Is sunlight an aetiological agent in the genesis of retinoblastoma?

Abstract

The incidence of unilateral, but not bilateral, retinoblastoma in human populations at different geographical locations increases significantly with ambient erythemal dose of ultraviolet B radiation from sunlight. This supports the hypothesis that sunlight plays a role in retinoblastoma formation.

Full text

Page 1

Retinoblastomas are childhood retinal tumours of which about
40% of all cases are familial (reviewed by Knudson, 1993).
Familial retinoblastoma cases are usually bilateral, unlike sporadic
cases in which only one eye is normally affected. Knudson (1971)
proposed that retinoblastomas arise as a result of two mutational
events, of which one is present in the germline in familial cases.
Recent molecular analysis has confirmed that this is indeed the
case, and demonstrated that both events involve modification or
loss of an allele of the same gene, RB-1, which is commonly
regarded as the paradigm for oncosuppressor genes (Mittnacht,
1998). More than 90% of individuals constitutively heterozygous
for an RB-1 mutation develop retinoblastoma as a result of a
somatic event occurring in one or more cells of the retina, or of its
precursor tissues, that eliminates the function of the wild-type
allele, usually by allele loss (Knudson, 1993).
Three different null alleles of the corresponding murine gene,
designated Rb-1, have been generated independently by gene
targeting (reviewed by Hooper, 1994). In contrast to the situation
in humans, no retinoblastomas have to date been detected in mice
heterozygous for any of the null alleles of Rb-1. There are a
number of possible explanations for this (Hooper, 1994), of which
one is that laboratory mice differ from humans in the level of
exposure of their retinas to sunlight. Here, I have analysed
retinoblastoma incidence at geographical locations differing in
sunlight exposure to test the hypothesis that sunlight plays a role in
its causation.
MATERIALS AND METHODS
Cumulative incidence of retinoblastoma for children of ages 0Ð14
at different geographical locations during various time intervals
between 1969 and 1987 were taken from Parkin et al (1988),
Parkin and Stiller (1995), and from data used in the compilation of
Parkin et al (1992) which were communicated by Dr Max Parkin.
For each population under study, mean values of latitude and
longitude were estimated, and these were used to determine the
annual ambient erythemal dose of UVB by the method of Diffey
and Elwood (1993) using tabulated dose values for clear skies at
each latitude and plots of average percentage cloud cover as a
function of latitude and longitude, which they provide, at intervals
throughout the year and using linear interpolation between tabu-
lated and plotted values. For the purposes of this analysis, cases of
eye tumour with unspecified histology were distributed between
retinoblastoma and other tumour types in the same ratio as the
known cases for the same centre, and this accounts for entries
showing fractional numbers of cases. The data were analysed by
linear regression, weighted to take account of population size,
using the GLIM statistical package (Numerical Algorithms Group,
Oxford, UK).
RESULTS
In a preliminary analysis, I have reported that published values of
the incidence of retinoblastoma taken from Parkin et al (1988) fall
significantly with increasing geographical latitude (Hooper, 1994),
consistent with an aetiological role for sunlight. However, because
of the earthÕs axial tilt, sunlight exposure is greatest at the tropics
rather than at the equator, and it is also influenced by cloud cover
(Diffey and Elwood, 1993). I have, therefore, refined the analysis
(Table 1 and Figure 1) using annual ambient erythemal doses of
UVB calculated from published tables. This analysis shows that
incidence does indeed increase significantly with dose, the
difference between the slope of the regression line and zero being
significant at the 1% level.
Most centres do not report on the laterality of tumours.
However, analysis of data from those centres that do report later-
ality (Table 2 and Figure 2) shows no significant increase in the
incidence of bilateral tumours with annual ambient UV dose, but a
highly significant one in the case of unilateral tumours.
Is sunlight an aetiological agent in the genesis of
retinoblastoma?
ML Hooper
Sir Alastair Currie CRC Laboratories, Department of Pathology, Molecular Medicine Centre, University of Edinburgh, Western General Hospital, Crewe Rd,
Edinburgh EH4 2XU, UK
Summary The incidence of unilateral, but not bilateral, retinoblastoma in human populations at different geographical locations increases
significantly with ambient erythemal dose of ultraviolet B radiation from sunlight. This supports the hypothesis that sunlight plays a role in
retinoblastoma formation.
Keywords: retinoblastoma; epidemiology; sunlight; ultraviolet radiation
1273
British Journal of Cancer (1999) 79(7/8), 1273–1276
© 1999 Cancer Research Campaign
Article no. bjoc.1998.0204
Received 2 July 1998
Revised 31 July 1998
Accepted 5 August 1998
Correspondence to: ML Hooper

Page 2

DISCUSSION
The effects on retinoblastoma incidence documented here, even
when unilateral cases are considered separately, are less severe
than those reported for squamous cell carcinoma of the eye, the
incidence of which varies over more than two orders of magnitude
between centres with high and low UVB exposure (Newton et al,
1996) and which may explain why they have not previously been
detected.
A number of possible reasons must be considered for the effect
seen in Figure 1. First, it could be a consequence of ascertainment
bias, due either to variations in under- or over-reporting of
retinoblastoma or in estimating the size of the population from
which the patients are drawn. It might, for instance, be argued that
data from equatorial regions are least reliable in this context
because they contain a higher proportion of information from
developing countries. Second, it could be due to genetic differ-
ences between the populations under study in the form of differ-
ences in frequency of germline RB-1 mutations, or differences in
frequency of alleles at other loci that influence the frequency of
somatic events leading to loss of heterozygosity at the RB-1 locus
resulting, for instance, from variations in the racial composition of
populations. Third, it could be a direct effect of exposure to UVB,
or to other wavelengths present in sunlight, or to another environ-
mental variable correlated with sunlight exposure; the most
obvious hypothesis is that it is due to an effect of sunlight on
somatic events leading to RB-1 inactivation. Based on this hypo-
thesis, because individuals inheriting a germline RB-1 mutation
almost invariably develop bilateral retinoblastomas, one would
predict that the effect would be due largely or entirely to unilateral
tumours, which in general occur sporadically and require two
somatic events. The data of Figure 2 are consistent with this
prediction.
The analysis shown in Figure 2 argues against the hypothesis
that the effect in Figure 1 is due to ascertainment bias, which
would be expected to affect data on unilateral and bilateral
tumours in a similar fashion; it also argues that it is not due to
differences in RB-1 germline mutation frequency, which would
1274
ML Hooper
British Journal of Cancer (1999) 79(7/8), 1273–1276
© Cancer Research Campaign 1999
Table 1
Incidence of retinoblastoma at different locations
Location LatitudeAnnual
ambient
erythemal
dose
(MED year–1)
Number of
cases
studied
(0–14 years)
Cumulative
incidence
(cases per
million)
Denmark
England and Wales
Estonia
Finland
France
German Democratic
Republic
German Federal
Republic
Hungary
Italy
Netherlands
Norway
Poland, Warsaw
Scotland
Slovakia
Slovenia
Spain
Sweden
Switzerland
China, Shanghai
Hong Kong
Israel (Jews)
Israel (non-Jews)
India, Bangalore
India, Bombay
Japan
Kuwait (Kuwaitis)
Kuwait
(non-Kuwaitis)
Philippines
Singapore
Taiwan, Taipei
Thailand
Nigeria, Ibadan
Uganda
Zimbabwe
56°N
52°N
59°N
61°N
47°N
1450
1620
1270
1140
2080
26.8
363.9
4
58
20
62
42
34
59
44
52°N1670 48 48
51°N
47°N
44°N
52°N
61°N
52°N
56°N
49°N
46°N
40°N
59°N
47°N
31°N
22°N
32°N
32°N
13°N
19°N
35°N
29°N
1740
2080
2560
1650
1100
1690
1350
1940
2140
2890
1270
2070
3450
4160
4320
4320
5450
5160
2990
4910
78
34
36
26
26
12
59
81.8
17
12
87
8
13
22
35
9
10
103
177
5
44
22
61
63
43
53
55
62
41
54
64
32
39
46
35
45
42
70
59
31
29°N
15°N
1°N
25°N
14°N
7°N
4910
4800
4700
3940
5010
5010
5340
5340
853
93
52
27
44
102
101
85
136.9
40
22
32.9
14
29
16
0°
20°S
Canada, western
provinces
Canada, Atlantic
provinces
Cuba
Jamaica
Puerto Rico
Atlanta (whites)
Connecticut (whites)
Detroit (whites)
Greater Delaware Valley 40°N
(whites)
Greater Delaware Valley 40°N
(non-whites)
Iowa (whites)
Los Angeles (blacks)
Los Angeles (whites)
New Mexico (whites)
New York (blacks)
New York (whites)
San Francisco (whites)
Seattle (whites)
Utah (whites)
Brazil, Fortaleza
Brazil, Recife
Brazil, Sao Paulo
Colombia, Cali
Costa Rica
Australia, New South
Wales
Australia, Queensland
South Australia
Australia, Victoria
Western Australia
Fiji (Fijians)
Fiji (Indians)
New Zealand
52°N172090.958
46°N
22°N
18°N
18°N
34°N
42°N
42°N
2100
4880
5400
5140
3340
2290
2430
2580
72.5
120
13
68
10
15
20
63
50
48
62
54
61
41
50
65
2580 19 83
42°N
34°N
34°N
35°N
42°N
42°N
38°N
48°N
40°N
4°S
8°S
24°S
4°N
10°N
2500
3600
3600
3660
2290
2290
3110
1960
2920
5750
5540
4880
4880
5160
31
13
24
70
65
48
37
35
35
67
69
60
111
68
78
88
98
8
10
46
20
26
23
12
27
108
12
30
34°S
25°S
34°S
37°S
32°S
18°S
18°S
41°S
3810
4850
3970
3180
3860
4790
4790
2740
76
25
55
73
49
46
87
50
28
85
5
14
10
6
4
72
LocationLatitude Annual
ambient
erythemal
dose
(MED year–1)
Number of
cases
studied
(0–14 years)
Cumulative
incidence
(cases per
million)
Retinoblastoma incidence data are taken from Parkin et al (1988) and from data used in the compilation of Parkin et al (1992).

Page 3

affect bilateral tumour incidence preferentially. The hypotheses
that it is due to differences in allele frequencies at other loci, or to
an environmental variable correlated with sunlight exposure,
cannot be formally excluded, but a direct effect of sunlight
provides the most ready explanation of the data. This would be
consistent with the mutation spectrum of sporadic retinoblastoma:
first-hit lesions include some C to T transitions, consistent with
action of UV radiation, although instances of the CC to TT change
that is more diagnostic of UV mutagenesis have not been reported,
whereas second-hit lesions are uninformative usually involving
gross chromosomal change (reviewed by Murphree and Munier,
1994). To test this hypothesis further, it would be valuable for
more centres to report the laterality of tumours, and also to report
eye pigmentation so that its effect on incidence can be assessed. It
would also be of interest to determine the critical period for UV
exposure. Most retinoblastomas develop in the first 5 years of life
(Knudson, 1971) and the period of susceptibility to UV may be
restricted to a relatively small fraction of this 5-year interval,
perhaps immediately after birth when the retina is least mature. If
this period is less than a year, it may be possible to detect a varia-
tion in retinoblastoma incidence at a given location as a function of
the month of birth. If not, a study of patients who had migrated
between locations with different UV exposure would be necessary,
posing daunting problems in defining a sufficiently large dataset
and in identifying and quantifying the corresponding Ôat riskÕ
populations.
We have tested the hypothesis that differences in exposure to
ultraviolet radiation between mice and humans are alone sufficient
to account for the lack of retinoblastomas in Rb-1+/Ðmice by
controlled exposure of the mice to fluorescent light with a daylight
spectrum, but have not observed any retinoblastomas in the
exposed mice (JF Armstrong, MH Kaufman and MLH, unpub-
lished observations). This leaves open the possibility that it is
important in combination with other differences. The latter could
include a possible need in the mouse for an additional genetic
event or events involving genes such as p53, p107 or p130
(Hooper, 1994). We have carried out controlled exposure of
Rb-1+/–p53Ð/Ðmice to light without obtaining any retinoblastomas.
However, it has recently been reported (Robanus Maandag et al,
1998) that retinoblastomas developed in 6 out of 14 chimaeric eyes
in mice containing cells doubly homozygous for mutant alleles of
Sunlight and retinoblastoma
1275
British Journal of Cancer (1999) 79(7/8), 1273–1276
© Cancer Research Campaign 1999
Table 2
Incidence of unilateral and bilateral retinoblastoma at different locations
Location Annual
ambient
erythemal
dose
(MED year–1)
Number of
retinoblastoma
cases studied
(0–14 years)
Cumulative
incidence of
unilateral
retinoblastoma
(cases per million)
Cumulative
incidence of
bilateral
retinoblastoma
(cases per million)
Great Britain
India, Bombay
Japan, Osaka
Nigeria, Ibadan
South Africa
(blacks)
USA, Greater
Delaware Valley
(blacks)
USA, Greater
Delaware Valley
(whites)
USA (Navajo
Indians)
USA, SEERa
registry (blacks)
USA, SEER
registry (whites)
Australia,
Queensland
1620
5160
3020
5010
4610
699
63
56
97
78
27.4
59
50
75
80
16.6
10
18
25
18
2580 19 6217
2580633721
3660 15125 28
2790 2843 17
279016741 13
485025 5922
aData from nine population-based registries in different parts of the USA which have no overlap with any other data listed in this table. Incidence data are taken
from Parkin and Stiller (1995) and references cited therein. No other independent datasets in which incidences of unilateral and bilateral retinoblastoma were
recorded separately were identified in a literature search.
120
100
80
60
40
20
0
0 1000
Annual ambient erythemal dose (MED year–1)
200030004000 50006000
Cumulative incidence of retinoblastoma (cases million–1)
Figure 1
for the locations listed in Table 1, plotted as a function of annual ambient
erythemal dose of UVB. The bold continuous line is a linear regression,
whose slope, 0.004100 ± s.e. 0.001334, is significantly different from zero
(t64= 3.073, 0.001 < P < 0.01)
Cumulative incidence of retinoblastoma for children of ages 0–14

Page 4

Rb-1 and p107. This demonstrates that in addition to Rb-1 inacti-
vation, retinoblastoma formation in the mouse requires mutation
of p107 and probably a further event involving an unidentified
gene that occurred somatically in the chimaeras. There is not, at
present, a ready explanation for a difference in requirement for
these additional events between mouse and human. Nonetheless,
whatever the complete explanation for the difference in retino-
blastoma incidence between Rb-1+/–mice and RB-1+/–humans, it
has stimulated an analysis that has revealed a previously unsus-
pected association between UVB exposure and the incidence of
unilateral, but not bilateral, retinoblastoma in human populations.
ACKNOWLEDGEMENTS
I am grateful to Max Parkin for helpful discussion and for commu-
nication of retinoblastoma incidence data, and to Jane Armstrong,
Matt Kaufman and Anton Berns for helpful discussion. Work on
mutant mice in my laboratory was supported by the Medical
Research Council and the Cancer Research Campaign.
REFERENCES
Diffey BL and Elwood JM (1993) Tables of ambient solar ultraviolet radiation for
use in epidemiological studies of malignant melanoma and other diseases. In
Melanoma Epidemiology. Gallacher R and Elwood JM (eds), pp. 81Ð105.
Kluwer: New York
Hooper ML (1994) The role of the p53 and Rb-1 genes in cancer, development and
apoptosis. J Cell Sci 18 (Suppl): 13Ð17
Knudson AG (1971) Mutation and cancer: statistical study of retinoblastoma. Proc
Natl Acad Sci USA 68: 820Ð823
Knudson AG (1993) Antioncogenes and human cancer. Proc Natl Acad Sci USA 90:
10914Ð10921
Mittnacht S (1998) Control of pRb phosphorylation. Curr Opin Genet Dev 8:
21Ð27
Murphree AL and Munier FL (1994) Retinoblastoma. In Retina, 2nd edn. Ryan SJ
(editor-in-chief), pp. 571Ð586. Mosby: St. Louis
Newton R, Ferlay J, Reeves G, Beral V and Parkin DM (1996) Effect of ambient
solar ultraviolet radiation on incidence of squamous-cell carcinoma of the eye.
Lancet 347: 1450Ð1451
Parkin DM and Stiller CA (1995) Childhood cancer in developing countries:
environmental factors. Int J Pediat Hematol Oncol 2: 411Ð417
Parkin DM, Stiller CA, Bieber A, Draper GJ, Terracini B and Young JL (1988)
International Incidence of Childhood Cancer. IARC Scientific Publication no.
87. IARC: Lyon
Parkin DM, Muir CS, Whelan SL, Gao Y-T, Ferlay J and Powell J (1992) Cancer
Incidence in Five Continents, Vol. VI. IARC Scientific Publication no. 120.
IARC: Lyon
Robanus Maandag E, Dekker M, van der Valk M, Carrozza M-L, Jeanny J-C,
Dannenberg J-H, Berns A and te Riele H (1998) p107 is a suppressor of
retinoblastoma development in pRb-deficient mice. Genes Dev 12:
1599Ð1609
1276
ML Hooper
British Journal of Cancer (1999) 79(7/8), 1273–1276
© Cancer Research Campaign 1999
120
100
80
60
40
01000
Annual ambient erythemal dose (MED year–1)
20003000 400050006000
Cumulative incidence (cases million–1)
20
0
140
Figure 2
retinoblastoma for children of ages 0–14 at the locations listed in Table 2,
plotted as a function of annual ambient erythemal dose of UVB. Regression
lines have been fitted separately to unilateral data: broken line, slope
0.01295 ± s.e. 0.001958, significantly different from zero (t9= 6.614,
P < 0.001); and to bilateral data, bold continuous line, slope 0.0003361 + s.e
0.0008433, not significantly different from zero (t9= 0.3986, P > 0.5)
Cumulative incidence of unilateral (squares) and bilateral (circles)