- Email: [email protected]
Pineal Parenchymal Tumours: II
Clinical Oncology (2004) 16: 244–247 doi:10.1016/j.clon.2003.12.005
Case Report Pineal Parenchymal Tumours: II On the Aggressive Behaviour of Pineoblastoma in Patients with an Inherited Mutation of the RB1 Gene P. N. Plowman*, B. Pizer†, J. E. Kingston‡ *Department of Clinical Oncology, St Bartholomew’s Hospital, London, U.K.; †Department of Paediatric Oncology, Alder Hey Hospital, Liverpool, U.K.; ‡Department of Paediatric Oncology, Barts and the London NHS Trust, London U.K. ABSTRACT: This report relates to a retrospective analysis of two non-randomised cohorts of patients with pineoblastoma, with some differences in presenting features and treatment characteristics. We have identified a large difference in survival depending on the possession or otherwise of the mutated RB (retinoblastoma) gene in the genome/karyotype. Eight children with familial retinoblastoma (non-metastatic at presentation) developed pineoblastoma and were treated by chemotherapy and radiotherapy. The survival of these patients was compared with the survival of nine non-metastatic sporadic cases of pineoblastoma similarly staged and treated. One out of eight children having the RB mutation in the genome survived compared with seven out of nine in the group with sporadic pineoblastoma (P=0.002). It is suggested that the inheritance of the mutated retinoblastoma gene is not only causal in the generation of this tumour type but, in a way that is yet to be defined, renders such tumours more aggressive or less responsive to therapy. With the current interest in the role of RB mutations in other cancers (where the prognostic import of single genes is less easily identified), this observation may have wider relevance. Plowman P. N. et al. (2004). Clinical Oncology 16, 244–247 2004 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved. Key words: Pineal parenchymal tumours, pineoblastoma, RB1 gene Received: 4 August 2003
Revised: 5 November 2003
Introduction
Retinoblastoma is a well-researched retinal cancer afflicting young children. Genetic (hereditary) and nongenetic (sporadic) forms occur. The discovery of the retinoblastoma gene RB1, on chromosome 13q14 [1] led to the validation of the Knudson ‘two hit hypothesis’ of this cancer’s causation at the molecular level [2]. The RB1 gene encodes a ubiquitously expressed nuclear protein whose level of phosphorylation importantly varies throughout the cell cycle and thereby influences cell-cycle progression in a negative or inhibitory manner. Mutations of this RB1 gene, when carried in the genome (as shown in hereditary cases) [3], lead to multiple retinal tumours (both retinae) occurring at an early age. Furthermore, the pineal gland whose parenchymal cells were once, specifically in primitive fish the ‘retinal carpet’ of the pineal eye, is also at risk, with pineal tumours identical (in all respects other than location) to retinoblastoma occurring in 4% of children with the Author for correspondence: P. N. Plowman, Department of Clinical Oncology, St Bartholomew’s Hospital, London EC1, U.K.; E-mail: [email protected] 0936-6555/04/040244+4 $30.00/0
Accepted: 17 December 2003
inherited form of the disease [4,5]. The terms pineoblastoma, ectopic intracranial retinoblastoma and ‘trilateral’ retinoblastoma have been used for these growths, which are distinct primary cancers [4,5]. The retinal staging of retinoblastoma is closely correlated with prognosis [4]; however, it has not been possible to ascribe tumour aggressiveness to the possession of the RB1 gene because somatic mutations are present in the sporadic form of the disease, and familial cases are currently picked up earlier in their natural history (in Western countries) as a result of screening programmes and because the inherited form of the disease is multifocal, which is of prognostic importance. This is not the case with the associated pineoblastoma. Sporadic pineoblastoma is another ‘primitive neuroectodermal tumour (PNET)’, and one whose morphology is similar or identical to retinoblastoma, and indeed medulloblastoma; however, this tumour is not as closely aligned with a causal mutation in the RB1 gene and certainly not a major inherited mutation [6]. In 1985, we reported the poor prognosis of trilateral retinoblastoma [5] despite the use of chemoradiotherapy. The survival of the sporadic form of
2004 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
PINEAL PARENCHYMAL TUMOURS: II
245
Fig. 1 – Survival distributions from time of diagnosis in children with pineoblastoma associated with hereditary retinoblastoma and those with the sporadic form of the disease.
pineoblastoma, a rare childhood tumour, has been unclear until recently. An analysis of a recently completed clinical study for PNETs, UKCCSG/SIOP PNET III trial, clearly showed better survival for pineal/pineoblastoma patients compared with other supratentorial patients, after standard platinum-based chemotherapy and radiotherapy [7], an observation of interest in its own right. The results of this study has now afforded the opportunity to compare the outcome of patients with ‘trilateral’ retinoblastoma with patients with sporadic pineoblastoma, and this comparison forms the basis of this report. Materials and Methods
Fourteen children presenting to the National Retinoblastoma service at St Bartholomew’s Hospital, London, U.K., with familial retinoblastoma and associated pineoblastoma, were staged for metastatic disease; those children without such metastases (n=8) were included in this study; four other children with an ectopic intracranial ‘retinoblastoma’ in the suprasellar location were excluded in order to confine the data to a purely pineal location. No patient had uncontrolled ocular retinoblastoma. Therefore, this case report relates to eight consecutive children with nonmetastatic pineal disease at presentation, with an age range of 1–10 years and mean age of 3.5 years. Early data from St Bartholomew’s on the chemoradiotherapy of these patients have previously been published [8]; carboplatin, etoposide and vincristine were used as the combination chemotherapy in most patients [8].
The SIOP PNET III trial (1991–2000) examined the benefit of pre-irradiation chemotherapy with carboplatin, cyclophosphamide, etoposide and vincristine in children aged at least 3 years. The main study was designed principally for children with medulloblastoma; children were randomised to receive sandwich chemotherapy before craniospinal radiotherapy or the radiotherapy alone. Fourteen children were entered into the trial with sporadic pineoblastoma, nine with nonmetastatic disease at presentation. No child had hereditary retinoblastoma. Of the nine non-metastatic disease children, the age range was 3.8–15.0 years, with a mean of 7.7 years. Seven of these received chemotherapy and radiotherapy, and two received radiotherapy alone. In general, both chemotherapy regimens were based on platin chemotherapy and both regimens led to 10-day white cell count nadirs in the region of 0.5109/l or less, although the specifics varied.
Results
Kaplan–Meier survival curves for the two groups of patients from the time of diagnosis of pineoblastoma are shown in Fig. 1. Patients with hereditary retinoblastoma were much more likely to relapse and die of their disease after chemo-radiotherapy than those suffering sporadic pineoblastoma (P=0.002). Three years from diagnosis, five patients with hereditary retinoblastoma had died compared with none with sporadic disease. By 5 years from diagnosis, seven patients with hereditary retinoblastoma had relapsed and died compared with two patients with sporadic pineoblastoma (Fig. 1).
246
CLINICAL ONCOLOGY
Furthermore, relapsing patients with hereditary retinoblastoma relapsed earlier after therapy than compared with those with relapsing sporadic cases (Fig. 1). For both groups of patients, the event-free survival mirrored the overall survival.
Discussion
In the genesis of most adult cancers, it is generally regarded that many steps or mutations are required for oncogenesis (i.e. it is rare to have a true mono-aetiology or single genetic mutation) that can account for any given cancer. The situation may be more straightforward in hereditary retinoblastoma. The normal function of the RB gene is to negatively regulate and thereby control the cell-division cycle (i.e. it is a tumour suppressor gene). In 1971 Knudson [2] proposed a simple model to explain the genesis of multiple bilateral retinoblastomas in the eyes of children with the inherited or heritable form of the disease. All subsequent data support his so-called ‘two hit hypothesis’, which contends that these children are born with an inherited loss of one normal RB gene and acquire a mutation in the second copy (the second ‘hit’), and that this is sufficient to cause tumour formation. In addition, the pineal ‘retinal carpet’ is also at risk for tumour development. In the genetic form of retinoblastoma, a mutation in one RB gene occurs in the germ line, thus only one allele carries the normal gene. During subsequent development, the single copy of the normal RB gene is lost or mutated, and the chance of developing retinoblastoma is then (almost) inevitable. The most common means by which the second copy is lost are whole chromosome loss, large deletions, gene conversion, or both. In the sporadic form of the disease, mutation or loss of both RB genes must occur in the same somatic retinal cell for the disease to occur. Patients with the heritable form of the disease also seem at greatly increased risk of developing second primary tumours particularly osteogenic sarcoma (although this was not a feature that occurred in the current series). Data indicate that such osteosarcomas had become homozygous around the chromosomal locus for RB. Furthermore, these same chromosomal mechanisms, resulting in losses of constitutional heterozygosity, were observed in sporadic osteosarcoma, suggesting a similarity in the pathogenetic causality [9]. Furthermore, aberrations of function or mutations in RB expression have been observed in other cancers, suggesting that RB is pleiotropically active and therefore of greater importance in oncology. Thus, molecular analyses of small-cell lung cancer have revealed RB1 structural abnormalities in 15% of cases [10], and loss of heterozygosity for chromosome 13 has been detected in about 25% of breast cancers [11]. Lesser aberrations in the RB locus, including unmasking mutations, may have a role. Stratton et al. [12] noted the frequent concurrence of p53 gene abnormalities with inactivation of the RB1 susceptibility gene.
Some other inherited oncogenes have been associated with more aggressive phenotypes or behaviour in subsequent cancers. An example of this is the BRCA gene inheritance, which leads to a more aggressive breast cancer phenotype, although there is some debate whether overall survival differs from sporadic breast cancer [13,14]. The prognostic importance of the inherited RB1 mutation in the prognosis of retinoblastoma is, however, not established for reasons that have been discussed above. In a study of 91 curatively resected non-smell cell lung cancer patients, DosakaAkita et al. [15] found no correlation between RB mutations and prognosis, either alone or in combination with p53 or ras mutations; this result differed from results predicted by the authors on the basis of earlier work from their centre. However, in a smaller subanalysis of 19 patients with adenocarcinoma of lung, patients having normal RB and ras-p21 protein expression had improved survival over deficient tumour cases. Given that the same authors found that ras (and p53) genes independently marked for poor prognosis, at present we can conclude that no certain correlation between RB mutation presence and prognosis has been established. In the most common type of PNET of childhood, the cerebellar medulloblastoma, any prognostic import of retinoblastoma RB1 gene mutation also remains an unknown quantity. Recently, Pomeroy et al. [6] probed DNA gene expression by microarray technology to try to find associations between gene expressions and prognosis in medulloblastoma. They found that genes characteristic of cerebellar differentiation correlated with favourable outcome: vesicle coat protein betaNAP, NSCLI, TRKC) and genes encoding extracellular matrix proteins (PLOD lysyl hydroxylase, collagen typeV alpha-1, elastin). By contrast, these genes were underexpressed in poor-prognosis tumours that were dominated by the overexpression of genes related to cell proliferation and metabolism (MYBL2, enolase 1, LDH, HMG1(Y), cytochrome C oxidase) and multidrug resistance (sorcin). Although we believe that such prognostic predictions for tumours, based on microarray gene expression data, will become more important in the future, the comparison of outcome of two groups of patients with histologically identical cancers, but differing in an inherited oncogene, provides another formidable method of comparing the prognostic impact of a particular gene. Although the prognostic implication of the inherited RB1 mutation (as distinct from its causal effect) has been hitherto unproven, it is an important issue throughout oncology, as this gene has now been found in a somatically mutated form in several common cancers, such as lung cancer, breast cancer and osteogenic sarcoma. In lung cancer, Dosaka et al. [15] found concomitant mutations of ras and p53, which clouds interpretation of the prognostic impact of single gene. In soft-tissue sarcoma, p53 may also be mutated concomitantly with another important tumour suppressor
PINEAL PARENCHYMAL TUMOURS: II
gene viz. p53, again clouding the issue about the prognostic (as distinct from causal) importance of the possession of a mutated RB1 gene by itself [12]. This final association is yet to be fully worked out. In inbred mice carrying both RB1 and p53 mutations in their genome, Williams et al. [16] found pineoblastomas, retinal dysplasia, bronchial epithelial hyperplasia and islet cell tumours. The authors concluded that RB1 and p53 mutations co-operate in the oncogenic (causal) process. It may well be that one gene mutation frequently works in combination with others in oncogenesis. Of interest is our observation that pineoblastomas, occurring in the presence of an inherited or mutated RB1 gene in the genome, behave in a more aggressive fashion and respond more poorly to treatment, than in the non-inherited form of the disease. Whether the inherited RB1 gene mutation is acting in association with others or not is unknown, and this will need more careful exploration. Nevertheless, its possession in the genome leads to a more aggressive form of pineoblastoma than occurs sporadically, whatever somatic mutations may have been causal or present in the non-inherited disease. It is remarkable that only one of the eight heritable RB1 mutation cases survived, and that the other seven died with a median survival of only 7 months, compared with an 89% 5-year and 76% 8-year survival for sporadic pineoblastoma cases, respectively. Although the platinum-based chemotherapy and possibly the craniospinal-dose prescriptions of radiotherapy differed in detail and timing between the two groups, the scale of the difference strongly suggests that the possession of a germ-line mutation of the RB1 gene carries with it an aggressive pattern of clinical behaviour. It is clear that future treatment protocols will need to be intensified for individuals with the hereditary form of the disease. Acknowledgements. The authors are very grateful to Dr Robert Hills of Birmingham University Clinical Trials Unit for statistical analysis, to Claire Weston (statistician to the UKCCSG) and Kath Robinson (PNET III trial co-ordinator).
247
References 1 Gallie BL, Squire JA, Goddard A, et al. Mechanism of oncogenesis in retinoblastoma. Lab Investig 1990;62:394–408. 2 Knudson AG. Mutation and cancer: a statistical study of retinoblastoma. Proc Natl Acad Sci USA 1971;68:820–823. 3 Onadim Z, Woolford AJ, Kingston JE, et al. The RB1 gene mutation in a child with ectopic intracranial retinoblastoma. Br J Cancer 1997;76:1405–1409. 4 Bader JL, Meadows AT, Zimmerman LE, et al. Bilateral retinoblastoma with ectopic intracranial retinoblastoma: ‘trilateral’ retinoblastoma. Cancer Genet Cytogenet 1982;5:203–213. 5 Kingston JE, Plowman PN, Hungerford JL. Ectopic intracranial retinoblastoma in childhood. Br J Ophthalmol 1985;69:742–748. 6 Pomeroy SL, Tamayo P, Gaasenbeek M, et al. Prediction of central nervous system embryonal tumour outcome based on gene expression. Nature 2002;415:436–442. 7 Pizer B, Taylor R, Weston C, et al. Analysis of patients with supratentorial PNET entered into the SIOP PNT 111 trial [abstract 0027]. Proc Soc Int Pediatr Oncol Sept 2002; Oporto. 8 Douek E, Plowman PN, Kingston JE. Platinum based chemotherapy for recurrent non-gliomatous brain tumours in young patients. J Neurol Neurosurg Psychiatr 1991;54:722–725. 9 Hansen MF, Koufos A, Gallie BL, et al. Osteosarcoma and retinoblastoma: a shared chromosomal mechanism revealing a shared chromosomal predisposition. Proc Natl Acad Sci USA 1985;82:6216–6223. 10 Harbour J, Lai SL, Whang Peng J, et al. Abnormalities in the structure and expression of the retinoblastoma genein SCLC. Science 1988;242:263–269. 11 Bookstein R, Lee EYHP, Peccei A, et al. Human retinoblastoma gene: long range mapping and analysis of its deletion in a breast cancer cell line. Mol Cell Biol 1989;9:1628–1635. 12 Stratton MR, Moss S, Warren W, et al. Mutation of the p53 gene in soft tissue sarcomas: association with abnormalities of the RB1 gene. Oncogene 1990;5:1297–1301. 13 Verhoog LC, Brekelmans CTM, Stratton MR, Seynaeve C, et al. Survival and tumour characteristics of breast cancer patients with germline mutations of BRCA1. Lancet 1998;351:316–321. 14 Hafty BG, Harrold E, Khan AJ, et al. Outcome of conservatively managed early onset breast cancer by BRCA1/2 status. Lancet 2002;359:1471–1477. 15 Dosaka-Akita H, Hu SX, Fujino M, et al. Altered retinoblastoma protein expression in non-small cell lung cancer: its synergistic effects with altered ras and p53 status on prognosis. Cancer 1997; 79:1329–1337. 16 Williams BO, Remington L, Albert DM, et al. Co-operative tumorigenic effects of germline mutations in Rb and p53. Nat Genet 1994;7:480–484.
Case Report Pineal Parenchymal Tumours: II On the Aggressive Behaviour of Pineoblastoma in Patients with an Inherited Mutation of the RB1 Gene P. N. Plowman*, B. Pizer†, J. E. Kingston‡ *Department of Clinical Oncology, St Bartholomew’s Hospital, London, U.K.; †Department of Paediatric Oncology, Alder Hey Hospital, Liverpool, U.K.; ‡Department of Paediatric Oncology, Barts and the London NHS Trust, London U.K. ABSTRACT: This report relates to a retrospective analysis of two non-randomised cohorts of patients with pineoblastoma, with some differences in presenting features and treatment characteristics. We have identified a large difference in survival depending on the possession or otherwise of the mutated RB (retinoblastoma) gene in the genome/karyotype. Eight children with familial retinoblastoma (non-metastatic at presentation) developed pineoblastoma and were treated by chemotherapy and radiotherapy. The survival of these patients was compared with the survival of nine non-metastatic sporadic cases of pineoblastoma similarly staged and treated. One out of eight children having the RB mutation in the genome survived compared with seven out of nine in the group with sporadic pineoblastoma (P=0.002). It is suggested that the inheritance of the mutated retinoblastoma gene is not only causal in the generation of this tumour type but, in a way that is yet to be defined, renders such tumours more aggressive or less responsive to therapy. With the current interest in the role of RB mutations in other cancers (where the prognostic import of single genes is less easily identified), this observation may have wider relevance. Plowman P. N. et al. (2004). Clinical Oncology 16, 244–247 2004 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved. Key words: Pineal parenchymal tumours, pineoblastoma, RB1 gene Received: 4 August 2003
Revised: 5 November 2003
Introduction
Retinoblastoma is a well-researched retinal cancer afflicting young children. Genetic (hereditary) and nongenetic (sporadic) forms occur. The discovery of the retinoblastoma gene RB1, on chromosome 13q14 [1] led to the validation of the Knudson ‘two hit hypothesis’ of this cancer’s causation at the molecular level [2]. The RB1 gene encodes a ubiquitously expressed nuclear protein whose level of phosphorylation importantly varies throughout the cell cycle and thereby influences cell-cycle progression in a negative or inhibitory manner. Mutations of this RB1 gene, when carried in the genome (as shown in hereditary cases) [3], lead to multiple retinal tumours (both retinae) occurring at an early age. Furthermore, the pineal gland whose parenchymal cells were once, specifically in primitive fish the ‘retinal carpet’ of the pineal eye, is also at risk, with pineal tumours identical (in all respects other than location) to retinoblastoma occurring in 4% of children with the Author for correspondence: P. N. Plowman, Department of Clinical Oncology, St Bartholomew’s Hospital, London EC1, U.K.; E-mail: [email protected] 0936-6555/04/040244+4 $30.00/0
Accepted: 17 December 2003
inherited form of the disease [4,5]. The terms pineoblastoma, ectopic intracranial retinoblastoma and ‘trilateral’ retinoblastoma have been used for these growths, which are distinct primary cancers [4,5]. The retinal staging of retinoblastoma is closely correlated with prognosis [4]; however, it has not been possible to ascribe tumour aggressiveness to the possession of the RB1 gene because somatic mutations are present in the sporadic form of the disease, and familial cases are currently picked up earlier in their natural history (in Western countries) as a result of screening programmes and because the inherited form of the disease is multifocal, which is of prognostic importance. This is not the case with the associated pineoblastoma. Sporadic pineoblastoma is another ‘primitive neuroectodermal tumour (PNET)’, and one whose morphology is similar or identical to retinoblastoma, and indeed medulloblastoma; however, this tumour is not as closely aligned with a causal mutation in the RB1 gene and certainly not a major inherited mutation [6]. In 1985, we reported the poor prognosis of trilateral retinoblastoma [5] despite the use of chemoradiotherapy. The survival of the sporadic form of
2004 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
PINEAL PARENCHYMAL TUMOURS: II
245
Fig. 1 – Survival distributions from time of diagnosis in children with pineoblastoma associated with hereditary retinoblastoma and those with the sporadic form of the disease.
pineoblastoma, a rare childhood tumour, has been unclear until recently. An analysis of a recently completed clinical study for PNETs, UKCCSG/SIOP PNET III trial, clearly showed better survival for pineal/pineoblastoma patients compared with other supratentorial patients, after standard platinum-based chemotherapy and radiotherapy [7], an observation of interest in its own right. The results of this study has now afforded the opportunity to compare the outcome of patients with ‘trilateral’ retinoblastoma with patients with sporadic pineoblastoma, and this comparison forms the basis of this report. Materials and Methods
Fourteen children presenting to the National Retinoblastoma service at St Bartholomew’s Hospital, London, U.K., with familial retinoblastoma and associated pineoblastoma, were staged for metastatic disease; those children without such metastases (n=8) were included in this study; four other children with an ectopic intracranial ‘retinoblastoma’ in the suprasellar location were excluded in order to confine the data to a purely pineal location. No patient had uncontrolled ocular retinoblastoma. Therefore, this case report relates to eight consecutive children with nonmetastatic pineal disease at presentation, with an age range of 1–10 years and mean age of 3.5 years. Early data from St Bartholomew’s on the chemoradiotherapy of these patients have previously been published [8]; carboplatin, etoposide and vincristine were used as the combination chemotherapy in most patients [8].
The SIOP PNET III trial (1991–2000) examined the benefit of pre-irradiation chemotherapy with carboplatin, cyclophosphamide, etoposide and vincristine in children aged at least 3 years. The main study was designed principally for children with medulloblastoma; children were randomised to receive sandwich chemotherapy before craniospinal radiotherapy or the radiotherapy alone. Fourteen children were entered into the trial with sporadic pineoblastoma, nine with nonmetastatic disease at presentation. No child had hereditary retinoblastoma. Of the nine non-metastatic disease children, the age range was 3.8–15.0 years, with a mean of 7.7 years. Seven of these received chemotherapy and radiotherapy, and two received radiotherapy alone. In general, both chemotherapy regimens were based on platin chemotherapy and both regimens led to 10-day white cell count nadirs in the region of 0.5109/l or less, although the specifics varied.
Results
Kaplan–Meier survival curves for the two groups of patients from the time of diagnosis of pineoblastoma are shown in Fig. 1. Patients with hereditary retinoblastoma were much more likely to relapse and die of their disease after chemo-radiotherapy than those suffering sporadic pineoblastoma (P=0.002). Three years from diagnosis, five patients with hereditary retinoblastoma had died compared with none with sporadic disease. By 5 years from diagnosis, seven patients with hereditary retinoblastoma had relapsed and died compared with two patients with sporadic pineoblastoma (Fig. 1).
246
CLINICAL ONCOLOGY
Furthermore, relapsing patients with hereditary retinoblastoma relapsed earlier after therapy than compared with those with relapsing sporadic cases (Fig. 1). For both groups of patients, the event-free survival mirrored the overall survival.
Discussion
In the genesis of most adult cancers, it is generally regarded that many steps or mutations are required for oncogenesis (i.e. it is rare to have a true mono-aetiology or single genetic mutation) that can account for any given cancer. The situation may be more straightforward in hereditary retinoblastoma. The normal function of the RB gene is to negatively regulate and thereby control the cell-division cycle (i.e. it is a tumour suppressor gene). In 1971 Knudson [2] proposed a simple model to explain the genesis of multiple bilateral retinoblastomas in the eyes of children with the inherited or heritable form of the disease. All subsequent data support his so-called ‘two hit hypothesis’, which contends that these children are born with an inherited loss of one normal RB gene and acquire a mutation in the second copy (the second ‘hit’), and that this is sufficient to cause tumour formation. In addition, the pineal ‘retinal carpet’ is also at risk for tumour development. In the genetic form of retinoblastoma, a mutation in one RB gene occurs in the germ line, thus only one allele carries the normal gene. During subsequent development, the single copy of the normal RB gene is lost or mutated, and the chance of developing retinoblastoma is then (almost) inevitable. The most common means by which the second copy is lost are whole chromosome loss, large deletions, gene conversion, or both. In the sporadic form of the disease, mutation or loss of both RB genes must occur in the same somatic retinal cell for the disease to occur. Patients with the heritable form of the disease also seem at greatly increased risk of developing second primary tumours particularly osteogenic sarcoma (although this was not a feature that occurred in the current series). Data indicate that such osteosarcomas had become homozygous around the chromosomal locus for RB. Furthermore, these same chromosomal mechanisms, resulting in losses of constitutional heterozygosity, were observed in sporadic osteosarcoma, suggesting a similarity in the pathogenetic causality [9]. Furthermore, aberrations of function or mutations in RB expression have been observed in other cancers, suggesting that RB is pleiotropically active and therefore of greater importance in oncology. Thus, molecular analyses of small-cell lung cancer have revealed RB1 structural abnormalities in 15% of cases [10], and loss of heterozygosity for chromosome 13 has been detected in about 25% of breast cancers [11]. Lesser aberrations in the RB locus, including unmasking mutations, may have a role. Stratton et al. [12] noted the frequent concurrence of p53 gene abnormalities with inactivation of the RB1 susceptibility gene.
Some other inherited oncogenes have been associated with more aggressive phenotypes or behaviour in subsequent cancers. An example of this is the BRCA gene inheritance, which leads to a more aggressive breast cancer phenotype, although there is some debate whether overall survival differs from sporadic breast cancer [13,14]. The prognostic importance of the inherited RB1 mutation in the prognosis of retinoblastoma is, however, not established for reasons that have been discussed above. In a study of 91 curatively resected non-smell cell lung cancer patients, DosakaAkita et al. [15] found no correlation between RB mutations and prognosis, either alone or in combination with p53 or ras mutations; this result differed from results predicted by the authors on the basis of earlier work from their centre. However, in a smaller subanalysis of 19 patients with adenocarcinoma of lung, patients having normal RB and ras-p21 protein expression had improved survival over deficient tumour cases. Given that the same authors found that ras (and p53) genes independently marked for poor prognosis, at present we can conclude that no certain correlation between RB mutation presence and prognosis has been established. In the most common type of PNET of childhood, the cerebellar medulloblastoma, any prognostic import of retinoblastoma RB1 gene mutation also remains an unknown quantity. Recently, Pomeroy et al. [6] probed DNA gene expression by microarray technology to try to find associations between gene expressions and prognosis in medulloblastoma. They found that genes characteristic of cerebellar differentiation correlated with favourable outcome: vesicle coat protein betaNAP, NSCLI, TRKC) and genes encoding extracellular matrix proteins (PLOD lysyl hydroxylase, collagen typeV alpha-1, elastin). By contrast, these genes were underexpressed in poor-prognosis tumours that were dominated by the overexpression of genes related to cell proliferation and metabolism (MYBL2, enolase 1, LDH, HMG1(Y), cytochrome C oxidase) and multidrug resistance (sorcin). Although we believe that such prognostic predictions for tumours, based on microarray gene expression data, will become more important in the future, the comparison of outcome of two groups of patients with histologically identical cancers, but differing in an inherited oncogene, provides another formidable method of comparing the prognostic impact of a particular gene. Although the prognostic implication of the inherited RB1 mutation (as distinct from its causal effect) has been hitherto unproven, it is an important issue throughout oncology, as this gene has now been found in a somatically mutated form in several common cancers, such as lung cancer, breast cancer and osteogenic sarcoma. In lung cancer, Dosaka et al. [15] found concomitant mutations of ras and p53, which clouds interpretation of the prognostic impact of single gene. In soft-tissue sarcoma, p53 may also be mutated concomitantly with another important tumour suppressor
PINEAL PARENCHYMAL TUMOURS: II
gene viz. p53, again clouding the issue about the prognostic (as distinct from causal) importance of the possession of a mutated RB1 gene by itself [12]. This final association is yet to be fully worked out. In inbred mice carrying both RB1 and p53 mutations in their genome, Williams et al. [16] found pineoblastomas, retinal dysplasia, bronchial epithelial hyperplasia and islet cell tumours. The authors concluded that RB1 and p53 mutations co-operate in the oncogenic (causal) process. It may well be that one gene mutation frequently works in combination with others in oncogenesis. Of interest is our observation that pineoblastomas, occurring in the presence of an inherited or mutated RB1 gene in the genome, behave in a more aggressive fashion and respond more poorly to treatment, than in the non-inherited form of the disease. Whether the inherited RB1 gene mutation is acting in association with others or not is unknown, and this will need more careful exploration. Nevertheless, its possession in the genome leads to a more aggressive form of pineoblastoma than occurs sporadically, whatever somatic mutations may have been causal or present in the non-inherited disease. It is remarkable that only one of the eight heritable RB1 mutation cases survived, and that the other seven died with a median survival of only 7 months, compared with an 89% 5-year and 76% 8-year survival for sporadic pineoblastoma cases, respectively. Although the platinum-based chemotherapy and possibly the craniospinal-dose prescriptions of radiotherapy differed in detail and timing between the two groups, the scale of the difference strongly suggests that the possession of a germ-line mutation of the RB1 gene carries with it an aggressive pattern of clinical behaviour. It is clear that future treatment protocols will need to be intensified for individuals with the hereditary form of the disease. Acknowledgements. The authors are very grateful to Dr Robert Hills of Birmingham University Clinical Trials Unit for statistical analysis, to Claire Weston (statistician to the UKCCSG) and Kath Robinson (PNET III trial co-ordinator).
247
References 1 Gallie BL, Squire JA, Goddard A, et al. Mechanism of oncogenesis in retinoblastoma. Lab Investig 1990;62:394–408. 2 Knudson AG. Mutation and cancer: a statistical study of retinoblastoma. Proc Natl Acad Sci USA 1971;68:820–823. 3 Onadim Z, Woolford AJ, Kingston JE, et al. The RB1 gene mutation in a child with ectopic intracranial retinoblastoma. Br J Cancer 1997;76:1405–1409. 4 Bader JL, Meadows AT, Zimmerman LE, et al. Bilateral retinoblastoma with ectopic intracranial retinoblastoma: ‘trilateral’ retinoblastoma. Cancer Genet Cytogenet 1982;5:203–213. 5 Kingston JE, Plowman PN, Hungerford JL. Ectopic intracranial retinoblastoma in childhood. Br J Ophthalmol 1985;69:742–748. 6 Pomeroy SL, Tamayo P, Gaasenbeek M, et al. Prediction of central nervous system embryonal tumour outcome based on gene expression. Nature 2002;415:436–442. 7 Pizer B, Taylor R, Weston C, et al. Analysis of patients with supratentorial PNET entered into the SIOP PNT 111 trial [abstract 0027]. Proc Soc Int Pediatr Oncol Sept 2002; Oporto. 8 Douek E, Plowman PN, Kingston JE. Platinum based chemotherapy for recurrent non-gliomatous brain tumours in young patients. J Neurol Neurosurg Psychiatr 1991;54:722–725. 9 Hansen MF, Koufos A, Gallie BL, et al. Osteosarcoma and retinoblastoma: a shared chromosomal mechanism revealing a shared chromosomal predisposition. Proc Natl Acad Sci USA 1985;82:6216–6223. 10 Harbour J, Lai SL, Whang Peng J, et al. Abnormalities in the structure and expression of the retinoblastoma genein SCLC. Science 1988;242:263–269. 11 Bookstein R, Lee EYHP, Peccei A, et al. Human retinoblastoma gene: long range mapping and analysis of its deletion in a breast cancer cell line. Mol Cell Biol 1989;9:1628–1635. 12 Stratton MR, Moss S, Warren W, et al. Mutation of the p53 gene in soft tissue sarcomas: association with abnormalities of the RB1 gene. Oncogene 1990;5:1297–1301. 13 Verhoog LC, Brekelmans CTM, Stratton MR, Seynaeve C, et al. Survival and tumour characteristics of breast cancer patients with germline mutations of BRCA1. Lancet 1998;351:316–321. 14 Hafty BG, Harrold E, Khan AJ, et al. Outcome of conservatively managed early onset breast cancer by BRCA1/2 status. Lancet 2002;359:1471–1477. 15 Dosaka-Akita H, Hu SX, Fujino M, et al. Altered retinoblastoma protein expression in non-small cell lung cancer: its synergistic effects with altered ras and p53 status on prognosis. Cancer 1997; 79:1329–1337. 16 Williams BO, Remington L, Albert DM, et al. Co-operative tumorigenic effects of germline mutations in Rb and p53. Nat Genet 1994;7:480–484.