- Email: [email protected]
Enucleation versus preservation of blind eyes following plaque radiotherapy for choroidal melanoma
Enucleation versus preservation of blind eyes following plaque radiotherapy for choroidal melanoma James J. Augsburger,* MD; Lucie Khouri,t MD; Athanasios Roumeliotis, t MD; Robert C. Kersten, t§ MD; Dwight R. Kulwin,t§ MD; Susan Schneider,*:j: MD ABSTRACT Background: Currently available information about patients with posterior uveal melanoma treated by plaque radiotherapy is insufficient to determine what to do about eyes that become blind as a consequence of the tumour and its treatment. Should they be enucleated, or is ocular preservation just as good in terms of survival? Methods: We performed a retrospective survival analysis of secondary enucleation versus ocular preservation in patients with a posterior uveal melanoma treated by plaque radiotherapy whose irradiated eye became completely blind following that treatment. Of the 79 patients who fulfilled defined inclusion criteria, 25 underwent secondary enucleation of the blind eye, and 54 retained their irradiated blind eye. Results: Most of the baseline demographic and tumour-related variables evaluated were similarly distributed between the subgroups. The 5-year, I0-year and IS-year all-cause death rates in the secondary enucleation subgroup were 24.7%, 51.5% and 52.0% respectively, and those in the ocular preservation subgroup were 7.4%, 32.9% and 48.1% respectively. In spite of the apparent slight difference between the curves, the difference was not statistically significant (p = 0.41, Mantei-Haenszel test). Interpretation: Although a retrospective study of this type has several limitations, our results suggest that secondary enucleation is not likely to substantially improve survival of patients whose irradiated eye becomes totally blind following plaque radiotherapy for choroidal or ciliochoroidal melanoma.
C
urrently available survival data for patients with a primary posterior uveal melanoma treated by plaque radiotherapy are inadequate to determine the advisability of enucleating or retaining the irradiated eye if it becomes blind subsequent to the treatment. Some experienced clinicians believe that secondary enucleation of such eyes is likely to improve the patient's survival prognosis and should therefore be
recommended uniformly. With equal conviction, other experienced clinicians believe that enucleation of such eyes is unlikely to improve survival and should therefore not be recommended unless other factors (such as severe eye pain or intraocular tumour relapse 1) prompt this surgery. We attempted to address this issue by means of a retrospective survival study. The principal objective
From *the Oncology Service, tthe Oculoplastic and Reconstructive
Correspondence to: Dr. James J. Augsburger, Department of Ophthalmology, Suite 350, Health Professions Building, 3223 Eden Ave., Cincinnati OH 45267-0527, USA; fax (513) 558-3108, [email protected]
Surgery Service and tthe Ophthalmic Pathology Service, Department of Ophthalmology, University of Cincinnati College of Medicine, Cincinnati, Ohio, and §Cincinnati Eye Institute, Cincinnati, Ohio
372
Enucleation vs. preservation-Augsburger et al
Can J Ophthalmol 2004;39:372-9
Enucleation vs. preservation-Augsburger et al of our study was to determine the effect of secondary enucleation on melanoma-specific mortality in plaque-treated eyes of patients whose affected eye became blind subsequent to plaque radiotherapy. A secondary objective was to identify significant prognostic factors for secondary enucleation of eyes that become blind following plaque therapy and for melanoma-specific mortality in such patients. METHODS
We performed a retrospective analysis of patients in the senior author's (J.J.A.) computerized database of patients with primary choroidal or ciliochoroidal melanoma treated by plaque radiotherapy between 1980 and 1997 (n = 1074). We first identified all patients who had become completely blind (i.e., visual acuity of no light perception) in the tumourcontaining eye subsequent to plaque radiotherapy for the intraocular tumour. We excluded from this group all patients whose irradiated eye still had some vision (i.e., visual acuity of light perception or better) at the time of secondary enucleation, all patients in whom metastatic uveal melanoma developed before the eye became completely blind, and all patients whose blind eye had not been enucleated but who had been followed for less than 6 months after the eye went blind. We classified the patients who fulfilled our inclusion criteria into two subgroups: those treated by secondary enucleation, and those in whom the blind eye continued to be managed without enucleation. We evaluated numerous baseline demographic and tumour-related variables in this study. The senior author evaluated all these variables prospectively at the time of comprehensive standardized ophthalmologic examination, which immediately preceded radioactive plaque implantation. The methods used to evaluate these variables have been reported elsewhere. 2- 5 Plaque radiotherapy was performed using cobalt 60 plaques (1980 through 1987), iodine 125 plaques (1982 through 1997), iridium 192 plaques (1982 through 1988) or ruthenium 106 applicators (1984 through 1989). The methods used to implant and remove the plaques and to estimate radiation dose rates and dosage distributions in these cases have previously been reported. 6 The target dosage of radiation to the tumour apex was 80 Gy to 100 Gy for all patients in this group. The senior author performed the plaque surgery in all cases. Following plaque radiotherapy, patients were mon-
itored periodically to assess local tumour response to therapy and various other ocular variables, including visual acuity in the treated eye and presence or absence of specific treatment-related complications (radiation-induced cataract, radiation retinopathy, radiation optic papillopathy, iris neovascularization, neovascular glaucoma, vitreous hemorrhage and radiation-induced scleral necrosis). The recommended ophthalmologic follow-up schedule for all patients was every 3 months after plaque therapy for 3 years, every 6 months for the next 2 years, and yearly thereafter. The systemic status of plaque-treated patients (i.e., presence or absence of metastatic uveal melanoma and survival) was determined by communication with their family practitioners, internists and oncologists. We evaluated the survival status of patients who had been lost to ophthalmologic followup by searching the US National Death Index database.7 The cause of death for patients known to have died before the study closure date (March 2003) was determined by communication with the attending physician at the time of death, review of medical records from the terminal illness and, when available, review of official death certificates. We performed the following predetermined data analyses on the tabulated variables: 1) comparative analysis of the distributions of the baseline clinical variables in the two patient subgroups, using the unpaired two-tailed t test for comparison of the means of the intrinsically continuous numeric variables and Pearson's X2 test for comparison of the frequency distributions of the categoric variables; 2) analysis of potential prognostic factors for secondary enucleation using both univariate and multivariate Cox proportional hazards modelling; 3) computation and plotting of Kaplan-Meier cumulative actuarial survival curves based on deaths from metastatic uveal melanoma and deaths from any cause for the two patient subgroups; 4) comparison of the survival curves of the two patient subgroups using the Mantel-Haenszel X2 test; 5) analysis of potential prognostic factors for death from metastatic uveal melanoma and death from any cause using both univariate and multivariate Cox proportional hazards modelling; and 6) estimation of the adjusted effect of secondary enucleation on death from metastatic uveal melanoma by computing the rate ratio (and 95% confidence interval) for secondary enucleation when added to the best Cox proportional hazards regression model containing only significant predictor variables. Because of the exploratory nature
CAN JOPHTHALMOL-VOL.39, NO.4, 2004
J 73
Enucleation vs. preservation-Augsburger et al
of this study, the multiplicity of variables being evaluated and the expectation of a limited number of patients who would fulfil all the inclusion criteria, we selected the conservative a value of 0.01 for assigning statistical significance to observed differences between the subgroups in the various comparative analyses. We performed the multivariate Cox proportional hazards modelling using a stepwise, step-down method and p > 0.05 to reject variables. RESULTS
Our computerized database search identified 185 patients (17.2% of the 1074 patients treated by plaque radiotherapy between 1980 and 1997) whose melanoma-containing eye had become completely blind during available follow-up. We excluded from this group 99 patients whose eye became blind at the time of secondary enucleation even though it still retained visual acuity of light perception or better up to that instant, 3 patients in whom metastatic uveal melanoma had developed before the date the eye became blind, and 4 surviving patients whose blind eye had been preserved but who had been followed for less than 6 months after the eye became blind. We thus studied 79 patients, of whom 25 (31.6%) had undergone secondary enucleation of the blind eye and 54 (68.4%) retained the blind eye through all available follow-up. Tables 1 and 2 contain summary information for the
79 patients and their tumours at baseline examination before plaque therapy. The two groups were reasonably similar in terms of all of the baseline variables evaluated. The median length of follow-up (the interval between the date of plaque radiotherapy and the date of the most recent patient encounter) was 10.5 years for the secondary enucleation subgroup and 11.4 years for the ocular preservation subgroup. The mean interval between plaque radiotherapy and total loss of vision in the eye was 3.2 (standard deviation 2.3) years and 4.4 (standard deviation 2.6) years respectively, a nonsignificant difference (t = 1.86, p = 0.066). Fourteen eyes (56.0%) in the secondary enucleation subgroup became painful as well as blind during available follow-up, compared with 6 eyes (11.1%) in the ocular preservation subgroup, a highly significant difference (X 2 = 15.9, p < 0.001). This finding indicates that secondary enucleation was much more likely when the eye was painful than when it was not. During available follow-up, vitreous hemorrhage developed in the treated eye in 10 eyes (40.0%) in the secondary enucleation subgroup and 24 eyes (44.4%) in the ocular preservation subgroup. This difference was not statistically significant (X 2 = 0.036, p = 0.85). During follow-up, local relapse of the irradiated tumour was detected in four eyes (16.0%) in the secondary enucleation subgroup and in seven eyes (13.0%) in the ocular preservation subgroup, a nonsignificant difference (X 2 < 0.01, p = 0.99).
Table !-Baseline continuous numeric clinical variables in patients with primary choroidal or ciliochoroidal melanoma treated by plaque radiotherapy who underwent secondary enucleation or whose blind eye was managed without enucleation Group; mean (and standard deviation)
Variable Age,yr Anteroposterior diameter of tumour, mm Circumferential diameter of tumour, mm Largest basal diameter of tumour, mm Tumour thickness, mm Largest linear dimension of tumour, mm
J 74
CAN J OPHTHALMOL-VOL.39, NO.4, 2004
Secondary enucleation (n = 25)
Ocular preservation (n =54)
61.2 ( 12.1) 11.3 10.5
value
p value
62.5 (11.3)
0.46
0.64
(3.4)
12.7 (3.6)
1.64
0.11
(3.3)
11.9 (3.7)
1.63
0.11
11.6 (3.2) 8.2 (3.4)
13.0 (3.5) 7.9 (2.9)
1.71 0.38
0.09 0.71
11.7 (3.3)
13.1
1.66
0.10
(3.5)
t
Enucleation vs. preservation-Augsburger et al
Table 2-Baseline categoric clinical variables in the two subgroups Group; no. (and %) of patients Variable Sex Male Female Location of anterior tumour margin At or posterior to ocular equator Anterior to equator but at or posterior to ora serrata Anterior to ora serrata in ciliary body Location of posterior tumour margin relative to optic disc Extending to optic disc margin Extending to within 3 mm of optic disc margin Over 3 mm from optic disc margin Location of posterior tumour margin relative to foveola Extending to or beneath foveola Extending to within 3 mm of foveola Over 3 mm from foveola Visual acuity before treatment ~ 20/SO < 20/SO to ~ 20/200 < 20/200
Secondary enucleation
Ocular preservation
12 (48.0) 13 (S2.0)
30 (SS.6) 24 (44.4)
II (44.0)
16 (29.6)
s (20.0)
22 (40.7)
9 (36.0)
16 (29.6)
3 (12.0)
3
14 (2S.9)
16 (64.0)
37 (68.S)
(8.0)
I
O.IS
0.70
3.40
0.18
1.01
0.60
1.80
0.41
1.04
O.S8
(1.8)
7 (28.0) 16 (64.0)
IS (27.8) 38 (70.4)
2 (8.0) 8 (32.0) IS (60.0)
9 ( 16.7) IS (27.8) 30 (SS.6)
Univariate analysis of potential prognostic factors for secondary enucleation identified eye pain (p < 0.001) and the interval between plaque therapy and onset of blindness (p = 0.003, shorter interval unfavourable) as the only significant factors. Multivariate Cox proportional hazards regression analysis identified eye pain as the only independently significant prognostic factor associated with secondary enucleation following plaque radiotherapy for choroidal or ciliochoroidal melanoma (p < 0.001). The interval between plaque radiotherapy and onset of blindness (shorter interval unfavourable) and the largest linear tumour dimension (larger size unfavourable) were of borderline significance (0.10 ~p > 0.01). Surprisingly, neither local tumour relapse nor
p value
(S.6)
6 (24.0)
2
x2
vitreous hemorrhage was significantly associated with secondary enucleation by multivariate analysis. During available follow-up, 13 patients (52.0%) in the secondary enucleation subgroup and 22 patients (40.7%) in the ocular preservation subgroup died. The cause of death for the 35 patients is shown in Table 3. The Kaplan-Meier survival curves based on the end point "death from metastatic uveal melanoma" in the two study subgroups are presented in Fig. 1, and those based on the end point "death from any cause" are presented in Fig. 2. There were slight apparent differences between the corresponding curves, but these differences were not statistically significant (see figure legends for details).
CAN JOPHTHALMOL-VOL. 39, NO.4, 2004
J 75
Enucleation vs. preservation-Augsburger et al
ative to the optic disc (p = 0.050, closer to disc unfavourable) as prognostic factors of borderline statistical significance for this end point. Multivariate Cox proportional hazards regression analysis identified largest linear dimension of the treated tumour at baseline as the only independently significant prognostic factor for death from metastatic uveal melanoma (p < 0.001). The interval between plaque radiotherapy and onset of blindness was of borderline statistical significance (p = 0.082, shorter interval unfavourable) in a two-term regression that included largest linear tumour dimension. Surprisingly, local tumour relapse was not significantly associated with death from metastatic melanoma by multivariate analysis. Using a Cox model that included both largest linear tumour dimension and secondary enucleation, we computed a rate ratio of 0.90 (95% confidence interval 0.56-1.46) for secondary enucleation. The fact that the rate ratio is less than 1.0 indicates that the adjusted probability of death from metastatic uveal melanoma was slightly higher in the ocular preservation subgroup than in the secondary enucleation subgroup. However, because the confidence interval
Table 3-Cause of death for the patients who died during follow-up Group; no. (and %) of patients Cause of death
Secondary enucleation
Ocular preservation
5 (20.0) 2 (8.0) 6 (24.0)
14 (25.9) 2 (3.7) 6 (II. I)
Metastatic uveal melanoma Nonmelanoma cancer Other
Univariate analysis of potential prognostic factors for death from metastatic uveal melanoma identified all of the variables related to tumour size except tumour thickness as significant (p < O.Ol).lt also identified the interval between treatment and onset of blindness (p = 0.012, shorter interval unfavourable), age at the time of treatment (p = 0.019, older age unfavourable), visual acuity in the eye at the time of treatment (p = 0.023, worse visual acuity unfavourable) and location of the posterior tumour margin rei-
1.0 t-
t- ,.....,
0.9 r-- · - t--
ta
0.8
-
0.7
·s;> ~
::::s
(/)
0.6
0
0.5
>.t:::
.c ta .c
-
1......_, ;--,
.., f-1-lh
0.4 0.3
0~
0.2
Q.
0.1
-
1-
r-- -·
'--
5
-
-
10
f--
15
20
Time from plaque therapy, yr Secondary enucleation
- - Ocular preservation
Fig. 1-Kaplan-Meier cumulative actuarial survival curves for the 25 patients in the secondary enucleation subgroup and the 54 patients in the ocular preservation subgroup. The curves are based on deaths from metastatic uveal melanoma only. The difference between the curves is not statistically significant (Mantei-Haenszel X2 = 0.15, p = 0.70).
J 76
CAN J OPHTHALMOL-VOL. 39, NO.4, 2004
Enucleation vs. preservation-Augsburger et a1
1.0 0.9
co> ·s:'-
0.8
tn
0.6
0
0.5
::::J
>
:!::::
:0
"'0
.c '-
c.
-c--r
i---4.
M 1-
0.7
0.4 0.3
-
w
L
L.
..__,
"1..,
-
h
-
..
~-
~~
r-- D
-
-r-
.,
-!--
....
-
1--
0.2 0.1
!--
I
-
!--
'
I I
I I
I
I 5
10
15
20
Time from plaque therapy, yr Secondary enucleation
- - Ocular preservation
Fig. 2-Kaplan-Meier cumulative actuarial survival curves for the two subgroups. The curves are based on deaths from any cause. The difference between the curves is not statistically significant (Mantei-Haenszel X2 = 0.68, p = 0.41 ).
broadly straddles the value 1.0, we regard the adjusted effect of secondary enucleation on melanoma-specific mortality to be not statistically significant. INTERPRETATION
The principal result of our study seems to be that secondary enucleation of eyes that become totally blind following plaque radiotherapy for choroidal or ciliochoroidal melanoma does not substantially reduce (or increase) the patient's risk of dying from metastatic uveal melanoma. This result should be reassuring to patients and surgeons faced with the decision to enucleate or to retain such eyes. It suggests that both enucleation and preservation of such eyes are acceptable management options because they are likely to yield a similar survival prognosis. Because of the counterintuitive nature of this result and several recognized limitations of our study, however, we urge readers to regard this result as inconclusive. Our finding needs to be confirmed (or refuted) in independent groups of patients before its validity can be accepted. The principal limitation of our work is the small
size of our study group. We had only 79 eyes in the total study group, only 25 eyes that had been enucleated secondarily, and only 19 deaths from metastatic uveal melanoma during available follow-up. The smaller the number of patients and achieved end points in a study, the greater the likelihood that the subgroups will differ substantially from the whole population of patients meeting the inclusion criteria of the study by chance alone. The fact that our data analysis did not identify local tumour relapse as an unfavourable prognostic factor for survival - even though two prior reports, based on the larger study group from which these patients were selected, demonstrated a substantial effect of local tumour relapse 8•9 - emphasizes this point. There is no way in which we can correct for this limitation by data analysis. We therefore acknowledge that the power of our study to identify a substantial favourable effect of secondary enucleation on survival is extremely low. A second limitation of our work is the restrictiveness of our inclusion and exclusion criteria. By requiring eyes to be completely blind (i.e., visual acuity of no light perception), we may have inadvertently eliminated the patients with the most aggressive uveal
CAN JOPHTHALMOL-VOL 39, NO.4, 2004
377
Enucleation vs. preservation-Augsburger et al
melanomas. In our study, the mean interval between plaque radiotherapy and onset of total blindness was 3.2 years in the secondary enucleation subgroup and 4.4 years in the ocular preservation subgroup. Patients in this study had to survive long enough for their irradiated eyes to go blind. Patients whose irradiated eyes might have gone blind but who experienced metastasis within the first 2 to 3 years after plaque treatment could not have appeared in our study group. The results also might have been substantially different if we had defined blindness as visual acuity of 5/200 or worse, counting fingers or worse, or light perception. Similarly, changes in our exclusion criteria could have influenced the observed results. A third limitation is the limited potential of survival modelling methods in general and of Cox proportional hazards modelling specifically to identify prognostic factors for specified patient outcomes. There is no inherent reason why clinical data should conform to any specific survival model. If we had used a parametric modelling method 10 or perhaps Cox modelling with time-dependent variables, 11 we might have obtained a better fit with the observed data. A fourth limitation is the nonrandomized assignment of patients to the two subgroups. It is entirely possible that unknown, unevaluated factors associated with the decision to enucleate or to preserve certain eyes influenced the survival outcome in indeterminate ways. A fifth limitation is the exploratory nature of our study. To our knowledge, no other group has published or presented results of a study designed to estimate the effect of secondary enucleation on death from metastatic melanoma in patients who were treated initially by plaque radiotherapy. We acknowledge the need for other investigators to study similar but independent groups of patients and provide either supportive or contradictory results. Finally, a sixth limitation is our probable failure to evaluate some pertinent clinical variables and the inability to assess pertinent pathological variables. It is certainly possible that we failed to evaluate one or more important clinical variables; however, we selected virtually all clinical variables that have been found by our group 2- 5 and other investigators 12•13 to be of predictive value for the outcomes evaluated in this study. A more global limitation that applies to all plaque-treated uveal melanoma groups is the inability to assess and therefore evaluate histopathological variables that have been associated with death from metastatic uveal melanoma. 14- 18 Assessing such vari-
3 78
CAN JOPHTHALMOL-VOL 39, NO.4, 2004
abies, of course, requires enucleating the eyes or, alternatively, excising the tumours intact. In conclusion, this exploratory retrospective study appears to show that secondary enucleation of eyes that become totally blind following plaque radiotherapy for choroidal or ciliochoroidal melanoma does not substantially decrease (or increase) the patient's risk of dying from metastatic uveal melanoma. For the reasons mentioned above, we regard these results as suggestive rather than definitive. We encourage other investigators to study this matter and report their independent results and conclusions in the near future. This work was supported by Research to Prevent Blindness, Inc., New York, and the Medical Education and Research Fund and Quest for Vision Fund, Department of Ophthalmology, University of Cincinnati College of Medicine, Cincinnati, Ohio. REFERENCES
1. Shields CL, Shields JA, Karlsson U, Markoe AM, Brady LW. Reasons for enucleation after plaque radiotherapy for posterior uveal melanoma: clinical findings. Ophthalmology 1989;96:919-24. 2. Augsburger JJ, Gamel JW, Sardi VF, Greenberg RA, Shields JA, Brady LW. Enucleation vs cobalt plaque radiotherapy for malignant melanomas of the choroid and ciliary body. Arch Ophthalmol1986;104:655-61. 3. Augsburger JJ, Gamel JW, Lauritzen K, Brady LW. Cobalt-60 plaque radiotherapy vs enucleation for posterior uveal melanoma. Am J Ophthalmol1990;109:585-92. 4. Augsburger JJ, Correa ZM, Freire J, Brady LW. Longterm survival in choroidal and ciliary body melanoma after enucleation versus plaque radiation therapy. Ophthalmology 1998;105:1670-8. 5. Augsburger JJ, SchneiderS, Freire J, Brady LW. Survival following enucleation versus plaque radiotherapy in statistically matched subgroups of patients with choroidal melanomas: results in patients treated between 1980 and 1987. Graefes Arch Clin Exp Ophthalmol 1999;237: 558-Q7. 6. Giblin ME, Shields JA, Augsburger JJ, Brady LW. Episcleral plaque radiotherapy for uveal melanoma. Aust N Z J Ophthalmol1989;17:153-Q. 7. National Center for Health Statistics, Centers for Disease Control and Prevention, US Department of Health and Human Services. National Death Index. Available: http://www.cdc.gov/nchs/r&dlndi/ndi.htm (accessed 2004 Apr 15). 8. Karlsson UL, Augsburger JJ, Shields JA, Markoe AM, Brady LW, Woodleigh R. Recurrence of posterior uveal melanoma after 6°Co episcleral plaque therapy. Ophthalmology 1989;96:382-8. 9. Vrabec TR, Augsburger JJ, Gamel JW, Brady LW,
Enucleation vs. preservation-Augsburger et al
10.
11.
12.
13.
14.
Hernandez C, Woodleigh R. Impact of local tumor relapse on patient survival after cobalt 60 plaque radiotherapy. Ophthalmology 1991;98:984-8. Gamel JW, McLean IW, Rosenberg SH. Proportion cured and mean log survival time as functions of tumour size. Stat Med 1990;9:999-1006. Fisher LD, LinDY. Time-dependent covariates in the Cox proportionru hazards regression model. Annu Rev Public Health 1999;20:145-57. Wilson MW, Hungerford JL. Comparison of episcleral plaque and proton beam radiation therapy for the treatment of choroidal melanoma. Ophthalmology 1999;106: 1579-87. Lommatzsch PK, Werschnik C, Schuster E. Long-term follow-up of Ru-1 06/Rh-106 brachytherapy for posterior uveal melanoma. Graefes Arch Clin Exp Ophthalmol 2000;238:129-37. Seddon JM, Albert DM, Lavin PT, Robinson N. A prognostic factor study of disease-free interval and survival following enucleation for uveal melanoma. Arch Ophthal-
moll983;101:1894-9. 15. Seddon JM, Polivogianis L, Hsieh CC, Albert DM, Gamel JW, Gragoudas ES. Death from uveal melanoma: number of epithelioid cells and inverse SD of nucleolar area as prognostic factors. Arch Ophthalmoll987;105:801-6. 16. Folberg R, Rummelt V, Parys-Van Ginderdeuren R, Hwang T, Woolson RF, Pe'er J, et al. The prognostic value of tumor blood vessel morphology in primary uveal melanoma. Ophthalmology 1993;100:1389-98. 17. Seregard S, Kock E. Prognostic indicators following enucleation for posterior uveal melanoma. A multivariate analysis of long-term survival with minimized loss to follow-up. Acta Ophthalmol Scand 1995;73:340-4. 18. Mlikitie T, Summanen P, Tarkkanen A, Kivela T. Microvascular density in predicting survival of patients with choroidal and ciliary body melanoma. Invest Ophthalmol Vis Sci 1999;40:2471-80. Key words: choroidal melanoma, ciliochoroidal melanoma, plaque radiotherapy, enucleation, survival
CAN J OPHTHALMOL-VOL. 39, NO.4, 2004
J 79
C
urrently available survival data for patients with a primary posterior uveal melanoma treated by plaque radiotherapy are inadequate to determine the advisability of enucleating or retaining the irradiated eye if it becomes blind subsequent to the treatment. Some experienced clinicians believe that secondary enucleation of such eyes is likely to improve the patient's survival prognosis and should therefore be
recommended uniformly. With equal conviction, other experienced clinicians believe that enucleation of such eyes is unlikely to improve survival and should therefore not be recommended unless other factors (such as severe eye pain or intraocular tumour relapse 1) prompt this surgery. We attempted to address this issue by means of a retrospective survival study. The principal objective
From *the Oncology Service, tthe Oculoplastic and Reconstructive
Correspondence to: Dr. James J. Augsburger, Department of Ophthalmology, Suite 350, Health Professions Building, 3223 Eden Ave., Cincinnati OH 45267-0527, USA; fax (513) 558-3108, [email protected]
Surgery Service and tthe Ophthalmic Pathology Service, Department of Ophthalmology, University of Cincinnati College of Medicine, Cincinnati, Ohio, and §Cincinnati Eye Institute, Cincinnati, Ohio
372
Enucleation vs. preservation-Augsburger et al
Can J Ophthalmol 2004;39:372-9
Enucleation vs. preservation-Augsburger et al of our study was to determine the effect of secondary enucleation on melanoma-specific mortality in plaque-treated eyes of patients whose affected eye became blind subsequent to plaque radiotherapy. A secondary objective was to identify significant prognostic factors for secondary enucleation of eyes that become blind following plaque therapy and for melanoma-specific mortality in such patients. METHODS
We performed a retrospective analysis of patients in the senior author's (J.J.A.) computerized database of patients with primary choroidal or ciliochoroidal melanoma treated by plaque radiotherapy between 1980 and 1997 (n = 1074). We first identified all patients who had become completely blind (i.e., visual acuity of no light perception) in the tumourcontaining eye subsequent to plaque radiotherapy for the intraocular tumour. We excluded from this group all patients whose irradiated eye still had some vision (i.e., visual acuity of light perception or better) at the time of secondary enucleation, all patients in whom metastatic uveal melanoma developed before the eye became completely blind, and all patients whose blind eye had not been enucleated but who had been followed for less than 6 months after the eye went blind. We classified the patients who fulfilled our inclusion criteria into two subgroups: those treated by secondary enucleation, and those in whom the blind eye continued to be managed without enucleation. We evaluated numerous baseline demographic and tumour-related variables in this study. The senior author evaluated all these variables prospectively at the time of comprehensive standardized ophthalmologic examination, which immediately preceded radioactive plaque implantation. The methods used to evaluate these variables have been reported elsewhere. 2- 5 Plaque radiotherapy was performed using cobalt 60 plaques (1980 through 1987), iodine 125 plaques (1982 through 1997), iridium 192 plaques (1982 through 1988) or ruthenium 106 applicators (1984 through 1989). The methods used to implant and remove the plaques and to estimate radiation dose rates and dosage distributions in these cases have previously been reported. 6 The target dosage of radiation to the tumour apex was 80 Gy to 100 Gy for all patients in this group. The senior author performed the plaque surgery in all cases. Following plaque radiotherapy, patients were mon-
itored periodically to assess local tumour response to therapy and various other ocular variables, including visual acuity in the treated eye and presence or absence of specific treatment-related complications (radiation-induced cataract, radiation retinopathy, radiation optic papillopathy, iris neovascularization, neovascular glaucoma, vitreous hemorrhage and radiation-induced scleral necrosis). The recommended ophthalmologic follow-up schedule for all patients was every 3 months after plaque therapy for 3 years, every 6 months for the next 2 years, and yearly thereafter. The systemic status of plaque-treated patients (i.e., presence or absence of metastatic uveal melanoma and survival) was determined by communication with their family practitioners, internists and oncologists. We evaluated the survival status of patients who had been lost to ophthalmologic followup by searching the US National Death Index database.7 The cause of death for patients known to have died before the study closure date (March 2003) was determined by communication with the attending physician at the time of death, review of medical records from the terminal illness and, when available, review of official death certificates. We performed the following predetermined data analyses on the tabulated variables: 1) comparative analysis of the distributions of the baseline clinical variables in the two patient subgroups, using the unpaired two-tailed t test for comparison of the means of the intrinsically continuous numeric variables and Pearson's X2 test for comparison of the frequency distributions of the categoric variables; 2) analysis of potential prognostic factors for secondary enucleation using both univariate and multivariate Cox proportional hazards modelling; 3) computation and plotting of Kaplan-Meier cumulative actuarial survival curves based on deaths from metastatic uveal melanoma and deaths from any cause for the two patient subgroups; 4) comparison of the survival curves of the two patient subgroups using the Mantel-Haenszel X2 test; 5) analysis of potential prognostic factors for death from metastatic uveal melanoma and death from any cause using both univariate and multivariate Cox proportional hazards modelling; and 6) estimation of the adjusted effect of secondary enucleation on death from metastatic uveal melanoma by computing the rate ratio (and 95% confidence interval) for secondary enucleation when added to the best Cox proportional hazards regression model containing only significant predictor variables. Because of the exploratory nature
CAN JOPHTHALMOL-VOL.39, NO.4, 2004
J 73
Enucleation vs. preservation-Augsburger et al
of this study, the multiplicity of variables being evaluated and the expectation of a limited number of patients who would fulfil all the inclusion criteria, we selected the conservative a value of 0.01 for assigning statistical significance to observed differences between the subgroups in the various comparative analyses. We performed the multivariate Cox proportional hazards modelling using a stepwise, step-down method and p > 0.05 to reject variables. RESULTS
Our computerized database search identified 185 patients (17.2% of the 1074 patients treated by plaque radiotherapy between 1980 and 1997) whose melanoma-containing eye had become completely blind during available follow-up. We excluded from this group 99 patients whose eye became blind at the time of secondary enucleation even though it still retained visual acuity of light perception or better up to that instant, 3 patients in whom metastatic uveal melanoma had developed before the date the eye became blind, and 4 surviving patients whose blind eye had been preserved but who had been followed for less than 6 months after the eye became blind. We thus studied 79 patients, of whom 25 (31.6%) had undergone secondary enucleation of the blind eye and 54 (68.4%) retained the blind eye through all available follow-up. Tables 1 and 2 contain summary information for the
79 patients and their tumours at baseline examination before plaque therapy. The two groups were reasonably similar in terms of all of the baseline variables evaluated. The median length of follow-up (the interval between the date of plaque radiotherapy and the date of the most recent patient encounter) was 10.5 years for the secondary enucleation subgroup and 11.4 years for the ocular preservation subgroup. The mean interval between plaque radiotherapy and total loss of vision in the eye was 3.2 (standard deviation 2.3) years and 4.4 (standard deviation 2.6) years respectively, a nonsignificant difference (t = 1.86, p = 0.066). Fourteen eyes (56.0%) in the secondary enucleation subgroup became painful as well as blind during available follow-up, compared with 6 eyes (11.1%) in the ocular preservation subgroup, a highly significant difference (X 2 = 15.9, p < 0.001). This finding indicates that secondary enucleation was much more likely when the eye was painful than when it was not. During available follow-up, vitreous hemorrhage developed in the treated eye in 10 eyes (40.0%) in the secondary enucleation subgroup and 24 eyes (44.4%) in the ocular preservation subgroup. This difference was not statistically significant (X 2 = 0.036, p = 0.85). During follow-up, local relapse of the irradiated tumour was detected in four eyes (16.0%) in the secondary enucleation subgroup and in seven eyes (13.0%) in the ocular preservation subgroup, a nonsignificant difference (X 2 < 0.01, p = 0.99).
Table !-Baseline continuous numeric clinical variables in patients with primary choroidal or ciliochoroidal melanoma treated by plaque radiotherapy who underwent secondary enucleation or whose blind eye was managed without enucleation Group; mean (and standard deviation)
Variable Age,yr Anteroposterior diameter of tumour, mm Circumferential diameter of tumour, mm Largest basal diameter of tumour, mm Tumour thickness, mm Largest linear dimension of tumour, mm
J 74
CAN J OPHTHALMOL-VOL.39, NO.4, 2004
Secondary enucleation (n = 25)
Ocular preservation (n =54)
61.2 ( 12.1) 11.3 10.5
value
p value
62.5 (11.3)
0.46
0.64
(3.4)
12.7 (3.6)
1.64
0.11
(3.3)
11.9 (3.7)
1.63
0.11
11.6 (3.2) 8.2 (3.4)
13.0 (3.5) 7.9 (2.9)
1.71 0.38
0.09 0.71
11.7 (3.3)
13.1
1.66
0.10
(3.5)
t
Enucleation vs. preservation-Augsburger et al
Table 2-Baseline categoric clinical variables in the two subgroups Group; no. (and %) of patients Variable Sex Male Female Location of anterior tumour margin At or posterior to ocular equator Anterior to equator but at or posterior to ora serrata Anterior to ora serrata in ciliary body Location of posterior tumour margin relative to optic disc Extending to optic disc margin Extending to within 3 mm of optic disc margin Over 3 mm from optic disc margin Location of posterior tumour margin relative to foveola Extending to or beneath foveola Extending to within 3 mm of foveola Over 3 mm from foveola Visual acuity before treatment ~ 20/SO < 20/SO to ~ 20/200 < 20/200
Secondary enucleation
Ocular preservation
12 (48.0) 13 (S2.0)
30 (SS.6) 24 (44.4)
II (44.0)
16 (29.6)
s (20.0)
22 (40.7)
9 (36.0)
16 (29.6)
3 (12.0)
3
14 (2S.9)
16 (64.0)
37 (68.S)
(8.0)
I
O.IS
0.70
3.40
0.18
1.01
0.60
1.80
0.41
1.04
O.S8
(1.8)
7 (28.0) 16 (64.0)
IS (27.8) 38 (70.4)
2 (8.0) 8 (32.0) IS (60.0)
9 ( 16.7) IS (27.8) 30 (SS.6)
Univariate analysis of potential prognostic factors for secondary enucleation identified eye pain (p < 0.001) and the interval between plaque therapy and onset of blindness (p = 0.003, shorter interval unfavourable) as the only significant factors. Multivariate Cox proportional hazards regression analysis identified eye pain as the only independently significant prognostic factor associated with secondary enucleation following plaque radiotherapy for choroidal or ciliochoroidal melanoma (p < 0.001). The interval between plaque radiotherapy and onset of blindness (shorter interval unfavourable) and the largest linear tumour dimension (larger size unfavourable) were of borderline significance (0.10 ~p > 0.01). Surprisingly, neither local tumour relapse nor
p value
(S.6)
6 (24.0)
2
x2
vitreous hemorrhage was significantly associated with secondary enucleation by multivariate analysis. During available follow-up, 13 patients (52.0%) in the secondary enucleation subgroup and 22 patients (40.7%) in the ocular preservation subgroup died. The cause of death for the 35 patients is shown in Table 3. The Kaplan-Meier survival curves based on the end point "death from metastatic uveal melanoma" in the two study subgroups are presented in Fig. 1, and those based on the end point "death from any cause" are presented in Fig. 2. There were slight apparent differences between the corresponding curves, but these differences were not statistically significant (see figure legends for details).
CAN JOPHTHALMOL-VOL. 39, NO.4, 2004
J 75
Enucleation vs. preservation-Augsburger et al
ative to the optic disc (p = 0.050, closer to disc unfavourable) as prognostic factors of borderline statistical significance for this end point. Multivariate Cox proportional hazards regression analysis identified largest linear dimension of the treated tumour at baseline as the only independently significant prognostic factor for death from metastatic uveal melanoma (p < 0.001). The interval between plaque radiotherapy and onset of blindness was of borderline statistical significance (p = 0.082, shorter interval unfavourable) in a two-term regression that included largest linear tumour dimension. Surprisingly, local tumour relapse was not significantly associated with death from metastatic melanoma by multivariate analysis. Using a Cox model that included both largest linear tumour dimension and secondary enucleation, we computed a rate ratio of 0.90 (95% confidence interval 0.56-1.46) for secondary enucleation. The fact that the rate ratio is less than 1.0 indicates that the adjusted probability of death from metastatic uveal melanoma was slightly higher in the ocular preservation subgroup than in the secondary enucleation subgroup. However, because the confidence interval
Table 3-Cause of death for the patients who died during follow-up Group; no. (and %) of patients Cause of death
Secondary enucleation
Ocular preservation
5 (20.0) 2 (8.0) 6 (24.0)
14 (25.9) 2 (3.7) 6 (II. I)
Metastatic uveal melanoma Nonmelanoma cancer Other
Univariate analysis of potential prognostic factors for death from metastatic uveal melanoma identified all of the variables related to tumour size except tumour thickness as significant (p < O.Ol).lt also identified the interval between treatment and onset of blindness (p = 0.012, shorter interval unfavourable), age at the time of treatment (p = 0.019, older age unfavourable), visual acuity in the eye at the time of treatment (p = 0.023, worse visual acuity unfavourable) and location of the posterior tumour margin rei-
1.0 t-
t- ,.....,
0.9 r-- · - t--
ta
0.8
-
0.7
·s;> ~
::::s
(/)
0.6
0
0.5
>.t:::
.c ta .c
-
1......_, ;--,
.., f-1-lh
0.4 0.3
0~
0.2
Q.
0.1
-
1-
r-- -·
'--
5
-
-
10
f--
15
20
Time from plaque therapy, yr Secondary enucleation
- - Ocular preservation
Fig. 1-Kaplan-Meier cumulative actuarial survival curves for the 25 patients in the secondary enucleation subgroup and the 54 patients in the ocular preservation subgroup. The curves are based on deaths from metastatic uveal melanoma only. The difference between the curves is not statistically significant (Mantei-Haenszel X2 = 0.15, p = 0.70).
J 76
CAN J OPHTHALMOL-VOL. 39, NO.4, 2004
Enucleation vs. preservation-Augsburger et a1
1.0 0.9
co> ·s:'-
0.8
tn
0.6
0
0.5
::::J
>
:!::::
:0
"'0
.c '-
c.
-c--r
i---4.
M 1-
0.7
0.4 0.3
-
w
L
L.
..__,
"1..,
-
h
-
..
~-
~~
r-- D
-
-r-
.,
-!--
....
-
1--
0.2 0.1
!--
I
-
!--
'
I I
I I
I
I 5
10
15
20
Time from plaque therapy, yr Secondary enucleation
- - Ocular preservation
Fig. 2-Kaplan-Meier cumulative actuarial survival curves for the two subgroups. The curves are based on deaths from any cause. The difference between the curves is not statistically significant (Mantei-Haenszel X2 = 0.68, p = 0.41 ).
broadly straddles the value 1.0, we regard the adjusted effect of secondary enucleation on melanoma-specific mortality to be not statistically significant. INTERPRETATION
The principal result of our study seems to be that secondary enucleation of eyes that become totally blind following plaque radiotherapy for choroidal or ciliochoroidal melanoma does not substantially reduce (or increase) the patient's risk of dying from metastatic uveal melanoma. This result should be reassuring to patients and surgeons faced with the decision to enucleate or to retain such eyes. It suggests that both enucleation and preservation of such eyes are acceptable management options because they are likely to yield a similar survival prognosis. Because of the counterintuitive nature of this result and several recognized limitations of our study, however, we urge readers to regard this result as inconclusive. Our finding needs to be confirmed (or refuted) in independent groups of patients before its validity can be accepted. The principal limitation of our work is the small
size of our study group. We had only 79 eyes in the total study group, only 25 eyes that had been enucleated secondarily, and only 19 deaths from metastatic uveal melanoma during available follow-up. The smaller the number of patients and achieved end points in a study, the greater the likelihood that the subgroups will differ substantially from the whole population of patients meeting the inclusion criteria of the study by chance alone. The fact that our data analysis did not identify local tumour relapse as an unfavourable prognostic factor for survival - even though two prior reports, based on the larger study group from which these patients were selected, demonstrated a substantial effect of local tumour relapse 8•9 - emphasizes this point. There is no way in which we can correct for this limitation by data analysis. We therefore acknowledge that the power of our study to identify a substantial favourable effect of secondary enucleation on survival is extremely low. A second limitation of our work is the restrictiveness of our inclusion and exclusion criteria. By requiring eyes to be completely blind (i.e., visual acuity of no light perception), we may have inadvertently eliminated the patients with the most aggressive uveal
CAN JOPHTHALMOL-VOL 39, NO.4, 2004
377
Enucleation vs. preservation-Augsburger et al
melanomas. In our study, the mean interval between plaque radiotherapy and onset of total blindness was 3.2 years in the secondary enucleation subgroup and 4.4 years in the ocular preservation subgroup. Patients in this study had to survive long enough for their irradiated eyes to go blind. Patients whose irradiated eyes might have gone blind but who experienced metastasis within the first 2 to 3 years after plaque treatment could not have appeared in our study group. The results also might have been substantially different if we had defined blindness as visual acuity of 5/200 or worse, counting fingers or worse, or light perception. Similarly, changes in our exclusion criteria could have influenced the observed results. A third limitation is the limited potential of survival modelling methods in general and of Cox proportional hazards modelling specifically to identify prognostic factors for specified patient outcomes. There is no inherent reason why clinical data should conform to any specific survival model. If we had used a parametric modelling method 10 or perhaps Cox modelling with time-dependent variables, 11 we might have obtained a better fit with the observed data. A fourth limitation is the nonrandomized assignment of patients to the two subgroups. It is entirely possible that unknown, unevaluated factors associated with the decision to enucleate or to preserve certain eyes influenced the survival outcome in indeterminate ways. A fifth limitation is the exploratory nature of our study. To our knowledge, no other group has published or presented results of a study designed to estimate the effect of secondary enucleation on death from metastatic melanoma in patients who were treated initially by plaque radiotherapy. We acknowledge the need for other investigators to study similar but independent groups of patients and provide either supportive or contradictory results. Finally, a sixth limitation is our probable failure to evaluate some pertinent clinical variables and the inability to assess pertinent pathological variables. It is certainly possible that we failed to evaluate one or more important clinical variables; however, we selected virtually all clinical variables that have been found by our group 2- 5 and other investigators 12•13 to be of predictive value for the outcomes evaluated in this study. A more global limitation that applies to all plaque-treated uveal melanoma groups is the inability to assess and therefore evaluate histopathological variables that have been associated with death from metastatic uveal melanoma. 14- 18 Assessing such vari-
3 78
CAN JOPHTHALMOL-VOL 39, NO.4, 2004
abies, of course, requires enucleating the eyes or, alternatively, excising the tumours intact. In conclusion, this exploratory retrospective study appears to show that secondary enucleation of eyes that become totally blind following plaque radiotherapy for choroidal or ciliochoroidal melanoma does not substantially decrease (or increase) the patient's risk of dying from metastatic uveal melanoma. For the reasons mentioned above, we regard these results as suggestive rather than definitive. We encourage other investigators to study this matter and report their independent results and conclusions in the near future. This work was supported by Research to Prevent Blindness, Inc., New York, and the Medical Education and Research Fund and Quest for Vision Fund, Department of Ophthalmology, University of Cincinnati College of Medicine, Cincinnati, Ohio. REFERENCES
1. Shields CL, Shields JA, Karlsson U, Markoe AM, Brady LW. Reasons for enucleation after plaque radiotherapy for posterior uveal melanoma: clinical findings. Ophthalmology 1989;96:919-24. 2. Augsburger JJ, Gamel JW, Sardi VF, Greenberg RA, Shields JA, Brady LW. Enucleation vs cobalt plaque radiotherapy for malignant melanomas of the choroid and ciliary body. Arch Ophthalmol1986;104:655-61. 3. Augsburger JJ, Gamel JW, Lauritzen K, Brady LW. Cobalt-60 plaque radiotherapy vs enucleation for posterior uveal melanoma. Am J Ophthalmol1990;109:585-92. 4. Augsburger JJ, Correa ZM, Freire J, Brady LW. Longterm survival in choroidal and ciliary body melanoma after enucleation versus plaque radiation therapy. Ophthalmology 1998;105:1670-8. 5. Augsburger JJ, SchneiderS, Freire J, Brady LW. Survival following enucleation versus plaque radiotherapy in statistically matched subgroups of patients with choroidal melanomas: results in patients treated between 1980 and 1987. Graefes Arch Clin Exp Ophthalmol 1999;237: 558-Q7. 6. Giblin ME, Shields JA, Augsburger JJ, Brady LW. Episcleral plaque radiotherapy for uveal melanoma. Aust N Z J Ophthalmol1989;17:153-Q. 7. National Center for Health Statistics, Centers for Disease Control and Prevention, US Department of Health and Human Services. National Death Index. Available: http://www.cdc.gov/nchs/r&dlndi/ndi.htm (accessed 2004 Apr 15). 8. Karlsson UL, Augsburger JJ, Shields JA, Markoe AM, Brady LW, Woodleigh R. Recurrence of posterior uveal melanoma after 6°Co episcleral plaque therapy. Ophthalmology 1989;96:382-8. 9. Vrabec TR, Augsburger JJ, Gamel JW, Brady LW,
Enucleation vs. preservation-Augsburger et al
10.
11.
12.
13.
14.
Hernandez C, Woodleigh R. Impact of local tumor relapse on patient survival after cobalt 60 plaque radiotherapy. Ophthalmology 1991;98:984-8. Gamel JW, McLean IW, Rosenberg SH. Proportion cured and mean log survival time as functions of tumour size. Stat Med 1990;9:999-1006. Fisher LD, LinDY. Time-dependent covariates in the Cox proportionru hazards regression model. Annu Rev Public Health 1999;20:145-57. Wilson MW, Hungerford JL. Comparison of episcleral plaque and proton beam radiation therapy for the treatment of choroidal melanoma. Ophthalmology 1999;106: 1579-87. Lommatzsch PK, Werschnik C, Schuster E. Long-term follow-up of Ru-1 06/Rh-106 brachytherapy for posterior uveal melanoma. Graefes Arch Clin Exp Ophthalmol 2000;238:129-37. Seddon JM, Albert DM, Lavin PT, Robinson N. A prognostic factor study of disease-free interval and survival following enucleation for uveal melanoma. Arch Ophthal-
moll983;101:1894-9. 15. Seddon JM, Polivogianis L, Hsieh CC, Albert DM, Gamel JW, Gragoudas ES. Death from uveal melanoma: number of epithelioid cells and inverse SD of nucleolar area as prognostic factors. Arch Ophthalmoll987;105:801-6. 16. Folberg R, Rummelt V, Parys-Van Ginderdeuren R, Hwang T, Woolson RF, Pe'er J, et al. The prognostic value of tumor blood vessel morphology in primary uveal melanoma. Ophthalmology 1993;100:1389-98. 17. Seregard S, Kock E. Prognostic indicators following enucleation for posterior uveal melanoma. A multivariate analysis of long-term survival with minimized loss to follow-up. Acta Ophthalmol Scand 1995;73:340-4. 18. Mlikitie T, Summanen P, Tarkkanen A, Kivela T. Microvascular density in predicting survival of patients with choroidal and ciliary body melanoma. Invest Ophthalmol Vis Sci 1999;40:2471-80. Key words: choroidal melanoma, ciliochoroidal melanoma, plaque radiotherapy, enucleation, survival
CAN J OPHTHALMOL-VOL. 39, NO.4, 2004
J 79