2077
Statistical Methods
The association between ReeseEllsworth group at diagnosis (I/
II/III vs. IV/V) and number of tumors at diagnosis was investigated
using the exact Wilcoxon rank sum test. The Spearman correlation
coefficient was used to examine the association between age at
diagnosis and number of tumors. The cumulative incidence of new
tumor formation was estimated. Although interocular correlation
2078
may exist in patients with bilateral disease,24 the eye was chosen
as the unit of analysis. For eyes with multiple new tumors, the date
of first new tumor formation was used. Enucleation and EBR
before new tumor formation were considered as competing events
in this analysis. Eyes that did not develop new tumors and did not
have other events (enucleation or EBR) were censored at the date
of last follow-up. The Gray test was used to test whether youngness, ReeseEllsworth group, and family history impacted new
tumor formation.25 P values and test statistics (with corresponding
degrees of freedom) resulting from the Gray tests were reported.
This was an exploratory retrospective study, and P values should
be interpreted with caution; no adjustments were made for multiple
tests.
Results
All 58 patients were observed with serial examinations under
anesthesia inclusive of digital fundus photography based on age
and stability of disease. Of them, 34 were male and 13 had a family
history of retinoblastoma (Table 1). Patients median age at
diagnosis was 6.6 months (range, 0.237.3). Forty-eight patients presented with bilateral retinoblastoma. Of the 106 eyes
with retinoblastoma at diagnosis, 52 were classified as being in
ReeseEllsworth groups IIII and 54 were classified as being in
ReeseEllsworth group IV or V. Most eyes were in group V at
diagnosis (n 50 eyes [47%]). There were a total of 237 tumors
at diagnosis in 106 eyes in 58 children. For the 106 eyes with
tumors at diagnosis, the median number of tumors per eye was 2
(range, 19). There was no evidence of an association between
ReeseEllsworth group at diagnosis (I/II/III vs. IV/V) and number
of tumors at diagnosis (P 0.98). There was also no evidence of
an association between age at diagnosis and total number of
tumors in patients (P 0.55, 0.08).
All patients were alive at the time of analysis, with a median
follow-up from diagnosis of 5.0 years (range, 1.0 10.1). Seven
patients (12%) developed a total of 36 new tumors in 11 eyes
(Tables 2, 3 [available at http://aaojournal.org]); 6 tumors formed
in the macula, 16 in the midzone, and 14 in the periphery. The
median time from start of chemotherapy to detection of the first
new tumor was 3 months (range, 115); the latest time a new
tumor was noted to form was 24 months from the start of chemotherapy. One of these 7 patients had bilateral disease at diagnosis
and developed new tumors in both eyes approximately 1 month
after starting chemotherapy. The other 6 patients had unilateral
disease at diagnosis. Three of these 6 patients developed new
tumors in the opposite eye at approximately 3.0, 7.2, and 7.5
months after initiation of chemotherapy. The other 3 patients
developed new tumors in both eyes. Of note, none of these eyes
had EBR before development of new tumors. In addition, none of
these eyes were subsequently enucleated.
Cumulative incidences of new tumor formation at 1 and 2 years
were estimated to be 83% and 103%, respectively. Figure 1
shows the cumulative incidence of new tumor formation for all
eyes. Family history was a significant predictor of new tumor
formation (P0.001, 21 17.8). Eyes of patients with a family
history were more likely to develop new tumors than eyes of
patients without a family history; 2-year cumulative incidences
were 319% and 32% for eyes with and without family histories, respectively (Fig 2). Age at diagnosis was also a significant
predictor of new tumor formation (P0.001, 21 13.9). Eyes in
patients who were younger than 6 months at diagnosis were more
likely to develop new tumors (2-year cumulative incidence,
216%) than eyes in patients who were at least 6 months old at
diagnosis (2-year cumulative incidence, 00%) (Fig 3). All pa-
N (%)
Gender
Male
Female
Age at diagnosis (mos)
Birth to 1
13
36
69
912
1215
1518
1821
2124
24
Unilateral
Bilateral
Family history
No
Yes
Chemotherapy
VC 4
VC 6
VC 7
VC 8
VC 8 VCE 3
VC 8 VT 3
VCE 6
ReeseEllsworth classification at diagnosis (n 106 eyes)
I
II
III
IV
V
34 (59%)
24 (41%)
6 (10%)
10 (17%)
11 (19%)
7 (12%)
10 (17%)
3 (5%)
0 (0%)
4 (7%)
3 (5%)
4 (7%)
10 (17%)
48 (83%)
Discussion
It has been estimated that hereditary retinoblastoma patients
will develop, on average, 2.7 tumors per eye.2 New tumors
will continue to form after initial diagnosis. Whether a
patients genetic predisposition to form new tumors can be
altered by treatment is unknown. We studied the impact of
primary systemic chemotherapy on the rate of new tumor
formation in a cohort of 58 patients with hereditary retinoblastoma. We found an overall cumulative incidence of 10%
at 2 years after the start of chemotherapy. Younger age at
diagnosis, family history of retinoblastoma, and lower
ReeseEllsworth classification were factors that significantly increased the likelihood of new tumor formation (P
0.001, P 0.001, and P 0.021, respectively). Eyes of
patients with a family history had a 31% incidence of new
tumor formation at 2 years, whereas those patients diagnosed before 6 months of age had a 21% incidence. Eyes
classified as ReeseEllsworth groups I to III had a 10%
incidence of new tumors. Each of these risk factors is intrinsically tied together. Patients with a family history are more
likely to be screened starting at birth and therefore diagnosed at
a younger age with less advanced intraocular disease.
We observed 36 new tumors in 11 eyes. Of these, 17%
(n 6) were in the macula, 44% (n 16) were in the
midzone, and 39% (n 14) were in the periphery. New
tumor formation in retinoblastoma appears to be closely tied
to retinal development. Immature retinal cells remain susceptible to second random mutations leading to inactivation
of the retinoblastoma protein and tumor development. As
the retina matures from the posterior pole towards the
45 (78%)
13 (22%)
1 (2%)
1 (2%)
5 (9%)
44 (76%)
2 (3%)
2 (3%)
3 (5%)
15 (14%)
24 (22%)
13 (12%)
4 (4%)
50 (47%)
0.15
0.10
0.05
0.00
0
10
12
Figure 1. Cumulative incidence estimates of new tumor formation for all eyes (n 116). X-axis, years.
2079
Figure 2. Cumulative incidence estimates of new tumor formation by family history (yes [solid line] vs. no [dashed line]) (P0.001, 21 17.8). X-axis,
years.
developed intrinsic vascularity may not have allowed cytotoxic levels of the chemotherapeutic agents to be achieved.
In these instances, the tumors growth may have been
blunted, but it continued to grow either at the completion of
chemotherapy or once resistance had been developed. This
hypothesis is in keeping with the finding of Gombos et al
that tumors 2 mm in diameter are less likely to respond to
chemotherapy.26 They concluded that the poorly developed
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
0
10
12
Figure 3. Cumulative incidence estimates of new tumor formation by age at diagnosis (6 months [solid line] vs. 6 months [dashed line]) (P0.001,
21 13.9). X-axis, years.
2080
0.15
0.10
0.05
0.00
0
10
12
Figure 4. Cumulative incidence estimates of new tumor formation by ReeseEllsworth classification (I/II/III [solid line] vs. IV/V [dashed line]) (P 0.021,
21 5.3). X-axis, years.
2081
References
1. Ellsworth RM. The practical management of retinoblastoma.
Trans Am Ophthalmol Soc 1969;67:462534.
2. Abramson DH, Gombos DS. The topography of bilateral
retinoblastoma lesions. Retina 1996;16:2329.
3. Abramson DH, Greenfield DS, Ellsworth RM. Bilateral
retinoblastoma: correlations between age at diagnosis and time
course for new intraocular tumors. Ophthalmic Paediatr Genet
1992;13:17.
4. Salmonsen PC, Ellsworth RM, Kitchin FD. The occurrence of
new retinoblastoma after treatment. Ophthalmology 1979;86:
837 43.
5. Merrill PT, Buckley EG, Halperin EC. New and recurrent
tumors in germinal retinoblastoma: is there a treatment effect?
Ophthalmic Genet 1996;17:115 8.
6. Hernandez JC, Brady LW, Shields JA, et al. External beam
radiation for retinoblastoma: results, patterns of failure and a
proposal for treatment guidelines. Int J Radiat Oncol Biol
Phys 1996;35:12532.
7. Messmer EP, Sauerwein W, Heinrich T, et al. New and recurrent
tumor foci following local treatment as well as external beam
radiation in eyes of patients with hereditary retinoblastoma.
Graefes Arch Exp Clin Exp Ophthalmol 1990;228:426 31.
8. Abramson DH, Frank CM. Second nonocular tumors in survivors of bilateral retinoblastoma: a possible age effect on
radiation-related risk. Ophthalmology 1998;105:5739.
9. Wilson MW, Rodriguez-Galindo C, Haik BG, et al. Multiagent chemotherapy as neoadjuvant treatment for multifocal
intraocular retinoblastoma. Ophthalmology 2001;108:2106
14, discussion 2114 5.
10. Rodriguez-Galindo C, Wilson MW, Haik BG, et al. Treatment
of intraocular retinoblastoma with vincristine and carboplatin.
J Clin Oncol 2003;21:2019 25.
11. Shields CL, De Potter P, Himelstein B, et al. Chemoreduction
in the initial management of intraocular retinoblastoma. Arch
Ophthalmol 1996;114:1330 8.
12. Friedman DL, Himelstein B, Shields CL, et al. Chemoreduction and local ophthalmic therapy for intraocular retinoblastoma. J Clin Oncol 2000;18:127.
2082
Gender
Laterality
Family History
Chemotherapy Agents
1
2
3
4
5
6
7
M
F
F
M
M
M
F
4.8
1.5
0.3
1.6
0.2
2.8
2.8
Unilateral
Unilateral
Unilateral
Unilateral
Unilateral
Bilateral
Unilateral
No
No
Yes
Yes
Yes
Yes
No
5.8
5.1
11.0
9.5
5.0
9.5
6.4
VC 8
VC 8
VC 8 VT 3
VC 8 VCE 3
VC 8
VC 8 VT 3
VC 8
ReeseEllsworth
Group
Age at
Diagnosis (mos)
Age at First
New Tumor (mos)
Ia
4.8
1.5
0.3
0.3
1.6
12
9
3
6
3
1
1
1
2
9
1.6
15
Right
0.2
5
6
Left
Right
1
3
Ia
IIa
0.2
2.8
15
4
1
10
Left
IIa
2.8
Right
2.8
Patient
Eye
Tumors at
Diagnosis
1
2
3
3
4
Left
Right
Right
Left
Right
0
0
1
0
1
Left
Ia
New Tumor
Location
Periphery, 1
Macula, 1
Midzone, 1
Macula, 2
Macula, 1;
midzone, 4;
periphery, 4
Macula, 1;
midzone, 3;
periphery, 1
Macula, 1;
midzone, 1
Midzone, 1
Midzone, 4;
periphery, 6
Midzone, 2;
periphery, 1
Periphery, 1
Time from
Chemotherapy
Start to First New
Tumor (mos)
7
7
2
5
1
13
3
15
1
1
3
2082.e1