Collaborative Ocular Melanoma Study (COMS) Medium Tumor Trial (MTT) patient disposition with respect to enrollment in the COMS Quality of Life Study (QOLS).
Two-state Markov model estimates of proportions of patients with a Hospital Anxiety and Depression Scale diagnosis of “definite anxiety” or “definite or possible anxiety.”
Observed vs fitted mean scores from generalized linear model for selected primary Quality of Life Study outcome scales. Asterisk indicates P<.05 for comparison of enucleation vs brachytherapy at this point. ADVS indicates Activities of Daily Vision Scale; CI, confidence interval; NEI-VFQ, National Eye Institute Visual Function Questionnaire.
Observed vs fitted mean scores from generalized linear model for selected primary Quality of Life Study outcome scales. ADVS indicates Activities of Daily Vision Scale; CI, confidence interval; NEI-VFQ, National Eye Institute Visual Function Questionnaire.
Observed vs fitted mean scores from generalized linear model for selected secondary outcome scales. Asterisk indicates P<.05 for comparison of enucleation vs brachytherapy at this point. NEI-VFQ indicates National Eye Institute Visual Function Questionnaire; CI, confidence interval.
Collaborative Ocular Melanoma Study–Quality of Life Study Group. Quality of Life After Iodine 125 Brachytherapy vs Enucleation for Choroidal Melanoma5-Year Results From the Collaborative Ocular Melanoma Study: COMS QOLS Report No. 3. Arch Ophthalmol. 2006;124(2):226-238. doi:10.1001/archopht.124.2.226
LESLIEHYMANPhDAuthors: Members of the writing team who signed authorship responsibility, financial disclosure, and copyright transfer statements for the group are listed below.
Collaborative Ocular Melanoma Study–Quality of Life Study Group Authors: Michele Melia ScM Claudia S.Moy PhD; and Sandra M. Reynolds MA, Wilmer Ophthalmological Institute, Johns Hopkins University, Baltimore, Md; James A. Hayman MD, Department of Radiation Oncology, University of Michigan, Ann Arbor; Timothy G. Murray MD, Bascom Palmer Eye Institute, University of Miami, Miami, Fla; Kenneth R. Hovland MD, Porter Adventist Hospital, Denver, Colo; John D. Earle MD, Department of Radiation Oncology, Mayo Clinic, Jacksonville, Fla; Natalie Kurinij PhD, National Eye Institute, Bethesda, Md; Li Ming Dong PhD, and Päivi H. Miskala PhD, Wilmer Ophthalmological Institute; Connie Fountain COT, Department of Ophthalmology, University of Iowa Hospitals and Clinic, Iowa City; David Cella PhD, Center on Outcomes Research and Education, Evanston Northwestern Healthcare, Evanston, Ill; and Carol M. Mangione MD, MSPH, Department of Medicine, David Geffen School of Medicine at the University of California, Los Angeles.
Dr Moy is currently with the National Institute of Neurological Disorders and Stroke, Bethesda.
To describe health- and vision-targeted quality of life following treatment with iodine 125 brachytherapy vs enucleation for choroidal melanoma in a subgroup of patients who were treated and observed prospectively as part of a large randomized clinical trial.
Main Outcome Measures
Difficulty with driving, near vision activities, and activities using stereopsis or binocularity; anxiety; and depression.
Two hundred nine patients who enrolled in the Collaborative Ocular Melanoma Study trial for medium-sized tumors between March 1995 and July 1998 and gave informed consent prior to randomization to participation in an ancillary study of quality of life.
Patients were interviewed by telephone by a trained interviewer from the Collaborative Ocular Melanoma Study Coordinating Center at baseline (prior to randomization), at 6 months, and on annual anniversaries of enrollment. The questionnaire battery included the Medical Outcomes Study Short Form 36, the Activities of Daily Vision Scale, the National Eye Institute Visual Function Questionnaire, and the Hospital Anxiety and Depression Scale. Additional questions concerning satisfaction with posttreatment appearance and concerns about cancer recurrence also were included in posttreatment interviews.
There was a significant increase in both treatment groups in levels of reported difficulty for most vision-oriented activities, and in bodily and ocular pain, 6 months following treatment. Differences in visual function between treatment groups reported during follow-up were relatively small, but significant differences favoring brachytherapy-treated patients were observed for driving during the first year of follow-up and for peripheral vision during the first 2 years of follow-up. Anxiety levels in both groups decreased significantly following treatment, but patients treated with brachytherapy with symptoms of anxiety were less likely to report later resolution of symptoms than patients with symptoms of anxiety who were treated with enucleation. This study was unable to assess impact of treatment on satisfaction with appearance and concern about cancer recurrence during the first year after treatment, but no treatment-related differences were found on these measures at 2 years and later follow-up times.
Patients treated with brachytherapy reported significantly better visual function than patients treated with enucleation with respect to driving and peripheral vision for up to 2 years following treatment. Differences between treatments in visual function diminished by 3 to 5 years posttreatment, paralleling decline in visual acuity in brachytherapy-treated eyes. Patients treated with brachytherapy were more likely to have symptoms of anxiety during follow-up than patients treated with enucleation.
Application to Clinical Practice
Given that no significant differences in survival between enucleation and brachytherapy have been found, the differences demonstrated here for driving and anxiety will allow the individual patient and physician to make informed choices regarding treatment based on personal preferences.
The Collaborative Ocular Melanoma Study (COMS) trial for patients with “medium-sized” choroidal melanoma was a randomized clinical trial sponsored by the National Eye Institute (NEI) and designed to test whether treatment with iodine 125 brachytherapy vs enucleation offers patients the better chance of future survival and cancer-free survival. The first patient was enrolled in 1987, and accrual to the trial was completed in 1998. The COMS Medium Tumor Trial (MTT) was not designed originally to evaluate impact of choroidal melanoma and its treatment on quality of life. However, it was recognized that the 2 treatments being compared in the COMS were likely to have differing impact on patients with respect to daily function, particularly for vision-related tasks, self-image, and mental health. In 1994, the NEI awarded funding to conduct an ancillary study of quality of life in COMS MTT patients, the COMS Quality of Life Study (QOLS), in recognition of the importance of quality-of-life outcomes in ophthalmologic studies in general and in patients with medium-sized tumors in particular. The impact of treatment on quality of life in choroidal melanoma is of particular interest and relevance given that no clear survival difference between treatments was demonstrated in the COMS1 and because of potential disparities in costs associated with alternative treatments. The purpose of this report is to describe quality-of-life results during the first 5 years of follow-up in patients who enrolled in the ancillary quality-of-life study and were interviewed prior to randomization in the MTT.
The COMS design has been described in detail elsewhere.1,2 Briefly, patients participating in the trial for medium-sized tumors were randomized with equal probability to treatment with iodine 125 brachytherapy or enucleation. All patients had unilateral choroidal melanoma at least 2.5 mm in apical height, no more than 10.0 mm in apical height (no more than 8.0 mm in height when the tumor was near the optic disc), and no more than 16.0 mm in basal diameter. Patients with visual acuity 20/200 or worse in the fellow eye or with expected survival of less than 5 years due to concomitant health conditions were not eligible for the study. Other eligibility and exclusion criteria were described in earlier publications.1,3 Visual acuity was measured by COMS-trained and certified examiners at baseline, every 6 months up to 5 years, and annually thereafter according to a standard protocol using best refractive correction and a Bailey-Lovie chart.3
Starting in February 1995, patients with choroidal melanoma newly enrolling in the COMS at participating centers were offered the opportunity to participate in the QOLS.4 Thirty of 40 COMS clinical centers participated in this phase of the QOLS. Institutional review board approval was obtained from all participating clinical centers, and written consent was obtained from all participating patients. The remaining centers elected not to participate (n = 5) or were no longer enrolling patients in the MTT (n = 5). Enrollment in the QOLS continued through July 1998, when enrollment in the MTT closed. A total of 209 (79%) of 265 patients who enrolled in the COMS at QOLS participating centers elected to participate in the QOLS (Figure 1).
The QOLS interview was administered by telephone from the coordinating center at baseline (prior to randomization), at 6 months of follow-up, and at annual anniversaries of enrollment. The QOLS questionnaires included the Medical Outcomes Study Short Form 36 (SF-36), the Activities of Daily Vision Scale (ADVS), the NEI Visual Function Questionnaire (NEI-VFQ), and the Hospital Anxiety and Depression Scale (HADS).5- 10 The NEI-VFQ was not available at the time the COMS-QOLS was initiated and was added to the interview in April 1997. Questions addressing concerns about cancer recurrence and satisfaction with appearance were developed for the QOLS and added to the follow-up interviews in April 1997 and July 1998, respectively.11 Hence, data for these subscales were not available for the majority of patients at baseline. Interviewers were unaware of the patients' treatment assignment; to maintain their masked status, patients were instructed at the beginning of the interview to try not to say anything that would reveal what treatment they had received. For 90% of interviews, the interviewer did not know the patient's treatment at the end of the interview; for the remaining interviews, the treatment assignment was revealed at some point during the interview.
This phase of the QOLS had a target sample size of 200 patients. With this sample size, it was projected that there would be 80% power to detect 5- to 10-point differences on most of the SF-36 and all of the ADVS scales with type I error of 5%, assuming 30% mortality over 5 years and 5% rate of missed interviews. An intertemporal correlation between subscale scores of any pair of interviews completed by the same patient of 0.60 also was assumed.
Analysis and reporting of study results is based on interviews completed and data received at the COMS Coordinating Center by June 30, 2003. As of that date, 178 patients (92% of patients still alive) had completed the 3-year follow-up interview, 168 (92%) patients had completed the 4-year follow-up interview, and 146 (91%) had completed the 5-year follow-up interview. Fifty patients had died. Time windows for all interviews up to and including 4 years had closed; 2 patients had yet to complete the 5-year interview and were within the time window for completing the 5-year interview.
The Wilcoxon rank sum test for continuous characteristics and the χ2 or Fisher exact test for discrete characteristics were used to compare patients who enrolled in the QOLS with those who were eligible but did not enroll and those who were ineligible. The χ2 goodness-of-fit test was used to determine whether the QOLS patients were representative of all patients enrolled in the MTT.
Estimated prevalences of concomitant health conditions at baseline and over follow-up were obtained using life-table methods to adjust for differing lengths of follow-up among study patients. Prevalences were compared between treatment groups using the log-rank test.12 The proportion of patients who stopped driving because of vision was compared between treatment groups using the Kaplan-Meier method13 with the log-rank test.
Change from baseline to 6 months and difference between treatment groups were estimated using a generalized linear model with serial correlation and measurement error that included baseline as 1 of the points.14 Time was included as a categorical variable in the model to avoid making any assumptions regarding the course over time of scale scores. Exploratory data analysis was used to guide the selection of a serial correlation structure that corresponded to observed data; for QOLS data, an autoregressive moving average correlation structure corresponded well to observed data. Primary analyses included as covariates only follow-up time, treatment, and interaction between follow-up time and treatment. When the overall time × treatment interaction term was statistically significant at P<.05, a t test was used to compare scores between treatments at each follow-up time with P<.05 being considered evidence of a significant treatment effect. Fitted pretreatment scores from the model were compared with 6-month posttreatment scores also using the t test. Adjustment for baseline QOL score was not performed in the primary analysis because baseline data were missing on a high proportion of patients for several of the primary outcomes. All analyses followed the intention-to-treat principle with patients included in the treatment group to which they were assigned by randomization, regardless of treatment actually received.
Secondary analyses included models with adjustment for other covariates, including age, sex, education, baseline QOL score, new cancers diagnosed during follow-up, and SF-36 physical component summary (PCS) score (as a surrogate for overall health status), and continuous time random effects models to determine whether there were significant trends over time within each treatment group. Distributions of change in scores from baseline also were compared between treatments using the Wilcoxon rank sum test. Results of the primary analysis are the focus of this report; however, it is noted whenever a secondary analysis gave qualitatively or significantly different results.
Anxiety and depression scores from HADS also were analyzed using a nonhomogeneous Markov model for discrete time intervals, with the diagnostic categories (no diagnosis, possible diagnosis, definite diagnosis) established by the HADS developers.10 This model estimates proportions of patients falling into each diagnostic category at each follow-up time and probabilities of transition to another category at the subsequent follow-up. Transition probabilities were compared between treatment groups using the log-rank test.15 Logistic regression with generalized estimating equations16 was used to estimate odds ratios for treatment of having a “diagnosis” vs no diagnosis based on the HADS diagnostic categories with adjustment for baseline diagnostic category and other baseline covariates.
Early in the QOLS, and prior to knowledge of any outcome data, a set of subscales was selected that would be considered the primary outcomes for the study based on hypothesized treatment differences or importance to patients in their everyday lives (Table 1). All other subscales were considered as secondary outcome variables. Statistically significant treatment differences in primary outcome variables were considered definitive evidence of important treatment effects while differences in secondary outcomes were considered of lesser importance and suggestive only of possible treatment effects. This strategy was adopted because of the multiplicity of subscales and loss of control of type I error if all were analyzed and considered equally important. It was judged that combining subscales into overall scores might mask important treatment differences because these were expected only in a few subscales. Hence, global scores or combining scales was not considered a valid analysis option.
Demographic, ocular, and tumor characteristics of study patients by treatment group are described in Table 2. Patient ages at time of QOLS enrollment ranged from 29 to 87 years (median, 62 years). Fifty-three percent of the patients were men, and nearly all (99%) were non-Hispanic white people. Sixty-three percent of patients had good visual acuity (better than 20/40) in both eyes at baseline, 31% had good visual acuity in the fellow eye and impaired visual acuity (less than 20/40) in the study eye, and 2% had good visual acuity in the study eye and impaired visual acuity (20/40 to 20/160) in the fellow eye. The remaining 4% of patients had impaired visual acuity in both eyes at baseline. There was an imbalance in study eye visual acuity at baseline between the treatment groups; patients randomized to brachytherapy had better visual acuity in the study eye (median, 20/25) as compared with patients randomized to enucleation (median, 20/32). The treatment groups were reasonably well-balanced at baseline with respect to other characteristics, including concomitant health conditions (not shown).
Three of the outcome scales, ADVS night driving, NEI-VFQ peripheral vision, and SF-36 mental health, showed imbalance in mean baseline score by treatment of 5 points or greater (mean differences, 5.2, 7.6, and 5.0, respectively; Table 3). The difference in night driving score favored brachytherapy while the other 2 differences favored enucleation. Baseline median scores were identical by treatment for night driving and peripheral vision but still favored enucleation for mental health. In addition, although well-balanced with respect to mean and median scores, the proportion of patients scoring in the “definite or possible anxiety” range on the HADS was higher at baseline for brachytherapy (32%) than for enucleation (27%). All other outcome scales were reasonably well-balanced by treatment; however, baseline scores were not available for 63% of patients for the NEI-VFQ scales because this questionnaire was added after the study began.
As of the cutoff date for this report, all but 2 patients were past the time window for the 5-year follow-up clinic visit and QOLS interview, and 95% of interviews required by protocol had been completed for both enucleation and brachytherapy patients. Vital status was known for all patients, and date of death was known for all deceased patients.
Baseline demographic, ocular, and tumor characteristics of patients enrolled in the QOLS were compared with characteristics of patients who were eligible but did not enroll and patients who were ineligible for the QOLS. Patients who were eligible but did not enroll were those who enrolled in the MTT but did not enroll in the QOLS for various reasons. Ineligible patients enrolled in the MTT at a non-QOLS participating center or enrolled in the MTT prior to the initiation of the QOLS. Enrolled patients had smaller tumors and better vision than ineligible patients (median tumor apical height, 4.0 vs 4.4 mm, and study eye visual acuity, 20/25 vs 20/32, in enrolled vs ineligible patients, respectively). This mainly was due to a change in COMS eligibility criteria that expanded the eligible range for tumor apical height, allowing enrollment of both smaller and larger tumors in the later years of the study when the QOLS also was enrolling patients. Enrolled patients were comparable with eligible patients who did not enroll on all baseline characteristics. Participants in QOLS were representative of all patients enrolled in the MTT based on comparison of baseline characteristics using the χ2 goodness-of-fit test, with the exception of a higher proportion of patients with tumor apical height in the 8.1 to 10.0 mm range and corresponding deficiency in the 6.1 to 8.0 mm range, a reflection of the change in tumor size eligibility criteria. There were no differences in tumor size between MTT and QOLS patients when the comparison is restricted to MTT patients who were enrolled after tumor size criteria were changed. Representativeness of MTT patients vs all patients with medium-sized choroidal melanoma who were screened at COMS centers was discussed in a previous COMS publication.17
Both treatment groups reported significantly increased difficulty with driving, near activities, and activities using stereopsis or binocularity and decreased levels of anxiety at the 6-month follow-up interview as compared with baseline (Table 4). The largest changes were reported for the driving scales (mean change, −6 to −13 points for enucleation and −5 to −9 points for brachytherapy, depending on the specific scale), followed by activities requiring stereopsis or binocularity (mean change, −12 for enucleation and −7 for brachytherapy). Smaller changes were seen for ADVS and NEI-VFQ 25 near activities (mean change, −4.8 and −6.5 respectively for enucleation, and −4.2 and 1.3 for brachytherapy; the latter change in the brachytherapy group was not statistically significant). There was no change in level or proportion of patients with depression (data not shown). Immediate impact of treatment on satisfaction with appearance and concerns about cancer recurrence could not be assessed because these questions were not asked at baseline.
Results of linear mixed model regression and Markov modeling of primary outcomes are reported in Table 4 and Figures 2, 3, and 4. For primary visual function outcomes, mean ± SE statistically significant (P<.05) treatment differences at specific follow-up times ranging from 9.3 ± 3.8 points to 12.8 ± 3.9 points that favored brachytherapy were seen for the ADVS night driving and NEI-VFQ driving scales. No significant treatment-related differences were seen for near vision activities or activities using stereopsis or binocularity. Ninety-one percent of patients were driving at baseline. Patients who stopped driving during follow-up for vision-related reasons were assigned a score of 0 on questions concerning driving and were included in the analyses. For the driving scales, the treatment difference diminished over follow-up time because of a decrease in brachytherapy patient scores, such that treatment differences were not significant for later follow-up times. There was no difference between treatments in the proportion of patients who stopped driving because of vision throughout the 5 years of follow-up (2% vs 0% at 6 months, 3% vs 2% at 1 year, and 7% vs 9% at 5 years for brachytherapy vs enucleation, respectively). A higher proportion of patients in both groups stopped driving at night (as compared with completely stopping driving), but again there were no significant differences by treatment (percent stopping driving at night, 7% vs 9% at 6 months, 9% vs 13% at 1 year, and 29% vs 20% at 5 years for brachytherapy vs enucleation, respectively [log-rank test, P = .31]).
The Markov model analysis of HADS anxiety scores over time by diagnostic category (“no anxiety” vs “possible or definite anxiety”) showed that brachytherapy patients with symptoms of anxiety were less likely to report resolution of symptoms than were enucleation patients (log-rank test, P<.001, Figure 2), although mean scores did not significantly differ between the 2 groups. Proportion of patients with “definite anxiety” also did not differ by treatment group, although the number of such patients was small (log-rank test, P = .29). Depression scores on HADS did not differ by treatment. Average scores for concerns about cancer recurrence and appearance scales were similar for the 2 treatments at 2 years and later follow-up times; there were insufficient data to assess impact of treatment on these scales prior to 2 years of follow-up.
As with primary visual function outcomes, both treatment groups reported increases in difficulty on secondary visual function outcomes from baseline to 6 months, including ADVS overall vision, distance activities (both ADVS and NEI-VFQ), ADVS glare problems, NEI-VFQ color vision, and NEI-VFQ peripheral vision; however, only increased difficulty on the ADVS overall vision and distance activities scales was statistically significant in both groups with mean decreases in the 6- to 7-point range (Table 5). Patients treated with enucleation experienced a large and highly significant decrease in peripheral vision score (mean, −29.2 points) while patients treated with brachytherapy had a much smaller decrease that did not differ significantly from 0 (mean, −4.0 points).
Both treatment groups reported a significant worsening from baseline to 6 months of ocular pain (−9.9 points for enucleation and −5.4 points for brachytherapy) and bodily pain (−4.2 points for enucleation and −5.9 points for brachytherapy). Patients treated with enucleation reported a significant 6-month decline in nearly all other NEI-VFQ vision-dependent general and mental health scales (general health, role difficulties, dependency, and social function) and in all of the SF-36 physical health measures (physical function, physical role function, and general health perceptions) with changes ranging from 4 to 14 points. Patients treated with brachytherapy reported smaller 6-month decreases or no change on these measures, and their estimated changes were not significantly different from 0. Both groups reported a small decline in the PCS score.
Patients treated with brachytherapy reported a significant improvement of 10 to 14 points 6 months following treatment in the SF-36 mental health and emotional role function scales, while patients treated with enucleation reported smaller improvements of 3 to 4 points; however, the difference between treatment groups was not statistically significant. The same trend was repeated for the SF-36 mental component summary score; patients treated with brachytherapy reported a statistically significant improvement (5 points), while patients treated with enucleation reported a smaller improvement (2 points).
As expected, significant treatment differences were not seen for most of the secondary outcomes. These included overall vision (both ADVS and NEI-VFQ), distance vision, glare problems, color vision, ocular and bodily pain, all of the vision-dependent general and mental health scales with the exception of vision-dependent social function, and all SF-36 scales. Not surprisingly, patients treated with brachytherapy reported significantly better function on the 1-item VFQ peripheral vision scale than did patients treated with enucleation at all follow-up times; the differences up to 2 years of follow-up were statistically significant (Figure 5). Patients treated with brachytherapy also reported significantly better vision-dependent social function as compared with patients treated with enucleation in the first year after treatment.
All of the physical health measures and the PCS score from the SF-36 declined slightly over time in both treatment groups (data not shown). This decline was consistent with increasing age in this group of patients, whose median age was 62 years at the time of study enrollment. There was little or no change over follow-up in SF-36 mental health-related measures (vitality, social function, emotional role function, mental health, and mental health component summary) (data not shown). In a comparison with age- and sex-specific norms from the National Survey on Functional Health Status,7 the QOLS patients, who were selected for the MTT to be healthy aside from their choroidal melanoma diagnosis, generally had higher than expected SF-36 scores for all but the general health and vitality scales, where they scored lower than expected.
Analyses adjusted for baseline score and baseline covariates (age, sex, education, and PCS score as a surrogate for general health status) gave similar results to the unadjusted analyses. The few exceptions were the following:
Differences on the driving scales between the 2 treatments were smaller after adjustment (3 to 8 points), with statistically significant benefits to brachytherapy confined to the first year after treatment. Differences on the day driving scale, although still favoring brachytherapy in early follow-up, were not statistically significant and significantly favored enucleation by 5 years of follow-up (mean difference, 6 points).
Differences on the SF-36 role functioning and mental health scales after adjustment supported a benefit of brachytherapy at 6 months for all 3 scales. Estimated differences adjusted for baseline covariates were 9 to 10 points for the role functioning scales and 5 points for the mental health scale.
There was a moderate although not statistically significant difference in the number of deaths as of the 5-year follow-up visit in the 2 treatment groups, with 25 deaths in the brachytherapy group and 17 deaths in the enucleation group. Estimated mortality in the brachytherapy group as of the 5-year visit was 24% vs 16% in the enucleation group (log-rank P = .14). Thus, there was some concern that observed treatment differences could be due to differences in mortality. However, the treatment-related differences in visual function were observed mainly in early follow-up when the number and rate of deaths in the 2 treatment arms were very similar (the Kaplan-Meier estimate of mortality at 2 years in the brachytherapy group was 2.9% vs 2.8% in the enucleation group). Hence, the observed treatment differences in early follow-up are highly unlikely to be due to differential mortality.
Thirteen patients treated with brachytherapy required enucleation at some time during follow-up; the Kaplan-Meier estimate of enucleation rate was 12.4% at 5 years. These patients had a clear trend for lower scores on all of the physical and mental health measures after enucleation as compared with other patients, but the number of such patients was too small to draw any definitive conclusions regarding this group.
There also was concern that higher anxiety levels in patients treated with brachytherapy could be due to factors other than treatment; hence, a number of analyses targeted factors associated with higher anxiety levels. In particular, an imbalance for this outcome at baseline favored the enucleation group. In addition, the brachytherapy group developed more new cancers and more new health conditions during follow-up than the enucleation group, and it was thought that this might contribute to the observed difference in anxiety symptoms.
The logistic regression model with generalized estimating equations that adjusted for baseline covariates, including baseline anxiety diagnostic group, showed higher odds of “definite or possible anxiety” and “definite anxiety” in patients treated with brachytherapy at all follow-up times, although this effect was not statistically significant. The odds ratio for brachytherapy vs enucleation of “definite or possible anxiety” pooling across follow-up times was 2.0 (95% confidence interval, 0.8-4.9) and of “definite anxiety” was 2.5 (95% confidence interval, 1.0-6.5). The odds ratio for “definite or possible” anxiety for patients treated with brachytherapy who were enucleated during follow-up at visits following the enucleation was 2.3 (95% confidence interval, 0.4-14.9) as compared with patients treated with enucleation; for patients treated with brachytherapy who were not enucleated, the corresponding odds ratio was 2.0 (95% confidence interval, 0.8-4.8). Thus, increased anxiety among patients treated with brachytherapy who require enucleation during follow-up does not appear to account for overall higher proportions of patients with anxiety in the brachytherapy group as compared with the enucleation group.
New cancer diagnosis did not have a significant impact on anxiety scores (most new cancers were basal cell carcinoma of the skin), and adjusting for new cancer diagnosis had no impact on the treatment difference. The SF-36 PCS score was used as a surrogate for overall physical health, and low PCS was associated with higher anxiety levels. However, adjusting for PCS had no impact on treatment-related differences, which still favored enucleation. Timing of the COMS clinic visit with respect to the QOLS interview also was related to anxiety score. Patients treated with brachytherapy who completed the QOLS interview before the COMS clinic visit had significantly higher anxiety scores than patients treated with brachytherapy who completed the interview after the clinic visit while timing of the interview was not associated with scores from patients treated with enucleation. The release of survival results for the MTT also correlated with anxiety scores. Anxiety scores indicated lower anxiety after release of survival information, particularly in patients treated with brachytherapy. Prior to release of survival information, patients treated with brachytherapy had greater average anxiety scores than patients treated with enucleation regardless of follow-up time (mean ± SE differences over follow-up, 1.8 ± 2.5 to 7.6 ± 3.8) with statistically significant differences at 4 and 5 years of follow-up. After release of survival information, there was no difference between treatments (mean ± SE difference, −2.4 ± 2.7 to 2.5 ± 3.3). These results support that the observed difference in anxiety scores was related to treatment rather than a chance finding, but we cannot rule out that this difference was due to a baseline imbalance with patients who were prone to anxiety being more likely to be assigned to brachytherapy.
The COMS now has results comparing brachytherapy with enucleation with respect to length of life, visual and physical function, mental health, and perceptions about health and appearance. Results of the MTT indicate that the 2 treatments are similar with respect to length of life. The QOLS demonstrated some early posttreatment functional advantages to brachytherapy with respect to driving and peripheral vision and possibly short-term advantages with respect to mental health. Certain patients treated with brachytherapy, particularly those with pre-existing symptoms of anxiety, may suffer from increased risk of symptoms of anxiety as compared with patients treated with enucleation during follow-up. The MTT had a high degree of external validity,1,17 and the QOLS patients were a representative subgroup of MTT patients; this supports that the QOLS results also are externally valid and applicable to all patients who meet the MTT eligibility criteria.
Regardless of whether patients received brachytherapy or enucleation, reported levels of difficulty for many vision-related activities (driving, near and distance activities, and activities using stereopsis or binocularity) and pain increased significantly immediately following treatment for choroidal melanoma. Patients treated with enucleation generally reported larger changes, although only driving and peripheral vision differed significantly by treatment. Although treatment differences were confined to the first 2 years of follow-up (1 year for driving), this length of time could be important, particularly for the older patient. Neither treatment group reported significant improvement in any aspect of visual function during the 5 years of follow-up, with the exception of ocular pain, which improved over time in patients treated with enucleation. Following the initial decrease in visual function, patients treated with enucleation had stable visual function for all visual function measures except ocular pain, while patients treated with brachytherapy reported a slow decline on nearly all visual function measures (data not shown). This finding is consistent with increasing loss of visual acuity in the brachytherapy-treated eye over follow-up and possible need for enucleation of the eye as already reported for the COMS.18,19 The QOLS results add to the growing body of literature supporting that although overall quality of life is more highly determined by vision in the better-seeing eye, vision in the worse-seeing eye significantly impacts visual function for many eye conditions.20- 25 Levels of impairment seen in COMS patients were generally lower than those reported for many eye diseases, such as cataract, age-related macular degeneration, and diabetic retinopathy.8,9,26,27 In the sense that most patients did not report high levels of impairment, they adapted well to both treatments.
As compared with changes seen from precataract to postcataract surgery,21 changes 6 months after surgery for choroidal melanoma among COMS patients were approximately one half the magnitude (for driving) or smaller (for other visual function scales) and opposite in direction (worsening function rather than improvement). Treatment-related visual function differences at 1 to 2 years in the COMS were similar or slightly larger in magnitude to differences reported for groups of people with vs without monocular vision impairment from observational studies.28,29 Another nonrandomized, retrospective study of quality of life in patients with choroidal melanoma reported little overall difference between enucleation and radiotherapy30; however, this study consisted largely of patients at later posttreatment times than the current study (mean follow-up time, 4.9 years for radiotherapy and 6.3 years for enucleation). Our study also found few differences between treatments during later follow-up.
Patients treated with brachytherapy by chance had better vision in the study eye at baseline than did patients treated with enucleation; this was reflected in better baseline scores on a number of visual function scales, including the driving scales. Baseline adjustment lessened the estimated difference between treatments for these scales; for the ADVS day driving scale, the treatment difference was no longer statistically significant after baseline adjustment. Hence, we cannot rule out that the advantage to brachytherapy with respect to driving applies only to driving at night.
Patients treated with brachytherapy in our study had higher posttreatment rates of symptoms of anxiety than did patients treated by enucleation. A high rate of psychological distress, including anxiety, following eye-conserving radiotherapy for choroidal melanoma has been reported by others31- 33; however, these investigators did not compare radiotherapy with enucleation. One prospective study of quality of life in choroidal melanoma reported no difference in levels of anxiety as measured by the HADS in patients treated with enucleation as compared with patients treated with radiotherapy34; however, this study was not randomized and had an imbalance toward greater anxiety in patients treated with enucleation prior to treatment. Although it has not been studied to our knowledge, it is plausible that patient anxiety level influences the choice of treatment.
The QOLS investigators had hypothesized a priori that patients treated with brachytherapy would have more anxiety over follow-up because these patients still had tumors that required careful monitoring for signs of growth, and patients may feel that presence of the tumors combined with late-occurring adverse effects from radiotherapy are indicators of an increased risk of metastasis. The finding of higher anxiety levels in patients treated with brachytherapy prior to a clinic visit as compared with after a clinic visit supports this hypothesis. We also found that treatment-related differences in anxiety were greater prior to the release of the MTT survival results than afterwards, when no treatment-related difference in overall survival was demonstrated. There was a slight imbalance between treatment groups with respect to age and educational level, with patients treated with enucleation being slightly older and better educated, but analyses adjusting for age and educational level gave similar estimated treatment effects. Studies in other cancers have demonstrated that occurrence of treatment adverse effects and need for additional treatment correlate with increased concern about cancer recurrence,35,36 and the QOLS has demonstrated that patients treated with brachytherapy whose eyes are enucleated during follow-up have increased concern about cancer recurrence.11 These findings suggest anxiety level of current patients treated with brachytherapy may be reduced by providing them with the information that their treatment has been shown to provide chances of survival equal to enucleation, as long as the tumor is monitored for signs of growth, and most common adverse effects of treatment have not been shown to be related to survival.
Patients treated with brachytherapy had significantly less difficulty than patients treated with enucleation with respect to driving and peripheral vision during the first 1 to 2 years following treatment and there was a trend for less difficulty with other aspects of visual function, although these differences were small. Our group of patients treated with brachytherapy had evidence of increased symptoms of anxiety over follow-up as compared with patients treated with enucleation, but this difference did not exist once survival data from the MTT were released, suggesting that it may have been due to uncertainty regarding the relative survival benefits of each treatment. Unfortunately, we had insufficient data to assess whether satisfaction with appearance or concerns about cancer recurrence differed between treatments during the early posttreatment period; however, there was no evidence of a treatment difference after 2 years following treatment.
Although by 4 to 5 years posttreatment the observed treatment-related differences in the primary outcomes had largely disappeared, the early differences favoring brachytherapy may be important considerations for patients faced with choosing a therapeutic approach. Given that no significant differences in survival between enucleation and brachytherapy have been found, these results will allow the individual patient and physician to make more informed choices regarding treatment based on personal preferences with respect to visual function, anxiety and concern about cancer recurrence, and appearance.
Correspondence: Michele Melia, ScM, Jaeb Center for Health Research, 15310 Amberly Dr, Suite 350, Tampa, FL 33647 (email@example.com).
Submitted for Publication: January 26, 2004; accepted May 3, 2005.
Financial Disclosure: None.
Funding/Support: This study was supported by the National Eye Institute through cooperative agreement EY 10207 with the National Institutes of Health, Bethesda, Md.
Group Information: A complete list of the Collaborative Ocular Melanoma Study–Quality of Life Study Group was published in Arch Ophthalmol. 2003;121:1010-1020.
Acknowledgment: The COMS-QOLS Advisory Committee thanks Drs Scott Zeger and Donald Patrick for guidance and advice on the analysis and interpretation of data for this study.