal selection, shape the progression of the glioma [37]. Previous analyses of IDH mutated and 1p/19q codeleted gliomas suggest that protein p16 loss is not a consequence of deletion or mutation of the CDKN2A gene-as found in glioblastomas [24,37]-but rather to epigenetic changes [9,35,36] or impaired protein synthesis [32,33].
Our results suggest that 9p homozygous deletion leads to decreased protein p16 expression, loss of CDK4/6 inhibition, and an increase in cell proliferation as manifest by increased MIB-1, whereas in the absence of 9p deletion unfavorable evolution of OGs does not proceed via loss of p16 expression but probably via another mechanism resulting in the nuclear accumulation of p16 protein and its failure to fulfill its role of tumor suppressor. In a series of astrocytomas, p16 overexpression has been associated with short survival by univariate analysis but no mechanism for this finding was suggested [38]. Further studies are needed to confirm our observations and to explain their mechanism. A recent review underlines the contradictory facets of p16 expression and its significance in tumor pathobiology [13]. Genetic inactivation of p16 has been found in nearly 50% of all human cancers resulting in the bypass of this critical cell cycle control mechanism; on the other hand the overexpression of p16 at both mRNA and protein levels is associated with poor prognosis in several cancers including neuroblastoma, cervical, ovarian, breast, prostate and oral neoplasms [29] suggesting that p16 overexpression is a surrogate marker for rapid cell proliferation with failure to eliminate left-over p16 between mitoses [39].
Cyclin-D1 is a protein encoded by the CCND1 gene located on chromosome arm 11q13. It functions as a regulatory subunit of cyclin-dependent kinase 4 (CDK4) or 6 (CDK6). It dimerizes with CDK4 and CDK6 to form a complex which phosphorylates and inhibits the retinoblastoma protein (pRb) and promotes passage through the G1 phase to the S phase of the cell cycle. The binding of p16 to CDK4 or CDK6 disrupts their association with Cyclin D1 and leads to nuclear accumulation of Cyclin D1 [13]. Cyclin D1 overexpression has been shown to correlate with early cancer onset and with tumor progression [40] and has been observed in a large variety of tumors including gliomas [41,42]. As expected from its role in cell proliferation, Cyclin-D1 expression increases in anaplastic gliomas [42] and especially in OGs [11,43] showing a correlation with proliferative index [42,43]. Similar results are observed in our study with a strong correlation between Cyclin D1 overexpression and both OIII histological grade and high MIB1 proliferative index by both univariate and multivariate analysis. We also demonstrate a potential prognostic value for Cyclin D1 overexpression which appears correlated to EFS and to a lesser degree to OS in our series; this has been previously reported for astrocytomas [44] but never reported to our knowledge for OGs. The prognostic value of cyclin D1 immunostaining is not only due to the strong correlation between cyclin D1 overexpression and MIB1 overexpression (which was expected giving the molecular role of cyclin D1) but also to the strong correlation with MVP and necrosis, which represent the two major histological criteria that define anaplasia in OGs [3]. The average cyclin D1 labelling index in our series was constantly higher than that of MIB1 in the large majority of our cases (44/ 50 = 88%) and was not significantly correlated to hyperexpression of p16, which may be due, in addition to its role in cell cycle control, to the known abundance of cyclin D1 in Oligodendrocyte Precursor Cells (OPC) [11]. These precursor cells give rise to neurons and oligodendrocytes in the developing brain and are presumed to be the origin of OGs which would explain their proneural molecular signature including positivity for INA [11,25]. Given the present state of our data we would emphasize the diagnostic value of Cyclin D1 expression as a marker of anaplasia in OGs especially in poorly-sampled cases (biopsy and/or perilesional area). The prognostic utility of Cyclin D1 expression in OGs deserves to be confirmed by other studies.
Myc (c-Myc) is a regulatory gene located on 8q24. It belongs to the Myc family of transcriptor factors which includes N-Myc and L-Myc. It codes for a transcription factor and appears involved in the transcription of a very large panel of genes estimated to up to 15% of all human genes [45]. It codes for a nuclear phosphoprotein that plays a role in cell proliferation (by upregulating cyclins and downregulating p21), cell growth (by upregulating ribosomal RNA), apoptosis (by downregulating Bcl2) and inhibiting cell differentiation [46]. Myc is a protooncogene implicated in many type of neoplasm including brain tumors [47]. High levels of Myc protein expression secondary to gene amplification of Myc family members has been described in medulloblastomas [48], astrocytomas [49] and glioblastomas [50]. Genotyping studies showed variants mapping to 8q24, near the c-Myc locus, associated with increased risk for development of IDH mutant gliomas [51]. It has since been demonstrated that increased Myc activity is associated with malignant progression and worse prognosis in IDH-mutated gliomas [52,53] and especially in OGs [12]. Our study confirmed the high expression of Myc in all our OG cases compared to normal brain controls. It also confirmed the link between Myc overexpression and progression to anaplasia with a significant correlation with histological grading but not with proliferative index, MVP or necrosis as seen with Cyclin D1 and/or p16. Myc overexpression appeared only correlated to p16 overexpression which was unexpected giving the presumed inhibitory role of Myc on p16 function [14,46]. These data underline the complexity of molecular interactions in tumor process and may indicate an escape of p16 from Myc inhibitory control in OGs, but giving the absence of specific information on this subject in the literature, these findings need to be confirmed in future studies. In our study, the only significant prognostic impact of Myc expression was on EFS in the OIIIw subgroup. We would have expected an impact on OIIId EFS as observed by others [12]. This discrepancy is worth investigating further to see if it may be explained by differences in methodology, since other factors such as miRNA levels and post-translational modifications affecting protein turnover could affect the correlation between genomic expression data on one hand and protein levels on the other.
Recent molecular and epigenetic studies of OGs have identified a poor prognosis subgroup of anaplastic OG characterized by the presence of vascular proliferation, necrosis, 9p deletion and high Myc activity [12]. This subgroup presents a higher expression of oligodendrocyte precursor cell (OPC) markers, especially GPR17 an oligodendrocyte-specific G-protein-coupled receptor [54] and cyclin D1 [11]. In our study we confirm the presence of a similar poor prognosis subgroup of anaplastic OG characterized by 9p deletion, necrosis, high proliferative and mitotic index and loss of p16 expression. In this subgroup we also found Cyclin D1 and Myc overexpression but these expression levels were not significantly different from those in the rest of the cohort and there was no significant correlation with EFS and OS. In our study the prognostic impact of Myc overexpression in OG does not appear as clearly established that of 9p deletion or loss of p16 expression and deserves to be clarified in further studies.

Conclusions
This study confirms the reliability of automated 9p FISH analysis and its value in assessing the prognosis of anaplastic OGs. 9p deletion appeared strongly correlated with high grade OG status and with poor EFS and OS by univariate and multivariate analysis. Further stratification of OGs can be achieved by immunohistochemical studies available in most anatomic pathology laboratories. Analysis of p16 expression permits identification of a subgroup of OIII characterized by 9p homozygous deletion and p16 lack of expression with a particularly poor EFS and OS. Such patients would likely benefit from closer clinical follow up and might require a more intensive or a different chemotherapy protocol than usual, since OGs are usually chemosensitive. Cyclin D1 overexpression appeared to be a good marker of progression towards anaplasia in OGs and might be of prognostic interest if these findings are confirmed in further studies. Myc overexpression increased in parallel with the degree of malignancy in OGs, but showed no clear correlation with survival.