Characterization of retinoblastoma tumours without RB1

mutations: genomic, gene expression and clinical studies

Discovery of MYCN-amplified retinoblastoma with functional

retinoblastoma protein in very young children

Diane E Rushlow, Berber M Mol,* Jennifer Y Kennett,* Stephanie Yee,* Sanja Pajovic,

Brigitte L Thériault, Nadia L Prigoda-Lee, Clarellen Spencer, Helen Dimaras, Timothy W

Corson, Renéee Pang, Christine Massey, Roseline Godbout, Zhe Jiang, Eldad

Zacksenhaus, Katherine Paton, Annette C Moll, Claude Houdayer, Anthony Raizis, William

Halliday, Wan L Lam, Paul C Boutros, Dietmar Lohmann, Josephine C Dorsman, Brenda L

Gallie

*These authors share second authorship

Retinoblastoma Solutions (now Impact Genetics) and the Toronto Western Hospital

Research Institute, Campbell Family Cancer Research Institute, Princess Margaret Cancer

Centre, University Health Network; Informatics and Biocomputing Platform, Ontario Institute

for Cancer Research; Departments of Hematology/Oncology, Ophthalmology and Visual

Science and of Pathology, Hospital for Sick Children; and Departments of Molecular

Genetics, Ophthalmology, Medical Biophysics, Pathobiology and Lab Medicine, University

of Toronto, Toronto, ON, Canada (D E Rushlow, BSc, S Yee, MSc, S Pajovic, PhD, B L

Thériault, PhD, N L Prigoda-Lee, MSc, C Spencer, BSc, Z Jiang, BSc, E Zacksenhaus, PhD, R

Pang, MA, C Massey, MSc, H Dimaras, PhD, P C Boutros, PhD, W Halliday, MD, Prof B L Gallie,

MD); Departments of Clinical Genetics, Ophthalmology and Pediatric Oncology/Hematology,

VU University Medical Center Amsterdam, Amsterdam, The Netherlands (B M Mol, MSc, A C

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

Moll, MD, J C Dorsman, PhD); British Columbia Cancer Research Centre and Departments of

Ophthalmology and Pathology & Laboratory Medicine, University of British Columbia,

Vancouver, BC, Canada (J Y Kennett, MSc, K Paton, MD, Prof W L Lam, PhD); Eugene and

Marilyn Glick Eye Institute, Departments of Ophthalmology, Biochemistry and Molecular

Biology, Indiana University School of Medicine Indianapolis, Indiana, USA (Timothy W

Corson, PhD); University of Alberta, Cross Cancer Institute, Edmonton, Alberta (Prof Roseline

Godbout, PhD); Service de Génétique Oncologique, Institut Curie and Université Paris

Descartes Paris, France (C Houdayer, PhD); Department of Molecular Pathology, Canterbury

Health Laboratories Christchurch, New Zealand (A Raizis, PhD); and Institut für

Humangenetik, Universitätsklinikum, Essen, Germany (Prof D Lohmann, MD)

Correspondence to Dr Brenda L. Gallie, Campbell Family Cancer Research Institute, Princess Margaret

Cancer Centre, University Health Network, Rm 8-415, 610 University Ave, Toronto, ON, M5G 1M9, Canada,

brenda@gallie.ca

2

1

1

2

3

4

5

6

7

8

9

10

12

13

14

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

Word count 286300/300

Summary

Background Retinoblastoma is the childhood retinal cancer that defined tumour suppressor genes.

By analysing age of at diagnosis, Knudson proposed that two “hits” initiate retinoblastoma, later

attributed toidentified as mutation of both alleles of the retinoblastoma suppressor gene,

(RB1-/- )RB1, in tumours. Persons with hereditary retinoblastoma carry a heterozygous constitutional

RB1 mutation; one additional hit initiates retinoblastoma and/or other cancers. Non-hereditary

retinoblastoma arises when both RB1 alleles are damaged in developing retina (RB1-/-).We set out to

characterize the non-familial retinoblastoma tumours with no detectable RB1 mutations.

Methods Of 1068 We unilateral non-familial retinoblastoma tumours, analysed we compared those

with no evidence of RB1 mutations (RB1+/+ ) to the usualmajority RB1-/- tumors, studying clinical

data, genomic copy-number, histology, RB1 gene expression and protein function, and retinal gene

expression, histology, and clinical datain unilateral non-familial retinoblastomas with no evidence

of RB1 mutations (RB1+/+), compared to RB1-/- tumours.

Findings We found no evidence of RB1 mutations that in 2·7% (29/1068) of unilateral non-familial

retinoblastomas . had no evidence of RB1 mutations (RB1+/+). Surprisingly, half ofOf the RB1+/+

tumours, 15/29 had high-level MYCN oncogene amplification (28 to 121 copies) (RB1+/+ MYCNA ),

while 0/93no RB1-/- primary tumours showed MYCN amplification (p<0·0001). RB1+/+MYCNA

tumours cell lines expressed functional RB1 protein,. RB1+/+MYCNA tumours had fewer overall

genomic copy-number changes than RB1-/- tumours, and distinct, aggressive histology. MYCN

amplification was the sole copy-number change detected in one RB1+/+MYCNA retinoblastoma. Two

more RB1+/+ MYCNA tumours were discovered after the initial study. Median (IQR) age at diagnosis

of 17 RB1+/+MYCNA tumours was 4·5 (3·5-10) months, compared to 24 (15-37) months for 79 non-

3

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

familial unilateral RB1-/- retinoblastoma. Children diagnosed with unilateral non-familial large

retinoblastoma at six months of age or less have >18% chance of RB1+/+MYCNA tumour.

Interpretation Amplification of the MYCN oncogene may initiate RB1+/+MYCNA retinoblastoma

despite in the presence of a normal RB1 genes. Although they are very young, it is unlikely that

these children carry a hereditary risk for other retinoblastoma or RB1-related cancers. Since these

tumours may rapidly become extraocular, removal of the eye of young children with potential large

potential RB1+/+MYCNA retinoblastoma is good care.

Funding NCI-NIH; CIHR; DFG; Canadian Retinoblastoma Society; Hyland Foundation; Toronto

Netralya and Doctors Lions Clubs; Ontario Ministry of Health and Long Term Care; UK-Essen;

Toronto Netralya and Doctors Lions Clubs;and Foundations Avanti-STR and KiKa.

4

1

1

2

3

4

5

6

7

8

9

10

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

Word count 33752998/3000

Background

Retinoblastoma set the paradigm for tumour suppressor genes, with Knudson’s classic hypothesis

predicting that two rate-limiting hits initiate this childhood eye cancer.1 The two hits were later

attributed to the retinoblastoma gene (RB1).2 The 40% of children who have bilateral

retinoblastoma carry a heterozygous, constitutional RB1 mutation and are predisposed to

retinoblastoma; one additional hit, damages the second RB1 allele and initiates retinoblastoma,

and/or other cancers later in life. In 70% of tumours, the second hit involves loss of the normal

allele with duplication of the mutant allele (loss of heterozygosity, LOH). The 60% of children with

unilateral non-familial retinoblastoma usually have both RB1 alleles damaged only in developing

retina, but 15% of them carry a heritable, constitutional RB1 mutation. Accepted dogma is that

disruption of both RB1 alleles (RB1-/-) initiates all retinoblastoma.2-4

The heterozygous mutant RB1 allele is identifiable in blood of 95% of bilaterally affected

persons. Low-level mosaicism may account for the 5% "no mutation found" blood samples from

bilaterally affected patients.5 The Toronto lab identifies both RB1 mutations (or promoter

methylation) in >95% of tumours from unilateral probands with no known family history.5, 6 In 3-

4% of tumours only one RB1 mutation is identified, and in 7/441 (1·6%) no RB1 mutation is found.

Undetected RB1 mutations in fully tested tumours (RB1+/+ ) may include translocations, deep

intronic mutations, or mutations in unknown RB1 regulatory regions. The heterozygous mutant RB1

allele is identifiable in blood of 95% of bilaterally affected persons. RB1 mutations or promoter

methylation, are detected on both alleles (RB1-/-) in 95% of unilateral retinoblastomas.5, 6 The

possibility that some unilateral retinoblastomas with no detectable RB1 mutations occur by an

independent mechanism has not been previously explored.

5

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

We report the first identification of unilateral retinoblastomas with normal RB1 alleles and high-

level MYCN gene amplification (MYCNA). These unilateral RB1+/+MYCNA retinoblastomas are

characterized by distinct histology, few of the genomic copy-number changes characteristic of

retinoblastoma, and very early age of at diagnosis. This newly recognized sub-type of

retinoblastoma has immediate diagnostic, genetic counselling, and therapeutic implications.

6

1

1

2

3

4

5

6

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

Methods (details in webappendix)

Clinical samples

Tissues and clinical data were provided for identification of RB1 mutations in clinical care of

children and their families. The families tested were both local to each laboratory and international.

Research Ethics Board approvals to use de-identified data and tissues after clinical analyses, are on

file at each participating site. Although not required, the Toronto and Essen patients provided

additional informed consent.

Mutation analyses

Standard of care analyses that identify 95% of RB1 (Gen bank accession #L11910) mutant alleles5-7

were applied at each site, including DNA sequencing, quantitative multiplex PCR (QM-PCR) or

Multiplex Ligation-Dependent Probe Amplification (MLPA), and RB1 promoter methylation

testing. Intragenic and closely- linked RB1 microsatellite markers and SNPs were used to determine

zygosity of the RB1+/+ tumours.

Genome copy-number analyses

Genomic copy-numbers of five genes known to be commonly altered in RB1-/- retinoblastomas

(KIF14, DEK, E2F3, CDH11, and MYCN) were determined by QM-PCR (Toronto) or MLPA and

single nucleotide polymorphism (SNP) analyses (Amsterdam). Sub-megabase resolution array

comparative genomic hybridization (aCGH) or SNP array were used to assess overall genomic

copy-numbers and normal cell contamination.

Retinoblastoma cell lines

Three primary RB1+/+ MYCNA tumours were placed in the same culture conditions that support

growth of RB1-/- cell lines, and developed into cell lines.

7

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

Protein expression studies

Paraffin-embedded sections of retinoblastomas and adjacent normal retinas were stained for full-

length pRb protein (antibodies targeting N- and C-terminal pRb) and N-Myc.8 Western blots were

performed on RB1+/+MYCNA and, RB1-/- primary tumours aand cell lines, and control cell

linescontrol RB1+/+ human fetal and adult retina, and lymphoblastoid and neuroblastoma cell lines.

Function of pRB was assessed by immunoblotting with phospho-specific pRb antibody and co-

immunoprecipitation of pRb and its major effector molecule, E2F1.

RNA gene expression studies

Expression of RB1, MYCN, and genes reflecting the proliferative status and retinal derivation of

tumours were assessed using End-Point Reverse Transcriptase PCR (RT-PCR) and/or Quantitative

Real-Time PCR.

Age of Diagnosis analysis

Ages at diagnosis vs. proportion not yet diagnosed were analysed to assess minimal number of

events for tumour initiation. Likelihood of children having RB1+/+MYCNA tumours at different ages

of diagnosis was estimated.

Statistical Analyses

To evaluate the likelihood that RB1+/+ tumours originated from two undetected RB1 mutations, Chi-

squared goodness-of-fit test, using default probabilities of p1=0·0025 for RB1+/+ and p2=0·9975 for

RB1- /- samples was used. The bimodal distribution of MYCN copy number in RB1+/+ tumours was

established using a two-tailed t-test with Welch’s adjustment for heteroscedasticity. MYCN

amplification and frequencies of specific genomic copy number changes comparing RB1+/+ tumours

to RB1-/- tumours, used Fisher’s Exact Test. These statistical tests were performed in the R statistical

environment (v2.13.2).

8

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

Total bp altered per region comparing RB1+/+ MYCNA and RB1-/- tumours used t-test with Welch’s

adjustment. Normalized intensity values were processed using the Circular Binary Segmentation

algorithm, with data filtered to identify only high-quality (>3 confirmatory probes), and suitably

sized (>25 kbp) regions of clear differential signal relative to normal (|mean-signal| > 0·1). Separate

terms were fitted to groups using a general linear model. A contrast matrix was used to identify the

number of differentially abundant probes relative to normal samples, with an empirical Bayes

moderation of standard error and a false-discovery rate adjustment for multiple testing. To compare

age of diagnosis between groups, pair-wise comparisons using the Wilcoxon rank-sum test were

performed. These analyses were performed in the R statistical environment (v2.11.1) using the

DNAcopy (v1.22.1) and limma (v3.4.4) packages. Copy number alterations detected by SNP array

were analyzed with the genomic segmentation algorithm of Partek Genomics Suite (Partek Inc., St.

Louis, MO, U.S.A.) with the following parameters: minimum number of markers 100, signal to

noise ratio 0.4, segmentation p-value: 0.0001, loss < 1.5, gain > 2.5, report p-value threshold: 0.01).

Role of the funding sources

The sponsors of the study had no roles in study design, data collection, analysis, interpretation, or

writing of the manuscript. No author was paid to write this article. All authors had full access to all

data in the study; the corresponding author (BLG) had final responsibility for the decision to submit

for publication.

9

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

Results

The Toronto lab identifies both tumour mutations (or promoter methylation) in >95% of tumours

from unilateral probands with no known family history. In 3-4% of tumours only one RB1 mutation

is identified, and in 1·6%, no RB1 mutation is found. Low-level mosaicism may account for the 5%

"no mutation found" blood samples from bilaterally affected patients.5 However, undetected RB1

mutations in fully tested tumours (RB1+/+) may include translocations, deep intronic mutations, or

mutations in unknown RB1 regulatory regions.

In clinical work in Toronto, we found seven 7/441 (1.6%) unilateral retinoblastoma tumours with

no RB1 mutations, no promoter methylation, and no RB1 LOH (RB1+/+). We used QM-PCR to

measure, in RB1+/+ tumours, copy-numbers of genes at 6p, 1q, 16q and 2p that are characteristically

gained or lost in RB1-/- retinoblastoma tumours.9 To our surprise, 5/7 RB1+/+ tumours showed

dramatic MYCN oncogene amplification (MYCNA). To validate this observation, we collaborated

with RB1 clinical laboratories in Germany, France, and New Zealand. , to study similar RB1+/+

tumours (table 1). After initial studies, we learned that the Amsterdam group had independently

discovered three RB1+/+MYCNA tumours. We combined our data on 1068 primary unilateral non-

familial retinoblastoma tumours. , to study similar RB1+/+ tumours (table 1).We report our

combined data after statistical analysis showed that frequencies were not different (p = 0·08) among

the five labs (Table 1). Subsequent to the initial discovery, rRetinoblastoma T101 was, predicted

clinically and pathologically to be an RB1+/+MYCNA tumour , confirmed by MYCN copy number,

(figure 4C), had 40 copies of MYCN, and is included in subsequent some analyses.

Since our standard sensitivity to find the RB1 mutation in blood of bilaterally affected persons is

>95%,5-7 the probability of finding no RB1 mutations in a tumour with no RB1 LOH is equivalent to

the probability of missing two independent RB1 mutations in one sample (0·05 x 0·05) or 0·25%.

By combining our data on 1068 unilateral non-familial tumours, we identified 29 RB1+/+tumours

10

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

(2·7%), 10-fold more than expected (p = 6 x 10-45) (table 1). This suggested that some RB1+/+

tumours might originate by a mechanism other than Knudson’s two RB1 mutations.

To characterize copy numbers of genes commonly gained or lost in retinoblastomas,9 we used

QM-PCR (Toronto) or MLPA/SNP (Amsterdam) analyses. MYCN copy-number was elevated in

27/30 (90%) RB1+/+ vs. 60/93 (65%) RB1-/- retinoblastomas (p = 3·4 x 10-4) (table S6). No RB1-/-

primary tumours (0/93) showed over >10 MYCN copies. In comparison, 16/30 (53%) RB1+/+

tumours showed high-level MYCN amplification (28 to 121 copies of MYCN) , (RB1+/+MYCNA

retinoblastoma) (p <0·0001) (tables 1, S6, figures 1, S2). The remaining 14 RB1+/+ tumours showed

between 2 and 10 MYCN copies. For ten children with RB1+/+MYCNA tumours, normal DNA from

available blood showed two MYCN copies.

Compared to RB1-/- retinoblastoma, the 16 RB1+/+MYCNA tumours showed lower frequency of

copy-number changes in four other genes characteristic of retinoblastoma: gain of oncogenes KIF14

(55/89, 62% vs. 3/16, 19%; p = 0·002), DEK (50/87, 55% vs. 1/16, 6%; p = 0·0002), and E2F3

(52/90, 55% vs. 1/16, 6%; p = 0·0002), (57% vs. 6%; p = 0·0002) and loss of tumour suppressor

gene CDH11 (50/90, 56% vs. 2/16, 13%; p = 0·002) (tables S6, S7). SNP analysis indicated that the

level of normal cell contamination in the RB1+/+MYCNA tumours was low and similar to RB1-/-

tumours (table S2).

To probe for other genomic gains/losses, we studied DNA from 48 unilateral retinoblastomas by

aCGH10 (14 RB1+/+ MYCNA , 12 RB1+/+ , 25 RB1-/- (+) , from the initial 4 study sites) and 3

RB1+/+ MYCNA tumours by SNP analysis (Amsterdam) (14 RB1+/+MYCNA, 12 RB1+/+, 25 RB1-/-(+))

(tables S6, S8, figures 2A, S25). None of the RB1+/+MYCNA retinoblastomas showed any evidence

of copy-number changes or translocations11 at the RB1 locus. Except for MYCN copy-number,

aCGH (figure 2A) confirmed a reduced frequency in RB1+/+MYCNA retinoblastomas of the specific

genomic copy-number changes characteristic of RB1-/- retinoblastomas (figure 1, table S7).9 The

11

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

RB1+/+MYCNA retinoblastomas also showed overall significantly fewer altered bp and aCGH clones

than the RB1-/- retinoblastomas (p = 0·033) (figures 2A, S3, table S8).

MYCN amplicons of 14/14 (11 by aCGH, 3 by SNP analysis) RB1+/+MYCNA retinoblastomas that

were studied by aCGH (11) or SNP analysis (3), as well as one RB1+/- tumour (T33) with 73 MYCN

copies ofon aCGH MYCN, and one RB1-/- primary tumour (RB381) with 9·2 copies of MYCN, were

narrow, spanning 1·1 to 6·3 Mbp encompassing MYCN (figures S4, S5, table S8). Importantly, The

sole copy-number change detected in one RB1+/+MYCNA retinoblastoma (E4) was 48 copies of

2p24·2–-24·3 encompassing MYCN. The minimal common amplicon defined by RB1+/+(-) MYCNA

retinoblastomas T33 and P2 spanneding 513 kbp, containinged only MYCN (figure 2B).

RB1+/+MYCNA retinoblastomas T5 and P2 also defined a minimal common amplicon including only

MYCN. Of the remaining RB1-/- (11), RB1-/+ (13) and RB1+/+ (12) unilateral 35 tumours tested by

aCGH, 24/35 unilateral tumours showed no gain or loss at MYCN, and 11/35 had moderate gain

spanning a broad region of at least 28 Mbp of chromosome 2p, too large to meet the definition of

amplification.12

Three of 15(23%) RB1+/+MYCNA tumours showed 17q21·3-qter or 17q24·3-qter gain; two

RB1+/+MYCNA tumours showed 11q loss. Both regions are commonly altered in neuroblastoma,13, 14

but rare in RB1-/- retinoblastoma (present and published data15, 16). Other changes in RB1+/+MYCNA

retinoblastomas not often seen in RB1-/- retinoblastoma were gains at 14q and 18q, and losses at 11p

(figures 2A, S5).

Retinoblastoma T33 (RB1+/-) showed high-level MYCN amplification and loss of one copy of

most of 13q, including RB1; we suspect that amplification of MYCN initiated proliferation, followed

by 13q deletion. Since T33 showed numerous characteristics of RB1+/+MYCNA tumours, we included

T33 with MYCNA retinoblastomas in many clinical analyses (figures 2, 4A, S4, S6).

RB1+/+MYCNA retinoblastomas expressed functional RB1 protein (pRb). The RB1+/+MYCNA

retinoblastomas and the expected retinal cell types expressed pRb (ganglion cells, specific inner

12

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

nuclear cells, and cone photoreceptors)17 and showed nuclear staining for C-terminal (figures 3A,

S6) and N-terminal (data not shown) epitopes, while RB1-/- and RB1+/- tumours were negative or

stained weakly, depending on their RB1 mutations. Western blots and immunoprecipitation in 3

RB1+/+MYCNA cell lines (A3, RB522, T101) derived from RB1+/+MYCNA primary retinoblastomas,

demonstrated functional nuclear pRb that was full-length (figures S6, S7), normally hypo- and

hyperphosphorylated (figure 3B, S7), and bound to E2F1 (figure 3C), the major target of pRb in

control of the cell cycle.18 Full-length 2·8 kbp RB1 transcripts were detected by end-point and real-

time RT-PCR in the 3 RB1+/+MYCNA primary retinoblastomas for which mRNA was available

(figures 3D, S8, table S9). In contrast, 4 primary most RB1-/- retinoblastomas expressed low RB1

transcript levels relative to fetal and adult retina..

RB1+/+MYCNA primary retinoblastomas (figure 3A) and three derived cell lines (data not shown)

stained strongly for N-Myc protein. MYCN and KI67 transcripts (indicative of proliferation) were at

low levels in adult retina, at higher levels in fetal retina and RB1-/- retinoblastomas without MYCN

amplification, and at very high levels in primary RB1+/+MYCNA retinoblastoma and RB1-/- cell lines

with MYCN amplification (figures 3A, D, S8A, table S8). RB1+/+MYCNA tumours showed reduced

mRNA expression of the oncogene KIF14,9 in contrast to normal fetal retina, and RB1-/- primary

retinoblastomas and cell lines which over-expressed KIF14 (figure 3DCF, table S9).

RB1+/+MYCNA tumours expressed embryonic retinal cell markers consistent with a retinal origin.

The mRNAs of cone cell marker X-arrestin19 and CRX, a marker of retinal and pineal lineage

tumours, (strongly expressed in RB1-/- retinoblastoma but not in neuroblastoma,20) were expressed in

fetal retina, human adult retina, and RB1+/+MYCNA and RB1-/- primary retinoblastomas (figure S8B).

Children with RB1+/+MYCNA retinoblastomas were much younger at diagnosis than children with

unilateral RB1-/- retinoblastomas. The median (IQR) age at diagnosis of 17 children (with 16

RB1+/+MYCNA and one RB1+/-MYCNA [T33] tumours) was 4·5 (3·5-10) months, significantly

younger than 79 children with unilateral sporadic RB1-/- (24·0 (15-37) months, p<10-4) or 14

13

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

children with RB1+/+ (21·5 (16-45) months, p<10-4) retinoblastomas (figure 4A, tables S5A & S6).

We estimate that 18·4% of children diagnosed with non-familial, unilateral retinoblastoma at age

six months or younger will have RB1+/+MYCNA retinoblastoma (table S5B).

Analysis of age at diagnosis vs. proportion not yet diagnosed led Knudson to propose that two

hits initiate retinoblastoma.1 Our data from 79 unilaterally affected RB1-/- patients was consistent

with Knudson’s model and fit a two-hit curve, representative of two independent mutational events

in a tumour suppressor gene (figure 4B). Similar analysis for RB1+/+MYCNA tumours was

inconclusive: while the data points for twelve children diagnosed before 10 months approximated

the calculated one-hit curve, ages for the older children deviated (figure 4B).

RB1+/+MYCNA tumours showed distinctive histology, displaying undifferentiated cells with large,

prominent, multiple nucleoli, and necrosis, apoptosis, and little calcification, similar to other

MYCNA embryonic tumours, such as neuroblastoma21 (figures 4C5A, S6). The Flexner-

Wintersteiner rosettes22 and nuclear molding characteristic of prototypic RB1-/- retinoblastoma

(figure 4D5B) were absent.

Clinically, the RB1+/+MYCNA retinoblastomas were large and invasive, considering the very

young age of these children (figures 4C, E5A, C, S7, S9; compare 5D). Three RB1+/+MYCNA

retinoblastomas (RB522, T101, and A3) from enucleated eyes grew rapidly into cell lines, unlike

the RB1-/- retinoblastomas that grow poorly in tissue culture. The cell lines still showed no RB1

mutations and maintained the MYCN amplification of the primary, confirming minimal normal cell

contamination of the primary tumour. One RB1+/+MYCNA retinoblastoma had already invaded the

optic nerve past the cribriform plate at age 11 months, a feature of aggressive disease23, 24 (figure

4E5C). However, with follow-up ranging from 2 to 25 years, all the children with RB1+/+MYCNA

tumours were cured had no further evidence of retinoblastoma after by removal of their affected eye

with no adverse outcomes and; none developed retinoblastomas in their other eye.

14

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

Discussion

Knudson’s analysis of retinoblastoma was fundamental to the concept that normal genes suppress

cancer,1 leading to the identification of the RB1 tumour suppressor gene,2 widely assumed to initiate

all retinoblastoma. We now describe a previously unrecognized type of retinoblastoma with no

evidence of RB1 mutations and with functional pRb, likely initiated by amplification of the MYCN

oncogene.

Our international study of 1,068 unilateral retinoblastomas revealed a distinct RB1+/+MYCNA

subtype comprising 1·4% of unilateral, non-familial retinoblastomas. At least two previously

described tumours may also be RB1+/+MYCNA retinoblastomas,25 although RB1 genetic status was

not defined. Despite the low incidence of RB1+/+MYCNA retinoblastoma, RB1+/+MYCNA tumours

were independently discovered and characterized in the Toronto and Amsterdam labs, with different

patient cohorts and technologies. The statistical analyses show that although rare, the RB1+/+ MYCNA

retinoblastoma are an important subset that challenges dogma.

We also identified 14 RB1+/+ retinoblastomas without MYCN amplification. aCGH showed

genomic gains and losses distinct from either RB1-/- or RB1+/+MYCNA retinoblastomas, but we

lacked RB1+/+ tumour materials for gene expression or protein studies. In particular, these RB1+/+

retinoblastomas showed frequent gain of 19p and q, 17p and q, 2p, and the telomeric end of 9q. The

RB1+/+ group appears heterogeneous and merits further study.

Our paper highlights how molecular diagnostics can identify novel malignancies that elude

histopathological recognition. Although RB1+/+MYCNA retinoblastomas were not previously

recognized as histologically distinct, they resemble large nucleolar neuroblastomas with MYCN-

amplification and poor outcome.21 Like MYCN-amplified neuroblastomas,14 RB1+/+MYCNA tumours

showed less complex genomic copy-number alterations than tumours without MYCN amplification,

suggesting that MYCNA may be the critical driver of malignancy. However, the early age of

15

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

diagnosis of RB1+/+MYCNA tumours may allow less time to accumulate genomic alterations.

Although whole genome sequencing recently suggested that point mutations (other than in the RB1

gene) are few in RB1-/- retinoblastoma,26 loss of RB1 induces mitotic changes and lagging

chromosomes,27 leading to genomic instability. RB1 loss in retinoblastoma and pre-malignant

retinoma8, 16 is associated with specific DNA copy-number changes, that are less frequent in

RB1+/+MYCNA retinoblastomas.

Are RB1+/+MYCNA tumours truly retinoblastomas? RB1+/+MYCNA retinoblastomas arise in, and

express markers of, embryonic retina, meeting the definition of retinoblastoma as ‘a blast cell

tumour arising from the retina’.19, 20, 28 We demonstrate intact RB1 genes in primary RB1+/+MYCNA

retinoblastomas, with strong nuclear pRb antibody staining. Three RB1+/+MYCNA cell lines

expressed full length RB1 mRNA, and inactive hyperphosphorylated and active

hypophosphorylated pRb that binds co-immunoprecipitates with to E2F1, indicating normal

functional pRb18.

The possibility of normal cell contamination masking an RB1 mutations in the RB1+/+MYCNA

tumours is ruled out by the 3 RB1+/+ MYCNA cell lines that grew rapidly from the primary tumors,

unlike most RB1-/- retinoblastoma. that They maintained the RB1+/+ genotype of the primary

tumour,and the very high level MYCN amplification of the primary tumours. While NextGen

Sequencing is currently a greatpowerful discovery tool, its clinical application has not yet achieved

sensitivity to match our RB1 mutation detection technologies. If an RB1 mutation were found

through NextGen enhanced read depth, that variant would likely represent convergent evolution to

knock out pRb with cancer progression, common in many cancers, rather than an initiating RB1

mutation. TBased on the cell line genotypes, the MYCN amplification would still be the event

common to all cells, thereby likely the initiating event., histology with the only normal cells being

vascular, and strong histological and biochemical evidence that pRb is functional.

16

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

The very early presentation of RB1+/+MYCNA retinoblastomas, lack of RB1 mutations, functional

pRb and high MYCN-amplification in a relatively copy-number stable genome, suggests that these

retinoblastomas arise by somatic MYCN oncogene amplification in a retinal progenitor cell. This is

supported by one RB1+/+MYCNA tumour in which MYCN amplification was the only identified

genomic copy-number change. The RB1+/+MYCNA retinoblastomas are already large in very young

children, so they likely initiate earlier in retinal development than RB1-/- retinoblastomas. In

children with an RB1 mutation of equivalent age, the tumours are usually much smaller (figure

5D4F). How MYCN -amplification is initiated, and whether MYCN -amplification alone suffices to

initiate these retinoblastomas remains to be formally demonstrated.

In retinal development the N-myc protein can support cell division in the absence of activating

E2fsE2Fs; presumably unregulated MYCN expression associated with high-level gene amplification

in RB1+/+MYCNA tumours promotes cell division indirectly through inactivation of pRb.29 The

relative genomic stability of RB1+/+MYCNA retinoblastomas suggests that anti-N-Myc therapeutic

agents30 may avoid the emergence acquisition of drug resistance acquired through progressive

genomic rearrangements that is apparent in RB1-/- retinoblastoma. Our preliminary MYCN knock

down experiments resulted in rapid death of RB1+/+MYCNA cell lines (data not shown). We predict

that children less than one year of age with extra-ocular retinoblastoma around the world,23 with

extraocular retinoblastoma , may have RB1+/+MYCNA retinoblastomas that might benefit from anti-

N-Myc therapy. This idea is consistent with our (BG) anecdotal observation during review of

retinoblastoma pathology slides in Kenya, of RB1+/+MYCNA histology in an orbital recurrence, later

confirmed to have 40 MYCN copies. of MYCN.

Young age at diagnosis of unilateral retinoblastoma is frequently interpreted as an indication of

heritable retinoblastoma, and is currently cited as a reason to try to cure the cancer without

enucleation. However, attempts to salvage an eye with a large RB1+/+MYCNA retinoblastoma could

be dangerous; in our study, prompt removal of the unilateral affected eyes was curative with no

17

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

adverse outcomes. The young patients with RB1+/+MYCNA tumours in our study had large,

aggressive tumours, including invasion into the optic nerve (figure 5C). At similar young ages, the

usual hereditary RB1-/- tumours are very much smaller, detected only by active surveillance (figure

4F5D). While our data predict that 18·4% of children diagnosed with non-familial, unilateral

retinoblastoma at age six months or younger will have RB1+/+MYCNA retinoblastoma (table S54B),

if tumour size is also considered, the risk will may be higher, and RB1+/+MYCNA tumours may be

clinically predictable, facilitating prompt removal of these eyes with good outcomes for the

children.

Standard care for unilateral non-familial retinoblastoma is identification of the RB1 mutant

alleles in tumour, and examination of blood to determine whether either mutant allele is germline.

Our study suggests that when no RB1 mutation is detected in a retinoblastoma tumour, determining

MYCN copy-number may assist on-going care. The diagnosis of RB1+/+MYCNA retinoblastoma

strongly suggests non-hereditary disease with normal population risks for retinoblastomas in the

other eye and other cancers throughout life, and no familial risks.

Our findings challenge the long-standing dogma that all retinoblastomas are initiated by RB1

gene mutations (summary figure 4Gtable 2). RB1+/+MYCNA retinoblastoma provides an intriguing

contrast to classical retinoblastoma, of immediate importance to patients.

Panel: Research in Context

Systematic Review

We systematically searched the published, peer-reviewed literature on PubMed

(http://www.ncbi.nlm.nih.gov/pubmed/) using the search terms “retinoblastoma”, “initiation” and

“genetics”; “retinoblastoma tumour genetics”; “retinoblastoma development”; and “retinoblastoma

initiation”. We reviewed publications with a main focus on genetic initiation and development of

human retinoblastoma. We found no data that challenged Knudson’s 1971 conclusion that two rate-

18

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

limiting events1, later shown to be loss of both RB1 gene alleles,2-5, 7 are essential but not necessarily

sufficient for development of retinoblastoma. We found no data to suggest another form of

retinoblastoma.

Interpretation

Our collaborative studies identify a previously unrecognized disease: retinoblastoma with a normal

RB1 genes, apparently driven by MYCN oncogene amplification. This newly recognised form of

retinoblastoma has immediate clinical implications for patients. RB1+/+MYCNA retinoblastomas are

diagnosed by molecular study of the tumour after removal of the eye of very young children with

unilateral non-familial disease. The children with RB1+/+MYCNA retinoblastoma and their relatives

are predicted to be at normal population risk for recurrence or other cancers. Attempts to salvage

the eye on the assumption of heritable disease in young children may incur high treatment

morbidity and failure to cure these aggressive oncogene-driven retinoblastomas.

19

1

1

2

3

4

5

6

7

8

9

10

11

12

13

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

Tables

Test Site Total RB1+/+ Proportion RB1+/+Number

RB1+/+MYCNA

Proportion

RB1+/+MYCNA

Canada 441 7 1·6% 5 1·1%

Germany 400 12 3·0% 4 1·0%

France 150 5 3·3% 2 1·3%

New Zealand 30 2 6·7% 1 3·3%

The Netherlands 47 3 6·4% 3 6·4%

Total 1068 29 2·7% 15# 1·4%

*Fisher's exact tests indicates that percentage of unilateral tumours with RB1+/+ is does not related tovary by site (p = 0·08).

Table 1. Frequency of RB1+/+ unilateral retinoblastoma at five international diagnostic labs

20

1

1

2

3

4

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

Figure Legends

Figure 1: Copy numbers of genes commonly altered in RB1-/-

retinoblastoma

(A) Box-plot of genomic copy-numbers determined by QM-PCR for indicated

genes in unilateral retinoblastoma, grouped by RB1 mutation status. T33 is an

outlier of the RB1+/- group, showing an RB1+/+MYCNA-like profile. Gray line, 2-

copies. (B) Heat-map for copy number for the profile genes (red, increased, and

blue, decreased copy-number; gray, 2 copies; white, not tested; n, number in

each group). Recently discovered RB1+/+MYCNA tumours not in (A) are indicated

by black triangles.

Figure 2: Genomic copy number alterations

(A) aCGH on 12 RB1+/+MYCNA (including T33) (red), 12 RB1+/+ (blue), 13 RB1+/-

(brown), and 11 RB1-/- (green) tumours; gains, right, and losses, left of

chromosome; minimal commonly gained/lost regions in RB1-/- tumours boxed;

*normally occurring copy number variations. Tumour T33 shows loss of most of

13q. (B) The minimal amplicon of 513 kbp is defined by two MYCNA tumours

(pink band); MYCN copy number by QM-PCR, red italics; aCGH individual

probes, green bars.

21

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

Figure 3: Expression of RB1 and MYCN

(A) Staining of adjacent retina and RB1+/+MYCNA or RB1+/- mutant

retinoblastoma for N-Myc protein and pRB (C-terminus antibody); T, tumour;

INL, inner nuclear layer of retina. (B) Western blots with pRb antibody that

recognizes both hypo- and hyperphosphorylated pRb, phospho-Rb (Ser795)

antibody, and E2F1 N-MycE2F1 antibody. Fetal retina (FR) and neuroblastoma

(NB) lysates included for comparison. (C) Cell lysates were immunoprecipitated

with antibodies to mouse IgG (negative control), pRb or E2F1, and western

blots performed with antibodies to pRb and E2F. (D) Real-time RT-PCR for RB1,

MYCN and KIF14 in human fetal (FR) and adult (HR) retina, RB1+/+MYCNA, and

RB1-/- primary tumours and cell lines; triplicate measurements normalized

against GAPDH, relative to FR; MYCN DNA copy-numbers in italics; #, KIF14 not

done.

22

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

Figure 4: Clinical features of children with RB1+/+MYCNA tumours

(A) (A) Children with RB1+/+MYCNA retinoblastoma are diagnosed

significantly younger than children with RB1-/- tumours (p<0·0001,

Wilcoxon rank sum test). (B) The Knudson plot of proportion not yet

diagnosed vs age at diagnosis, using birth as a surrogate for

initiation, showing a two-hit curve (blue) and a one-hit curve (red)

for children with RB1-/- tumours or RB1+/+MYCNA tumours;

scatterplot does not distinguish identically aged children. (C) Fundus

image of a large RB1+/+MYCNA unilateral tumour, extending from optic nerve

(white arrow) to anterior border of retina (double arrows) in a 4 month-old

child with characteristic calcification on ultrasound, and round nuclei with

prominent large multiple nucleoli on pathology, in comparison to (D) RB1-/-

tumour showing classic Flexner-Wintersteiner rosettes and nuclear molding;

hematoxylin-eosin staining. (E) RB1+/+MYCNA retinoblastoma in an 11 month-

old child (A2) with extra-ocular extension into the optic nerve (arrows) (2·5x,

hematoxylin-eosin staining). (F) In comparison, in 3·5 month-old child with

heritable RB1-/- retinoblastoma, a small tumour is present in the inner

nuclear layer of the retina on optical coherent tomography (OCT). (G)

Schema of data establishing RB1+/+MYCNA retinoblastoma as a novel

disease; months (m), data figures (f) and tables (t) indicated in grey on left.

Figure 5: Clinical features of RB1+/+ MYCNA tumours.

(A) Fundus image of an RB1+/+ MYCNA unilateral tumor, extending from optic

nerve (white arrow) to anterior border of retina (double arrows) in a 4

month-old child with characteristic calcification on ultrasound, and round

23

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

nuclei with prominent large multiple nucleoli on histopathology, in

comparison to (B) RB1-/- tumour showing classic Flexner-Wintersteiner

rosettes and nuclear moulding; hematoxylin-eosin staining. (C) RB1+/+ MYCNA

retinoblastoma in an 11 month-old child (A2) with extra-ocular extension

into the optic nerve (whiteblack arrows) (2.5x, hematoxylin-eosin staining).

(D) In comparison, in a 3.5 month-old child with heritable RB1-/-

retinoblastoma, a small tumor is present in the inner nuclear layer of the

retina on optical coherent tomography (OCT).

24

1

1

2

3

4

5

6

7

8

9

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

Contributors

Diane E Rushlow recognized the initial connection between MYCNA amplification and RB1

mutation status, performed literature search and QM-PCR analysis and supervised RB1 mutation

analysis, coordinated collaborations with the other sites and was the major contributor to manuscript

preparation. Berber M Mol recognized the RB1 and MYCN mutation status of the Amsterdam

samples and performed by MLPA and SNP array analysis, performed immunohistochemistry,

Western blots and co-immunoprecipitations. Jennifer Y Kennett performed aCGH and analysed

aCGH data. Stephanie Yee performed analysis of aCGH data, the MYCNA alignment,

immunohistochemistry and reverse transcriptase PCR. Sanja Pajovic performed literature search,

reverse transcriptase PCR and immunohistochemistry. Brigitte L Thériault performed literature

search and RNA expression studies. Nadia L Prigoda-Lee performed literature search, statistical

analysis, grew cell lines, and contributed to figure and manuscript preparation. Clarellen Spencer

performed immunohistochemistry. Helen Dimaras and Timothy W Corson performed literature

searches, assisted in data analysis and conceptualization of discussion, and contributed to figure and

manuscript preparation. Renée Pang performed statistical and bioinformatic analyses on the aCGH

data. Christine Massey performed statistical analyses. Roseline Godbout discovered and

characterised the first tumour with MYCN amplification and normal pRb (RB522) (now

RB1+/+MYCNA) long before anyone believed her; she provided the cell line and Western blot

showing multiple bands of pRb. Zhe Jiang and Eldad Zacksenhaus performed Western blots

specific to phosphorylated pRb. Katherine Paton and Annette C Moll provided and interpreted

clinical data and images. Claude Houdayer and Anthony Raizis provided RB1 mutation analysis,

and clinical data. William Halliday recognized and characterized the unique histological features of

the RB1+/+MYCNA retinoblastomas and prepared digital images for publication. Wan L Lam

supervised Jennifer Kennett and the aCGH experiments. Paul C Boutros performed detailed and

25

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

novel analysis of the aCGH data, and statistical analyses throughout the project, and supervised

Renéee Pang. Dietmar Lohmann performed literature search, provided RB1 mutation analysis, and

contributed to figure construction and development of concepts. Josephine C Dorsman coordinated

the Amsterdam study, and, with Berber Mol who she supervised, recognized the RB1 and MYCN

mutation status of the Amsterdam samples. Brenda L Gallie supervised overall, performed literature

searches, provided critical guidance on all components of the project, and contributed extensively to

figure and manuscript preparation. All authors contributed to concept development and manuscript

preparation.

Conflicts of interest

BLG was part-owner of Solutions by Sequence and Member of the Board of Retinoblastoma

Solutions, which are now merged into Impact Genetics with BLG as Medical Director. All other

authors declare that they have no conflicts of interest.

Acknowledgments

NCI-NIH grant 5R01CA118830-05 supported the early discovery at the Canadian site (BLG).

Canadian Institutes for Health Research grants (WLL) (MOP-86731, MOP-77903, MOP-110949)

supported the aCGH studies, and MOP-77710 (EZ) supported the pRb phosphorylation studies. The

Canadian Retinoblastoma Society, Hyland Foundation and Toronto Netralya and Doctors Lions

Clubs provided critical funding for additional experiments. The Ontario Ministry of Health and

Long Term Care provided infrastructure. Retinoblastoma Solutions and Solutions by Sequence

(Now merged into Impact Genetics) supported the overall project, data analysis and manuscript

preparation. The German study was made possible by grant DFG (Lo 530/6-2), while UK-Essen

provided infrastructure. The Dutch study was made possible by grants from Avanti-STR (JCD, J.

Cloos and ACM) and KiKa (JCD, H. te Riele, J. Cloos, ACM), while VUmc provided

26

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

infrastructure. This study was conducted with the support of the Ontario Institute for Cancer

Research to PCB through funding provided by the Government of Ontario. BMM was funded by a

grant from CCA/V-ICI/ Avanti-STR (to JCD, J. Cloos and ACM)., the Dutch research was also

funded in part by KIKA (JCD, H. te Riele, J. Cloos, ACM). SY was funded by the Vision Science

Research Program of the University Health Network and the University of Toronto. RP was funded

in part by a Great West Life Studentship from Queen’s University School of Medicine. We thank

Leslie MacKeen for the montage of RetCam images in figure 3B. We thank Dr. Valerie White of

U. British Columbia for providing clinical and pathological details and images. We thank Cynthia

Vandenhoven for the clinical images in figure 54F. We thank members of the VU University

Medical Center/The Netherlands Cancer Institute, Institut Curie, Institut für Humangenetik, Toronto

retinoblastoma teams and other, wise colleagues for useful discussions. We thank the children and

families who donated tissues for these studies for the benefit of future families.

27

1

1

2

3

4

5

6

7

8

9

10

11

12

13

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

REFERENCES

1. Knudson AG. Mutation and cancer: statistical study of retinoblastoma. Proceedings of the National Academy of Science, USA 1971; 68(4): 820-3.

2. Friend SH, Bernards R, Rogelj S, Weinberg RA, Rapaport JM, Albert DM, et al. A human DNA segment with properties of the gene that predisposes to retinoblastoma and osteosarcoma. Nature 1986; 323(6089): 643-6.

3. Cavenee WK, Hansen MF, Nordenskjold M, Kock E, Maumenee I, Squire JA, et al. Genetic origin of mutations predisposing to retinoblastoma. Science 1985; 228(4698): 501-3.

4. Lohmann DR, Gallie BL. Retinoblastoma: Revisiting the model prototype of inherited cancer. Am J Med Genet 2004; 129C(1): 23-8.

5. Rushlow D, Piovesan B, Zhang K, Prigoda-Lee NL, Marchong MN, Clark RD, et al. Detection of mosaic RB1 mutations in families with retinoblastoma. Hum Mutat 2009; 30(5): 842-51.

6. Lohmann D, Gallie B, Dommering C, Gauthier-Villars M. Clinical utility gene card for: Retinoblastoma. Eur J Hum Genet 2011; 19(3).

7. Houdayer C, Gauthier-Villars M, Lauge A, Pages-Berhouet S, Dehainault C, Caux-Moncoutier V, et al. Comprehensive screening for constitutional RB1 mutations by DHPLC and QMPSF. Hum Mutat 2004; 23(2): 193-202.

8. Dimaras H, Khetan V, Halliday W, Orlic M, Prigoda NL, Piovesan B, et al. Loss of RB1 induces non-proliferative retinoma: increasing genomic instability correlates with progression to retinoblastoma. Human molecular genetics 2008; 17(10): 1363-72.

9. Corson TW, Gallie BL. One hit, two hits, three hits, more? Genomic changes in the development of retinoblastoma. Genes Chromosomes Cancer 2007; 46(7): 617-34.

10. Ishkanian AS, Malloff CA, Watson SK, DeLeeuw RJ, Chi B, Coe BP, et al. A tiling resolution DNA microarray with complete coverage of the human genome. Nat Genet 2004; 36(3): 299-303.

11. Watson SK, deLeeuw RJ, Horsman DE, Squire JA, Lam WL. Cytogenetically balanced translocations are associated with focal copy number alterations. Hum Genet 2007; 120(6): 795-805.

12. Myllykangas S, Bohling T, Knuutila S. Specificity, selection and significance of gene amplifications in cancer. Semin Cancer Biol 2007; 17(1): 42-55.

13. O'Neill S, Ekstrom L, Lastowska M, Roberts P, Brodeur GM, Kees UR, et al. MYCN amplification and 17q in neuroblastoma: evidence for structural association. Genes Chromosomes Cancer 2001; 30(1): 87-90.

14. Mosse YP, Diskin SJ, Wasserman N, Rinaldi K, Attiyeh EF, Cole K, et al. Neuroblastomas have distinct genomic DNA profiles that predict clinical phenotype and regional gene expression. Genes Chromosomes Cancer 2007; 46(10): 936-49.

15. Chen D, Gallie BL, Squire JA. Minimal regions of chromosomal imbalance in retinoblastoma detected by comparative genomic hybridization. Cancer Genet Cytogenet 2001; 129(1): 57-63.

16. Sampieri K, Amenduni M, Papa FT, Katzaki E, Mencarelli MA, Marozza A, et al. Array comparative genomic hybridization in retinoma and retinoblastoma tissues. Cancer Sci 2009; 100(3): 465-71.

17. Spencer C, Pajovic S, Devlin H, Dinh QD, Corson TW, Gallie BL. Distinct patterns of expression of the RB gene family in mouse and human retina. Gene Expr Patterns 2005; 5(5): 687-94.

18. Templeton DJ, Park SH, Lanier L, Weinberg RA. Nonfunctional mutants of the retinoblastoma protein are characterized by defects in phosphorylation, viral oncoprotein association, and

28

1

1

23456789

10111213141516171819202122232425262728293031323334353637383940414243444546

2

MYCN amplified retinoblastoma with functional pRbRetinoblastoma without RB1 mutations

nuclear tethering. Proceedings of the National Academy of Sciences of the United States of America 1991; 88(8): 3033-7.

19. Murakami A, Yajima T, Sakuma H, McLaren MJ, Inana G. X-arrestin: a new retinal arrestin mapping to the X chromosome. FEBS Lett 1993; 334(2): 203-9.

20. Terry J, Calicchio ML, Rodriguez-Galindo C, Perez-Atayde AR. Immunohistochemical Expression of CRX in Extracranial Malignant Small Round Cell Tumours. The American journal of surgical pathology 2012; 36(8): 1165-9.

21. Tornoczky T, Semjen D, Shimada H, Ambros IM. Pathology of peripheral neuroblastic tumours: significance of prominent nucleoli in undifferentiated/poorly differentiated neuroblastoma. Pathol Oncol Res 2007; 13(4): 269-75.

22. Flexner S. A peculiar glioma (neuroepithelioma?) of the retina. Johns Hopkins Hosp Bull 1891; 2: 115.

23. Dimaras H, Kimani K, Dimba EA, Gronsdahl P, White A, Chan HS, et al. Retinoblastoma. Lancet 2012; 379(9824): 1436-46.

24. Chantada GL, Casco F, Fandino AC, Galli S, Manzitti J, Scopinaro M, et al. Outcome of patients with retinoblastoma and postlaminar optic nerve invasion. Ophthalmology 2007; 114(11): 2083-9.

25. Lillington DM, Goff LK, Kingston JE, Onadim Z, Price E, Domizio P, et al. High level amplification of N-MYC is not associated with adverse histology or outcome in primary retinoblastoma tumours. Br J Cancer 2002; 87(7): 779-82.

26. Zhang J, Benavente CA, McEvoy J, Flores-Otero J, Ding L, Chen X, et al. A novel retinoblastoma therapy from genomic and epigenetic analyses. Nature 2012; 481(7381): 329-34.

27. Manning AL, Longworth MS, Dyson NJ. Loss of pRB causes centromere dysfunction and chromosomal instability. Genes Dev 2010; 24(13): 1364-76.

28. Kobayashi M, Takezawa S, Hara K, Yu RT, Umesono Y, Agata K, et al. Identification of a photoreceptor cell-specific nuclear receptor. Proceedings of the National Academy of Sciences of the United States of America 1999; 96(9): 4814-9.

29. Chen D, Pacal M, Wenzel P, Knoepfler PS, Leone G, Bremner R. Division and apoptosis of E2f-deficient retinal progenitors. Nature 2009; 462(7275): 925-9.

30. Hook KE, Garza SJ, Lira ME, Ching KA, Lee NV, Cao J, et al. An integrated genomic approach to identify predictive biomarkers of response to the aurora kinase inhibitor PF-03814735. Molecular cancer therapeutics 2012; 11(3): 710-9.

29

1

123456789

101112131415161718192021222324252627282930313233

34

35

2