The epidemiology, biology and genetics of human astrocytic tumours

Abstract

In Sweden > 1000 primary central nervous system (CNS) tumours are diagnosed each year. The majority are malignant and in adults astrocytic gliomas dominate. These are classified into four malignancy grades according to WHO criteria, including Glioblastoma (GB), which is the most common and most malignant of all CNS tumours (malignancy grade IV), Anaplastic Astrocytoma (AA; malignancy grade Ill) and Astrocytoma (All; malignancy grade II). Prognosis is generally poor but varies markedly between the malignancy grades and even between individuals with the same malignancy grade. While numerous studies have addressed the issue of genetic prognostic markers in astrocytic tumours, little of clinical value has been established despite all the genetic information accumulated during the last two decades.

From 1987 to 1997 we collected >700 primary CNS tumours for molecular genetic analysis and the patients were followed-up clinically regarding age, sex, date of death and therapy received. We also collected blood samples from the patients, making it possible to pair-wise compare tumour DNA with normal DNA. The number of astrocytic glioma patients included in Paper I-III was 246 (172 GB, 54 AA and 20 All). The data was used for correlation analysis with the primary endpoint to identify associations between gene abnormalities and patient survival. We analyzed 129 GB and 37 AA for loss of six tumour suppressor genes (PTEN, RB1, CDKN2A, CDKN2B, p 14ARF and TP53) and amplification of three proto-oncogenes (CDK4, MDM2, and EGFR). These are the most commonly altered genes in astrocytic tumours. We found that abnormalities in any of the four genes coding for components of the Rb1 pathway (i.e. CDKN2A, CDKN2B, RB1 and CDK4), were associated with shorter survival (p=0.002) in GB.

In combination with loss of wild-type PTEN the association was even stronger (p<0.001). The survival difference (median survival in the two groups was 166 days compared to 437 days) was statistically significant in bivariate analysis adjusting for age (p=0.012). In AA we found that defects in any of the four Rb1 pathway-related genes were significantly associated with short survival (median survival 1.7 years compared to 9.9 years) in univariate (p=0.009) and multivariate analysis (p=0.013).

The Epidermal Growth Factor Receptor (EGFR) gene is frequently amplified and/or rearranged in astrocytic tumours, particularly in "primary" GB. The reported results on the prognostic impact of this have so far been contradictory and few reports have correlated both amplification and rearrangements with survival. We analyzed 221 astrocytic tumours for the presence of EGFR amplification and expression of the aberrant EGFRvIII transcript. In 160 "primary" GB patients we found no significant associations between EGFR abnormalities and survival. In 41 AA we found a tendency towards shorter survival in patients with tumours expressing the EGFRvIII transcript (p=0.069). These patients were significantly older than those with no EGFRvIII present (p=0.023).

As the aetiology of benign and malignant CNS tumours is largely unknown we performed an epidemiological study exploring if an association between CNS tumours and parathyroid adenomas exists (Paper IV). Known hereditary factors explain only a few percent of the incidence, as do the best characterized exogenous factors. We used the Swedish Cancer Registry (SCR) to identify all individuals operated for parathyroid adenomas between 1958 and 1999 (n=12,468) and using the SCR they were followed-up for the subsequent development of CNS tumours. There were 70 such cases observed. Compared to the 35 expected, the standard incidence ratio (SIR) was 2.0. This increased risk was independent of duration of follow-up and was confined to meningiomas (SIR=2.4), the most common type of benign tumour of the CNS, and neurinomas (SIR=3.4). For astrocytic tumours, after excluding the first year, the observed number was close to the expected (SIR=0.8).

With the studies included in this thesis, and the conclusions drawn, I hope to add something to the understanding of what causes CNS tumours as well as what dictates their prognosis.