LORYN NSELLNER, 1* EDWARD EDKINS, 1,2 AND NICHOLAS SMITH 3 ABSTRACT INTRODUCTION

Transcription

1 Pediatric and Developmental Pathology 0, , 0000 DOI: ª 2006 Society for Pediatric Pathology Screening for RB1 Mutations in Tumor Tissue Using Denaturing High Performance Liquid Chromatography, Multiplex Ligation-Dependent Probe Amplification, and Loss of Heterozygosity Analysis LORYN NSELLNER, 1* EDWARD EDKINS, 1,2 AND NICHOLAS SMITH 3 1 Molecular Genetics, Princess Margaret Hospital for Children, Roberts Road, Subiaco 6009, Perth, Western Australia 2 Centre for Human Genetics, Edith Cowan University, Perth, Western Australia 3 Department of Pathology, Princess Margaret Hospital for Children, Roberts Road, Subiaco 6009, Perth, Western Australia Received October 21, 2005; accepted November 3, 2005; published online Month, ABSTRACT?1 Retinoblastoma is a malignant retinal neoplasm arising in infancy as a result of inactivating mutations in both alleles of the retinoblastoma susceptibility gene, RB1. Identification of the causative RB1 mutations in a patient assists in the clinical management of the affected patient and risk assessment of family members, principally on the basis of whether there is a germline mutation. In this paper, we describe our experience with molecular analysis of RB1 mutations in tumor and nontumor samples from 18 retinoblastoma patients, using multiplex ligation dependent probe amplification (MLPA) to detect large deletions or duplications, microsatellite analysis to detect loss of heterozygosity (LOH), and denaturing high performance liquid chromatography (D-HPLC) analysis to detect point mutations and small insertions or deletions. We found LOH in 71% of all cases, and 83% of these were due to acquired isodisomy rather than chromosomal deletions. Small mutations identified by D-HPLC accounted for 78% of the non-loh mutations, and large deletions/duplications detected by MLPA accounted for the remaining 22%. We give the first report of a large, multiexon duplication in RB1 of exons 8 to 18. Key words: duplication, germline, LOH, MLPA, mutation, retinoblastoma *Corresponding author, geyerl01@tartarus.uwa.edu.au INTRODUCTION Retinoblastoma is a malignant retinal neoplasm with a reported incidence of between 1 in and 1 in live births [1]. It arises usually in infancy as a result of inactivating mutations in both alleles of the retinoblastoma susceptibility gene, RB1. Mutations in this gene may be germline, having been inherited from a parent or arising in gametogenesis, or somatic, occurring only in the tumor. In approximately 60% of cases, both mutations are somatic events (and are therefore nonhereditary) and the tumors are generally unilateral. In the remaining 40% of cases, one of the mutations is germline. This often results in multiple tumors, commonly bilateral, and the mutation can be passed on to offspring (hereditary retinoblastoma), although the penetrance can depend on the type of mutation and its functional consequences. Identification of the causative RB1 mutations in a patient assists in the clinical management of the affected patient and risk assessment of family members, principally on the basis of whether there is a germline mutation. If a patient is found not to have a germline mutation, the risk to family

2 members, subsequent siblings, and offspring is low. If a patient is found to have a germline mutation, parents are tested for carrier status (approximately 10% of carriers are asymptomatic), and if positive, they and any other offspring are at high risk. Even if parents are found not to carry the mutation, siblings are still at increased risk due to the possibility of germinal mosaicism in the parents. Children at risk of developing retinoblastoma must undergo repeated eye examinations under anesthetic (EUAs) in order to diagnose and treat tumors as early as possible to improve the outcome. The frequency and type of examinations depend on how closely related the child is to the index case, and conventionally ranges from 8 EUAs (performed on children under 4 y) plus 13 clinic visits, to 17 clinic visits with no EUA, until the child reaches 10 y old [2]. If an individual is known not to carry a germline mutation, these examinations can be avoided, at a considerable cost saving. Failure to detect a germline mutation when screening a nontumor sample from a retinoblastoma patient without a known familial mutation is not sufficient to rule out the presence of such a mutation, since 11% to 17% of mutations are not detectable using current methods such as heteroduplex analysis, SSCP, quantitative multiplex PCR, and sequencing [2,3]. Only when both mutations are identified in the tumor, and each of those is shown not to be present in the germline, can one confidently say a germline mutation is not present. This is the approach recommended by the European Molecular Genetics Quality Network [4]. We describe our experience with molecular analysis of RB1 mutations in tumor and nontumor samples from 18 retinoblastoma patients, using multiplex ligation-dependent probe amplification (MLPA) to detect large deletions or duplications, microsatellite analysis to detect loss of heterozygosity (LOH), and denaturing high performance liquid chromatography (D-HPLC) analysis to detect point mutations and small insertions or deletions. We report a high rate of LOH due to acquired isodisomy rather than chromosomal deletions and give the first report of a large, multiexon duplication. METHODS Patients and specimens All retinoblastoma patients identified in Perth, Western Australia, over a 10- year period (1995 to L.N. SELLNER ET AL. 2004) were included. The total number of patients was 18, representing an incidence of approximately 1 in live births. None of the cases were familial, and only 1 had bilateral disease. Fresh frozen tumor sample was available for analysis in 18 cases, and nontumor DNA available in 17 of these (WBC in 14 cases, uninvolved formalin-fixed paraffin-embedded tissue in 3 cases). DNA was extracted from tumor samples using the QIAamp DNA Micro kit (Qiagen, Hilden, Germany), from WBC using the GFX Genomic Blood DNA Purification Kit (Amersham Biosciences, Buckinghamshire, England), and from formalin-fixed paraffin-embedded tissue by dewaxing 20-lm sections by successive washes in xylene, 95% EtOH, and 70% EtOH; drying the tissue; and then digesting overnight in 10 mm Tris-HCl (ph 8.3), 50 mm KCl, 1.5 mm MgCl 2, 0.5% Tween 20, and 2 mg/ml proteinase K for 18 h at 558C, followed by heat inactivation of the proteinase K by incubation at 958C for 10 min. MLPA MLPA was performed as per manufacturer s instructions using the RB1 probe kit (MRC- Holland, Amsterdam). Briefly, 200 ng of genomic DNA in a volume of 5 ll was heated to 958C for 5 min, cooled, and then mixed with MLPA buffer and probe mix. The mix was then heated to 958C for 5 min and then incubated at 608C for 16 h for probe hybridization. Ligation buffer and ligase were then added and ligation was performed for 15 min at 548C. PCR was then performed on the ligation products as per manufacturer s instructions using a Beckman D4 labeled primer. PCR product (1 ll) was then mixed with Beckman fluorescent size standard and then run on the CEQ8000 Automated Sequencer (Beckman Coulter, Fullerton, CA, USA). After fragment analysis, the peaks corresponding to specific exons were identified and the relative peak heights were compared to those of a control sample to determine which peaks had increased or decreased, thus representing duplication or deletion of the corresponding exon. Loss of heterozygosity Paired DNA samples from tumor and WBC (or nontumor paraffin-embedded tissue where blood was not available) were PCR amplified at microsatellite sites on chromosome 13 using the

3 Table 1 Primer sequences and D-HPLC temperatures used for PCR and D-HPLC analysis Exon Forward primer Reverse primer Size D-HPLC temperature 1 GTG CGC GCG CGT CGT CCT CC GGC CCC TGG CGA GGA CGG GTC ACA GTA GTG TTA TGT GCA AAC TA TTA GCA GAG GTA AAT TTC CTC TG , 53 3 TAA CAT AGT ATC CAG TGT GTG A ATT TCC TTT TAT GGC AGA GG , 54 4 GTA GTG ATT TGA TGT AGA GCT G CAG AGT GTA ACC CTA ATA AAA TG AGC ATG AGA AAA CTA CTA TGA CT AAC CCT AAC TAT CAA GAT GTT TG , 53 6 CAC CCA AAA GAT ATA TCT GGA AA GGA ATT TAG TCC AAA GGA ATG , 52 7 TCT CAT ACA AAG ATC TGA ATC TC GAC ATT CAA TAA GCA ACT GCT GA TTG TAG TAG ATA TGG ATG AA ATC TAA ATC TAC TTT AAC TG , 56 9 TGC ATT GTT CAA GAG TCA AGA AAT TAT CCT CCC TCC ACA GTC TGT GCT GAG AGA TGT AAT GAC CTA CCT ATA TCA GTA TCA ACC TA GAG ACA ACA GAA GCA TTA TAC TG ATC TGA AAC ACT ATA AAG CCA TG , AGA TAC ATT TAA CTT GGG AGA TTG CTA CAT GTT AGA TAG GAG ATT AG GTA TCC TCG ACA TTG ATT TCT G TCT ATA GTA CCA CGA ATT ACA ATG GTG ATT TTC TAA AAT AGC AGG CTC TTT TAG TAG AGA CAG GGT TTC AC , 55, AAT GCT GAC ACA AAT AAG GTT TC GAT CTA AAA TAA GCA TTC CTT CTC , 53, CCA AAA AAA TAC CTA GCT CAA GG GTT AAG AAA CAC CTC TCA CTA AC , GGA AAA TTA TGC TTA CTA ATG TGG GCA GTT TGA ATG GTC AAC ATA AC , CAA CTT GAA ATG AAG ACT TTT CC TAG TTT CAG AGT CCA TGC TC , 54, GTA ATT CAA AAT GAA CAG TAA AAA TGA C AGT AAG TAG GGA GGA GAG AA , GAT TAA ACC TTT CTT TTT TGA GGC TA CCT ATG TTA TGT TAT GGA TAT GG , CTT TAT AAT ATG TGC TTC TTA CCA G TTT TGG TGG ACC CAT TAC ATT AG , ATG TAA TGG GTC CAC CAA AAC GAT CAA AAT AAT CCC CCT CTC AT , GAT GTA TTT ATG CTC ATC TCT GC GGA TGA GGT GTT TGA ATA ACT G TTG AGG TTG CTA ACT ATG AAA CAC GAT TCC CCA GAT GAC CAT , GTC ATC GAA AGC ATC ATA GTT AC CGA AAA GAC TTC TTG CAG TG , ATC AAT GCT GTT AAC AGT TCT TC TGT GAG AGA CAA TGA ATC C following primer pairs, of which one primer in each set was labeled with WellRED fluorescent?2 dye: D13S115 forward 5 0 -tct tag ctg ctg gtg gtg g- 3 0, D13S115 reverse 5 0 -tgt aag gag aga gag att tcg aca-3 0, D13S153 forward 5 0 -aca gaa atc ata ttt acc agg ac-3 0, D13S153 reverse 5 0 -cag cag tga agg tct aag cc-3 0, D13S292 forward 5 0 -taa tgg cgg acc atg c-3 0, D13S292 reverse 5 0 -ttt gac act ttc caa gtt gc Reaction conditions used 0.4 U of Platinum Taq DNA polymerase (Invitrogen, Carlsbad, CA, USA), 2 mm MgCl 2, 50 ng of each primer, 200 lm dntp, 20 mm Tris-HCl, ph 8.4, and 50 mm KCl, and the cycling parameters were 958C for 2 min, followed by 35 cycles of 958C for 30 s, 558C for 1 min, 728C for 2 min, and then 728C for 7 min. PCR products (0.5 ll) were run on the CEQ8000 Automated Sequencer (Beckman Coulter), and fragment analysis was performed. PCR and D-HPLC Exons 1 to 27 of the RB1 gene were amplified from tumor DNA using primers listed in Table 1. PCR was performed using 0.5 U of AmpliTaq Gold (Roche Molecular Systems, Branchburg, NJ, USA), 2 mm MgCl 2, 50 ng of each primer, 200 lm dntp, 10 mm Tris-HCl, ph 8.3, and 50 mm KCl, and the cycling parameters were 948C for 12 min and then 12 cycles of 958C for 15 s, 15 s at annealing temperature (with annealing temperature starting at 648C and reducing by 0.58C per cycle), 728C for 55 s, then 30 cycles of 948C for 15 s, 508C for 15 s, 728C for 55 s, followed by a final extension of 728C for 7 min. For tumor samples that exhibited genomic deletions by MLPA or LOH, patient PCR products for each exon were mixed with PCR products from a normal control to allow heteroduplex formation. PCR products were then heated to 948C and cooled to 658C, with 4- min holds every 58C. PCR products were injected onto a DNA separation column using a Varian Helix D-HPLC system (Varian, Walnut Creek, CA, USA) and separated over 9 min with a linear acetonitrile gradient. The temperatures required for optimal resolution of homoduplexes and heteroduplexes were predicted using the Stanford Genome Technology Centre Melt Program ( insertion.stanford.edu/dhplc.html). Chromatograms generated from patient samples were SCREENING FOR RB1 MUTATIONS IN TUMOR

4 Table 2 Patient Patient and mutation details Age at surgery Presentation Tumor mutations Mechanism of LOH Germline mutations 1 4 y Unilateral [c.1363c.t p.arg455x] þ NA [c.1-?_1695 þ?del] þ [¼] [c.1-?_1695 þ?del] 2 12 mo Bilateral [LOH] þ [c.1891c.t p.gln631x] Acquired isodisomy [c.1891c.t p.gln631x] þ [¼] 3 22 mo Unilateral [c.763c.t p.arg255x] þ NA [¼] þ [¼] [c.1-?_2787 þ?del] 4 12 mo Unilateral [c.958c.t p.arg320x] þ NA [¼] þ [¼] [c.2520 þ 5G.T] 5 6 y Unilateral [LOH] þ [c.dup719-?_1814 þ?] Acquired isodisomy [¼] þ [¼] 6 3 y Unilateral [LOH] þ [c.1399c.t p.arg467x] Acquired isodisomy [¼] þ [¼] 7 15 mo Unilateral [LOH] þ [c.2370c.a p.tyr790x] Acquired isodisomy [¼] þ [¼] y Unilateral [LOH] þ [c.1498 þ 1G.A] Hemizygosity [¼] þ [¼] 9 17 mo Unilateral [LOH] þ [c.1027_1028delct] Acquired isodisomy [¼] þ [¼] mo Unilateral [LOH] þ [c.1723c.t p.gly575x] Acquired isodisomy [¼] þ [¼] mo Unilateral [c.1-?_2787 þ?del] þ [?] NA [¼] þ [?] 12 6 mo Unilateral [LOH] þ [?] Hemizygosity 13 9 mo Unilateral [LOH] þ [?] Acquired isodisomy y Unilateral [LOH] þ [?] Acquired isodisomy 15 2 mo Unilateral [?] þ [?] mo Unilateral [?] þ [?] 17 4 mo Unilateral [?] þ [?] 18 4 y Unilateral [?] þ [?] [?] indicates mutation not identified; [¼], mutation not present (wild-type allele). compared to a normal control to identify exons with heteroduplexes and hence possible mutations. Sequencing PCR products were pretreated with 0.4 U of shrimp alkaline phosphatase and 2 U of exonuclease 1 (USB Corporation, Cleveland, OH, USA) per 1 ll of PCR product for 15 min at 378C and then 15 min at 808C. Sequencing reactions were then performed using the CEQ DTCS Quick Start Kit (Beckman Coulter) as per the manufacturer s recommendations. The thermal cycling program was 40 cycles of 968C for 20 s, 508C for 20 s, and 608C for 4 min. Samples were then ethanol precipitated, dried, and dissolved in 25 ll of formamide Sample Loading Solution. Sequencing products were then run on the CEQ2000 Automated Sequencer (Beckman Coulter). All mutations are described as per the recommendations at using the reference sequence GenBank accession number NM_ with the A of the ATG initiation codon as nucleotide þ1. RESULTS MLPA Changes in exon copy number were identified in 4 of the 18 cases (22%) (Table 2). This represented L.N. SELLNER ET AL. 29% of all non LOH-associated mutations found. This is similar to a previously reported rate of 21%, using a technique of quantitative multiplex PCR [5]. Two of these were deletions of the entire gene: 1 was a deletion of exons 1 through 17, and 1 was a duplication of exons 8 through 18 (Fig. 1). The duplication was confirmed by another laboratory using the quantitative multiplex PCR technique. The deletion of exons 1 through 17 was found to be present in the germline DNA from the patient; the other alterations were somatic. Loss of heterozygosity Paired tumor and nontumor samples were available from 17 of the 18 patients. There were no patients who were uninformative at all 3 loci tested. LOH was demonstrated for at least 1 locus in 12 of these 17 cases (71%) (Table 2). Examples are shown in Figure 2. This is comparable to other studies that have assessed LOH directly and found LOH in 60% to 64% of all tumors [6 8]. There were 4 cases in which no mutations could be identified (including the 1 case for which no normal issue was available), and therefore LOH was shown in 12 of 14 cases in which mutations were identified (86%). This compares to another study in which LOH was found in 69% of mutation-positive

5 Figure 1 MLPA analysis of patient 5 showing duplication of exons Top. Germline DNA. Bottom. Tumor DNA with duplicated exons indicated by arrows. tumors [6]. In 3 cases, no mutation other than LOH was identified. In 2 of the 12 cases (17%), LOH was due to a large deletion in 1 chromosome (as determined by loss of RB1 exons in MLPA); in the remaining 10 cases (83%), no reduction in gene copy number was shown by MLPA, indicating that LOH represented acquired homozygosity (isodisomy) rather than hemizygosity. A previous study has found LOH to be due to deletion in 6% of cases and isodisomy in 94% of cases [8]. D-HPLC and sequencing A total of 10 mutations were identified by detection of heteroduplexes on D-HPLC and sequencing of the appropriate exons. Seven of these were nonsense mutations, 2 were splice site mutations, and 1 was a 2-base deletion resulting in a frame shift (Table 2), which is a similar distribution to those previously reported [9]. Three of these mutations have not been previously described: the splice site mutation c.2520þ5g.t, the nonsense mutation c.2370c.a p.tyr790x, and the 2-base deletion c.1027_1028delct. Only 1 of the 10 mutations was found to be present in the patient s germline DNA. DISCUSSION In this study, we have investigated 18 retinoblastoma patients for RB1 mutations by analysis of tumor tissue with the aim of identifying mutations in both alleles, such that germline DNA can then be tested for the identified mutations to categorically determine germline status. The majority of retinoblastoma cases are sporadic and unilateral, SCREENING FOR RB1 MUTATIONS IN TUMOR

6 Figure 2 LOH in 2 patients. Top. Patients are heterozygous in the germline DNA at the given loci and homozygous in the tumor. In patient 8, LOH is due to acquired isodisomy; in patient 11, LOH is due to a large deletion of 1 allele. and by analysis of only germline DNA one can never be sure whether lack of identification of a mutation is due to absence of germline mutation or failure to identify a mutation that is present. Even for bilateral cases, the Best Practice Guidelines for Molecular Analysis of Retinoblastoma by the European Molecular Genetics Quality Network recommends analysis of tumor material if no mutation is identified in the germline DNA ( so it would seem a resource conserving measure to analyze tumor sample first in all cases where available. The exception to this would be familial cases in which the heritable mutation is already known. Assuming that 2 mutations are present in all tumors diagnosed as retinoblastoma, in this study we were able to detect 24 of 36 (67%) of mutations; however, no mutations were detected in 4 of 18 (22%) tumors. A previous study analyzing tumor tissue has detected 77% of mutations, with no mutations being found in 2 of 15 (13%) tumors [6]. Other studies analyzing blood DNA in bilateral or hereditary retinoblastoma cases in which a germline mutation is to be expected have detected mutations in 60% to 89% L.N. SELLNER ET AL. of cases [2,3,5,10]. Differences in detection rates may be due to techniques used, because, for example, not all studies use methods to detect deletions, and some studies use entire gene sequencing to identify small mutations while others screen for mutations using D-HPLC as a first pass, which may fail to detect some mutations. Promoter methylation has been suggested to account for gene silencing in sporadic tumors as a somatic event and has been found in 9% to 12% of tumors [2,11] but in none of 33 tumors in this and another study (results not shown and Choy and others [6]). Mutations outside the coding region that have deleterious effects on gene expression or exonic or intronic splice enhancers or suppressors may account for some of the mutations missed by techniques used to date, although 1 study that analyzed RNA from 8 retinoblastoma patients did not detect any unexplained splice variants [10]. Mosaicism may also account for a proportion of missed germline mutations but cannot explain missed tumor mutations. The importance of incorporating a technique to detect full or partial gene deletions in the mutation screening protocol is shown by the rate at which these mutations occur, accounting for 29% of

7 non-loh mutations in this study using the MLPA method and 21% of mutations in a previous study using quantitative multiplex PCR [5]. In our opinion, the MLPA method has several advantages over quantitative multiplex PCR. First, the setup costs are much less, because fluorescently labeled primers for every exon need not be purchased and optimization of the multiplexes can be arduous. Second, all exons are screened in 1 reaction by MLPA, whereas 4 reactions and electrophoresis runs are required for the PCR technique. In this study, we also found 1 patient with a multiexon duplication (exons 8 to 18) using the MLPA method, which is the first report of such a mutation in the RB1 gene. This method does not provide any information as to the arrangement of the duplicated DNA, so without further analysis we cannot determine whether the functional consequences are premature truncation of the protein due to outof-frame transcription of a head-to-head or head-totail duplication in introns 7 or 19 or a larger protein due to an in-frame duplication, or whether the duplication results in incorrect splicing, or whether the duplication is even within the normal gene. One point to note in considering gene dosage detection methods is that neither MLPA nor quantitative multiplex PCR will detect LOH when the mechanism behind the LOH is acquired homozygosity (isodisomy) rather than deletion of that region of one chromosome, since 2 copies of the gene are still present. In this study, we have found that LOH is due to acquired isodisomy in 83% of instances and hemizygosity in 17%, and a previous study has found isodisomy in 93% of instances and hemizygosity in 7% [8]. The isodisomy is thought to be caused by chromosomal nondisjunction roughly half the time, and somatic recombination in the remainder, resulting in segmental isodisomy [8]. It is important to perform LOH analysis specifically in addition to gene dosage analysis to differentiate between isodisomy and hemizygosity when a tumor appears homozygous for a particular mutation, since in the first case only 1 mutation could possibly be germline (but both alleles now carry it), and in the second case, there could be 2 possible germline mutations one a deletion and the other the identified mutation. Germline status of both mutations is required for risk assessment. The incidence of retinoblastoma in this population is slightly higher than previous estimates (1 in versus 1 in to ), and in fact 10 of the cases were in the last 2 years, giving an incidence over that period of 1 in Recently, there has been a suggestion that there may be an increased risk of retinoblastoma in children born after assisted reproductive techniques [12]. We do not have information as to whether the children in this study were conceived through in vitro fertilization. Other studies have suggested an increased risk of imprinting disorders in children born after assisted reproductive techniques [13]. The lack of identification of promoter methylation defects in this study suggests that either the apparently high incidence is not due to assisted reproductive techniques or aberrant methylation is not a common mechanism of increased incidence in retinoblastoma in in vitro fertilization children. ACKNOWLEDGMENTS We would like to thank the laboratory staff at the Division of Molecular Pathology at the Institute of Medical and Veterinary Science in Adelaide for performing the quantitative multiplex PCR analysis. REFERENCES 1. Vogel F. Genetics of retinoblastoma. Hum Genet 1979;52: Richter S, Vandezande K, Chen N, et al. Sensitive and efficient detection of RB1 gene mutations enhances care for families with retinoblastoma. Am J Hum Genet 2003;72: Lohmann DR, Brandt B, Hopping W, Passarge E, Horsthemke B. The spectrum of RB1 germ-line mutations in hereditary retinoblastoma. Am J Hum Genet 1996;58: Lohmann D, Scheffer H, Gaille B. Best Practice Guidelines for Molecular Analysis of Retinoblastoma Available at: www1.emqn.org/best_practice/guidelines.html 5. Houdayer C, Gauthier-Villars M, Lauge A, et al. Comprehensive screening for constitutional RB1 mutations by DHPLC and QMPSF. Hum Mutat 2004;23: Choy KW, Pang CP, Yu CBO, et al. Loss of heterozygosity and mutations are the major mechanisms of RB1 gene inactivation in Chinese with sporadic retinoblastoma. Hum Mutat Zhu X, Dunn JM, Goddard AD, et al. Mechanisms of loss of heterozygosity in retinoblastoma. Cytogenet Cell Genet 1992;59: Hagstrom SA, Dryja TP. Mitotic recombination map of 13cen-13q14 derived from an investigation of loss of heterozygosity in retinoblastomas. Proc Natl Acad Sci U S A 1999;96: Lohmann DR. RB1 gene mutations in retinoblastoma. Hum Mutat 1999;14: Jakubowska A, Zajaczek S, Haus O, et al. Hum Mutat 2001.?3 11. Ohtani-Fujita N, Dryja TP, Rapaport JM, et al. Hypermethylation in the retinoblastoma gene is associated with unilateral, sporadic retinblastoma. Cancer Genet Cytogenet 1997;98: Moll AC, Imhof SM, Cruysberg JRM, Schouten-van Meeteren AYN, Boers M, van Leeuwen FE. Incidence of retinoblastoma in children born after in-vitro fertilisation. Lancet 2003;361: Gosden R, Trasler J, Lucifero D, Faddy M. Rare congenital disorders, imprinted genes, and assisted reproductive technology. Lancet 2003;361: SCREENING FOR RB1 MUTATIONS IN TUMOR