Health-Related Quality of Life in Adolescent and Young Adult Retinoblastoma Survivors

Abstract

Background: Retinoblastoma (RB) is a malignant intraocular tumor diagnosed in early childhood that requires extensive medical and surgical treatment at a young age. Health-related quality of life (HRQOL) is thought to be diminished due to visual impairment, facial deformities, and fear of recurrence or secondary cancer. However, few studies have identified variables associated with HRQOL among those with RB. Purpose: To compare HRQOL of adolescents and young adults (AYAs) with RB to matched controls and to identify predictors of HRQOL in RB survivors. Methods: Using a cross-sectional design, 198 AYAs (101 RBs and 97 controls) completed HRQOL (PROMIS®-29 profile) and psychosocial questionnaires (Rosenberg self-esteem scale, multidimensional scale of perceived social support, and Hollingshead index for socioeconomic status). Clinical variables (age at diagnosis, visual acuity, laterality, heredity, treatment regime, and anesthesia exposure) were extracted from the medical record. Correlates of HRQOL were estimated using linear regression models. Results: RB survivors reported similar HRQOL compared to controls. Physical function (p < .001), social support (p = .013), and self-esteem (p = .028) were lower in the RB group compared to controls. Visual acuity and self-esteem accounted for 52% of the variance in PROMIS physical health summary scores and self-esteem accounted for 38% of the variance in mental health summary scores. Conclusion: Despite deficits in physical function and self-esteem HRQOL in RB survivors was comparable to healthy counterparts. However, the majority of RB survivors in this study had normal visual acuity. Clinicians should explore ways to enhance self-esteem in RB survivors.

Introduction

Retinoblastoma (RB) is a malignant central nervous system tumor that emerges from the developing retina and is caused by a hereditary or spontaneous mutation of the RB1 gene (Dimaras & Corson, 2019). RB is the most common intraocular childhood cancer, accounting for ∼11% of cancers diagnosed in the first year of life with 95% of cases identified before the age of five (Wong et al., 2014). Between 300 and 350 children are diagnosed with RB each year in the United States (U.S.; Gombos, 2012), with a worldwide incidence of one in every 16,000 live births (Dimaras & Corson, 2019). The overall five-year survival rate has steadily improved to 97% in the U.S. (Fernandes et al., 2017). This improved survival is attributed to advances in diagnosis and treatment including a shift toward chemotherapy rather than radiation (Tamboli et al., 2015; Fernandes et al., 2017).

The primary goal of RB treatment is the mitigation of cancer, while ocular salvage and vision preservation are secondary (Delhiwala et al., 2016). Options for treatment depend on tumor size and location, laterality, the threat of metastatic disease, and visual prognosis (Shields & Shields, 2004). Treatment for RB differs from other childhood cancers and includes a combination of systemic and/or intraarterial or intraocular chemotherapy, focal modalities such as laser photocoagulation and cryotherapy, external beam radiation, and/or enucleation (surgical removal of the eye; Gombos, 2012). Repeated eye examinations under general anesthesia to assure motionlessness are required to provide treatment and monitor for recurrence of disease. Unfortunately, a child with RB can be exposed to as many as 50 general anesthetics before the age of four (Wilson et al., 2006) with a long-term impact on health-related quality of life (HRQOL) unknown. RB survivors also require ophthalmology surveillance throughout the lifespan to monitor for reduction or loss of vision in the affected eye. If chemotherapy or radiation is used in treatment, survivors will also be followed by an oncologist due to the increased risk of secondary cancers.

RB survivors must live with their chronic disease and treatment effects into adulthood. Survivors often suffer from visual impairment, facial deformities, and the continual fear of a recurrence or secondary cancer (van Dijk et al., 2010). Adult RB survivors have an increased risk of chronic medical conditions including poor visual acuity, malignant neoplasms, cataracts, severe hearing loss, and thyroid nodules (Friedman et al., 2016). Outcomes in RB are dependent on age at diagnosis, disease stage at presentation, family history, laterality, and treatment regime (Aziz et al., 2012). While studies have identified long-term medical, visual, and psychosocial outcomes (Ford et al., 2015; Friedman et al., 2016; Narang et al., 2012), little is known about HRQOL in RB survivors in the U.S. Studies examining HRQOL in other childhood cancer survivors (CCS) rarely include survivors of RB, and the unique factors related to RB (diagnosis in early childhood, heritability, individualized treatments, anesthesia exposure, and visual outcomes) limit transferability of findings.

HRQOL is a multidimensional construct that includes functioning and well-being in physical, mental, and social health (Ruccione et al., 2013). Measuring HRQOL in the clinical setting can improve patient–provider communication and patient satisfaction, uncover unknown morbidities, facilitate clinical decision-making, and improve patient outcomes (Cantrell & Kelly, 2015). A recent literature review on HRQOL in RB survivors found few U.S.-based studies with mixed results reflecting both positive and negative outcomes (Belson, et al., 2020). Some studies reported worse HRQOL in RB survivors than controls or population norms while others found no difference or worse/better outcomes in individual domains (e.g., physical function). A variety of disease-specific and generic HRQOL measures were used, cohort age ranges were wide and included an aging era of treatment, studies predominantly from Europe/Asia countries, and few contained control groups limiting the generalizability of findings to the current U.S. RB population. Of interest, one study measured vision-related quality of life (QOL) revealing significant differences based on visual status or acuity and laterality of disease. These findings support the need for further investigation to identify potentially modifiable factors associated with HRQOL in RB survivors to aid in the development of future interventions. The purpose of this study is to describe HRQOL in a U.S. cohort of adolescent and young adult (AYA) RB survivors compared to matched controls and to identify factors associated with HRQOL in the RB group. Our central hypothesis is HRQOL reported by AYA RB survivors will be worse than matched healthy counterparts. Furthermore, younger age at diagnosis, decreased visual acuity, and hereditary status will be associated with worse HRQOL in RB survivors.

Methods

Study Design

The study used a comparative, cross-sectional design. Age-, gender-, and race/ethnicity-matched AYA with RB and healthy controls participated in the study between June 2019 and December 2020.

Sample and Setting

A convenience sample of AYA RB survivors was recruited from the Children's Hospital Los Angeles (CHLA) Vision Center. This center serves a racial and ethnically diverse population with various levels of medical complexity. Participants were included if they were between the ages of 14 to 26 years, had a diagnosis of RB, and were able to speak and read English. RB participants were excluded if they had severe developmental delay precluding active study participation and self-report or current diagnosis of a second malignant neoplasm. When COVID-19 occurred, procedures were modified. The principal investigator (PI) instead mailed recruitment letters and made phone calls to potential RB participants in lieu of on-site recruitment.

Healthy controls were recruited from posted community flyers, word of mouth referrals, and contact with controls in other clinical studies in Southern California. Controls were matched on age ±2 years, gender, and race/ethnicity, with an RB participant and self-reported no chronic medical (e.g., diabetes, hypertension, and asthma) or psychiatric (e.g., depression, anxiety, and bipolar) conditions. If eligible in either RB or control group, participation occurred on the same day, or a future appointment was made to participate in the study.

Procedures

This study was approved by the Institutional Review Boards at CHLA (IRB No. 19-00014) and the University of California Los Angeles (IRB No. 19-000305). At the time of study participation, written parental permission and child assent was obtained for participants under age 18, and informed consent was obtained from participants aged 18 and over. A study information sheet and consent form were read aloud to blind participants in the RB cohort by the PI and/or emailed to them to use with voiceover software if needed. A witnessed signature was obtained for adult blind RB participants and a waiver of signature for blind RB minors. Participants completed all measures utilizing an iPad while in a clinic or at another designated location with data collated via Research Electronic Data Capture (REDCap; Harris, 2009, 2019). When the COVID-19 lockdown occurred on March 13, 2020, procedures were modified, and a questionnaire link was emailed to participants. Participants received compensation and parking validation (for those at CHLA). The total time to complete all study measures was ∼10 to 30 min.

Data Collection

Variables

Clinical and demographic data were obtained by participant completion of a demographic form and/or extracted by medical record review. Demographic data included age, gender, race/ethnicity, and socioeconomic status (SES). The role of age as a predictor of HRQOL has been commonly measured in studies examining HRQOL with varying results. Adolescent RB survivors reported worse HRQOL than younger children in one previous study (van Dijk et al., 2007a). However, three other studies found no association between age and HRQOL in RB survivors (Batra et al., 2015, 2016; van Dijk et al., 2007b). Gender was explored as a predictor of HRQOL in four studies of RB survivors but was not found to be statistically significant (Batra et al., 2015, 2016; van Dijk et al., 2007a, 2007b). Race or ethnicity has not been previously examined in HRQOL in RB survivors (Batra et al., 2015, 2016; van Dijk et al., 2007a, 2007b), possibly due to the homogeneity of the study populations. SES and its relation to HRQOL have not been specifically examined in RB survivors. However, some components of SES (e.g., income, education, and occupation) have been included in previous European studies with no significant associations related to HRQOL (van Dijk et al., 2007a,  2007b).

For RB participants, clinical variables extracted from the medical record included age at diagnosis, visual acuity, laterality, heredity, treatment regime, and anesthesia exposure. Lower age at diagnosis was associated with better overall HRQOL in one study (Batra et al., 2015). However, age at diagnosis did not predict HRQOL in the other two studies (van Dijk et al., 2007a, 2007b). Visual acuity in child and adolescent RB survivors was negatively associated with self-perception (van Dijk et al., 2007a). Furthermore, survivors with normal visual acuity in their nonaffected eye reported better physical HRQOL than those that were visually impaired (van Dijk et al., 2007a). Corrected visual acuity was assessed by the ophthalmology team in the participants’ better eye using the Snellen scale (e.g., 20/20) and transformed to a logarithm of the minimal angle of resolution value by the PI (http://www.myvisiontest.com). Results were categorized according to the World Health Organization International Classification of Diseases (ICD-10) guidelines as normal vision (>0.3 or 20/40), visual impairment (0.05–0.3), and blindness (<0.05 or 20/400; www.who.int). Adult survivors with a hereditary risk for RB reported significantly worse HRQOL than nonhereditary survivors (van Dijk et al., 2007b). However, heredity and laterality did not predict HRQOL (van Dijk et al., 2007b).

The treatment regime consisted of enucleation, chemotherapy, a combination of enucleation/chemotherapy, or a combination of enucleation/chemotherapy/radiation. Previous studies examined treatment type as a predictor for HRQOL in the RB survivor but found no statistical significance (Batra et al., 2015, 2016; van Dijk et al., 2007a, 2007b). In one previous study, children who received chemotherapy reported significantly worse physical HRQOL than those who did not receive chemotherapy (Weintraub et al., 2011).

While no studies have assessed the relationship between anesthesia exposure and HRQOL, the risk of general anesthesia exposure, procedural fear/anxiety, and neurodevelopmental outcomes is well documented in the pediatric population (Flick et al., 2011; Ing et al., 2012; Stargatt et al., 2006; Wilder et al., 2009). Due to their frequent exposure to anesthesia at a young age and the potential for this exposure to cause long-term effects, this variable was included (e.g., number of exposures and total duration) to assess HRQOL.

Measures

Socioeconomic Status

SES was measured using the Hollingshead index (Hollingshead, 1975). The four factors of the Hollingshead index include parental marital status, employment status, educational attainment, and occupational prestige. Education (scale 1–7) and occupation (scale 1–9) scores are assigned to each parent/guardian and then weighted (education × 3, occupation × 5) to obtain a single score (range 8–66). SES was divided into five categories including lower status (8–19), lower-middle status (20–29), middle status (30–39), upper-middle status (40–54), and upper status (55–66) based on the social strata categories established (Hollingshead, 1975).

HRQOL Measure

The generic HRQOL measure used was the patient-reported outcomes measurement information system PROMIS®-29 profile v2.1 (Craig et al., 2014). The PROMIS® is a 29-item instrument that includes domains in physical (physical function, fatigue, sleep disturbance, pain interference, and pain intensity) mental (anxiety and depression), and social (ability to participate in social roles and activities) health. The PROMIS®-29 profile consists of two summary scores (mental and physical health), which are calculated from weighted correlations of the domains (Hays, et al., 2017). Raw scores are converted to T-scores with a mean of 50 and standard deviation (SD) of 10, which is referenced to the U.S. general population average (Liu et al., 2010). Higher T-scores on the mental and physical summary scores indicate better HRQOL. For the functional domains, greater T-scores represent increased or better function while for the symptom domains greater T-scores represent greater symptom burden (e.g., greater anxiety or depression). Internal consistency reliability (coefficient alpha) estimates for the PROMIS®-29 domains range from .77 to .94, and the physical and mental health summary scores at .98 and .97, respectively (Hays, et al., 2017). Multiple studies provide evidence of the validity of the PROMIS® measures (Cella et al., 2010; Cook et al., 2016; Rothrock et al., 2010), but they have not been previously used in the RB population. However, previous studies in the general pediatric oncology population support the validity of the PROMIS-29 (Hinds et al., 2013, 2019).

Internal consistency reliability estimates (Cronbach's alpha) for the PROMIS®-29 domains in this study were as follows: .69 physical function, .86 anxiety, .92 depression, .88 fatigue, .80 sleep, .83 social, and .93 pain interference. Estimated reliabilities of the physical and mental health summary scores were .77 and .96, respectively.

General Health Measure

The first question on the PROMIS® Global Health v1.2 scale was used to measure general perceived health (Hays, et al., 2017). The question is as follows: “In general, would you say your health is?” with answer choices including “excellent,” “very good,” “good,” “fair,” and “poor.” One's perception of health can influence their subjective appraisal of health; thus, general health perception is often a single-item measure used in many health statuses and HRQOL tools. This single item has a product–moment correlation with the PROMIS® four-item global physical health scale of .81, and thus can be used alone to estimate the global physical health scores (Hays, et al., 2015). However, the marginal reliability was only .52 for the single item compared to .81 for the PROMIS® Global health scale (Hays, et al., 2015).

General Overall QOL Measure

The second question on the PROMIS® Global Health v1.2 scale was used to measure overall QOL (Hays et al., 2017). The question is as follows: “In general, would you say your quality of life is?” with answer choices ranging the same as the previous question. This single question allows for the consideration of the components of QOL that are important to that individual and does not combine the ratings of physical, mental, and social domains together to one total score which can influence overall QOL assessment. Overall summary scores do not allow evaluation of whether the individual is more affected by their physical, mental, or social health. Overall QOL is a single item that measures subjective well-being and is one of four items in the PROMIS® global mental health scale (Hays, et al., 2017). Palimaru and Hays (2017) examined associations of the overall QOL item with other PROMIS® global health items to assess the overlap between overall QOL and HRQOL. All correlations were statistically significant (p < .001) and ranged from r = .36 (pain) to r = .82 (physical health; Palimaru & Hays, 2017).

Rosenberg Self-Esteem Scale

The Rosenberg Self-Esteem Scale (RSES) is a 10-item instrument and is the most widely used self-esteem scale (Blascovich & Tomaka, 1991; Rosenberg, 1989). The RSES uses an ordinal four-point Likert-type response scale ranging from strongly agree to strongly disagree. Total scores are recorded on a continuous scale with higher scores indicating higher self-esteem. Scores range from 0 to 30 with <15 suggesting low self-esteem. The mean internal reliability (Cronbach alpha) of the RSES was found to range from .81 to .91 (Schmitt & Allik, 2005; Sinclair et al., 2010) with test–retest correlations from .82 to .88 (Blascovich & Tomaka, 1991). The RSES has not been previously used in the RB population; however, it has been previously used in other CCS (Langeveld et al., 2002; Tonsing & Ow, 2018). Cronbach's alpha for the RSES in this study was .90.

Multidimensional Scale of Perceived Social Support

The Multidimensional Scale of Perceived Social Support (MSPSS) is a 12-item questionnaire measuring an individual's perception of their social support (Zimet et al., 1988). Respondents indicate how they feel about each statement (e.g., “There is a special person with whom I can share my joys and sorrows” and “My family really tries to help me”). The MSPSS has three subscales for social support (friends, family, and significant other) and uses a seven-point Likert-type scale ranging from very strongly disagree (1) to very strongly agree (7). The four items for each subscale are summed and divided by four. A total score is created by dividing the sum of all items by 12. The average scores range from one to seven with higher scores representing greater perceived support. Cronbach's alpha reliability was found to be high (.85 to .91) with test–retest values (.72 to .85) indicating good stability (Zimet et al., 1990). The MSPSS has not been used in the RB population but demonstrated good internal consistency in AYA cancer survivors with a Cronbach's alpha of .93 and the subscales of family, friends, and significant others were .91, .89, and .94, respectively (Tremolada et al., 2016). Cronbach's alpha for the MSPSS in this study was .93 for the total score and .94, .90, and .94 for the three subscales (family, friends, and significant other), respectively.

Data Analysis

An a priori power analysis was performed using G*Power 3.1 (Faul et al., 2007) to determine the sample size for this study. A sample size of 95 in the RB and control groups were determined using a two-tailed t-test with a significance level of .05 and 80% power to detect an effect size of .41. This effect size was based on the results of a previous study of HRQOL where children with RB perceived their QOL related to school (PedsQL) as significantly lower than an age-matched normative sample (t = 2.23, p < .01, Cohen's d = .41; Weintraub et al., 2011).

Sample characteristics are presented for RB and control participants as means with SD and/or medians with interquartile ranges for continuous variables. Variables were examined for normality and outliers. Outliers were examined for accuracy and representation of the target population and were not removed. The majority of continuous data had nonnormal distributions (per Shapiro–Wilk's test of normality). Differences in HRQOL between subgroups of RB survivors (gender, laterality, heredity, race/ethnicity, SES, and treatment regime) and comparison between groups (RB and controls) were examined using nonparametric statistics consisting of Mann–Whitney U or Kruskal–Wallis test for all continuous variables and chi-squared or analysis of variance for all categorical variables. Independent sample t-tests were used for a few dependent variables that were normally distributed. Transformation (both log and square root) of the physical health summary score was attempted without normality. Physical health summary scores did not have a symmetric distribution and were negatively skewed. Depending on the data distribution of the outcome variable, Pearson product–moment or Spearman rank-order correlation coefficients between all predictors (age, gender, race/ethnicity, SES, age at diagnosis, laterality, heredity, treatment regime, and anesthesia exposure) and the two outcome variables (physical and mental health summary scores) were examined. Variables with p values <.15 for the correlations were included in a stepwise multiple regression. The software identified the sequence of entry of covariates into the statistical models using the forward method. Although there was a nonnormality of the physical health summary scores, linear regression was still performed rather than less robust regression methods. Normality of the residuals was assessed and while Kolmogorov–Smirnov and Shapiro–Wilk have p < .001, plots of standardized residuals suggest that departure from normality may be only minor and should not substantially affect the interpretation of results. Assumptions of multicollinearity and homoscedasticity (equal variance of the residuals) were met. All analyses were conducted by using the Statistical Package for the Social Sciences (SPSS) Version 26.0 for MAC (IBM, Somers, NY, USA).

Results

A total of 184 RB survivors were identified from a clinic database, 61 were unable to be contacted via telephone or letter, and 123 were screened for eligibility. Of the 123 RB survivors, 7 were screened ineligible, 13 declined to participate, and 2 did not complete the study questionnaires for a total of 101 RB survivors enrolled (Figure 1). Furthermore, a total of 97 healthy controls matched on age ±2 years, gender, and race/ethnicity completed the questionnaires.

Figure 1.

Flow diagram of the recruitment process for retinoblastoma (RB) survivors.

Sample Characteristics

Demographic characteristics of the RB and control group are summarized in Table 1. There were no statistically significant differences in age, gender, race/ethnicity, and SES between groups. There were significant differences in insurance coverage between the groups with more RB survivors receiving state assistance and more controls with private coverage. RB is a California Children's Services eligible condition which provides assistance to families for treatment costs.

Table 1.

Comparison of Characteristics Between RB Survivors and Healthy Controls.

Characteristic	RB survivors (n = 101)		Healthy controls (n = 97)		p
Age, years (Mdn, IQR)	17 (15, 19)		17 (16, 19)		.194a
	n	%	n	%
Gender
Male	50	49.5	47	48.5	.995b
Female	51	50.5	50	51.5
Race/ethnicity
Latino/Hispanic	66	65.3	62	63.9	.995b
White	16	15.8	16	16.5
Black/African American	4	4	4	4.1
Asian/Pacific Islander	10	9.9	11	11.3
Other	5	5	4	4.1
SES					.085b
Lower status	14	14	14	14.6
Lower-middle status	22	22	17	17.7
Middle status	27	27	17	17.7
Upper-middle status	25	25	22	22.9
Upper status	12	12	26	27.1
Unknown	1	0	1	0
Insurance	(n = 99)		(n = 96)		<.001b
Private (PPO/HMO)	30	30.3	48	50
Public (MediCal/CCS)	62	62.6	29	30.2
No insurance (self-pay)	5	5.1	3	3.1
Unsure	2	2	16	16.7

Note. RB = retinoblastoma; Mdn = median; IQR = interquartile range; SES = socioeconomic status; PPO = preferred provider organization; HMO = Health Maintenance Organization; CCS = California Children's Services.

^a Mann–Whitney test. ^bChi-square.

The clinical characteristics of the RB group are summarized in Table 2. The median age at RB diagnosis was 15 months, 43 (43%) had bilateral disease and 13 (13%) had a previous family history of RB. Out of those who received genetic testing (n = 86), 40 (47%) had a germline pathogenic variant in the RB1 gene (8 unilateral and 32 bilateral). For treatment, 40 (40%) had a combination of chemotherapy and enucleation, 89 (88%) had at least one eye enucleation, with the majority 87 (86%) having normal vision. The median number of general anesthetics per survivor was 14 (range 3–63 anesthetics), while the median duration of exposure to general anesthesia was 693 min (range 226–3603 min). Lastly, 19 (19%) had another chronic medical condition (e.g., asthma, hearing loss, and thyroid problems), 5 (5%) had a psychological condition (e.g., anxiety, depression, or attention-deficit/hyperactivity disorder [ADHD]), and only 4 (4%) received treatment for a second malignant neoplasm or previous metastasis.

Table 2.

Clinical Characteristics of RB Survivors (n = 101).

Characteristics	n	%
Age at diagnosis, months (Mdn, IQR)	15	8, 23
Visual acuity
Normal vision	87	86
Visual impairment	8	7.9
Blindness	6	5.9
Laterality
Unilateral	58	57.4
Bilateral	43	42.6
Family history of RB
No	88	87.1
RB 1 mutation (n = 86)
Yes	40	46.5
Chemotherapy
Yes	64	63.4
Radiation therapy
No	86	85.1
Enucleation
Yes	89	88.1
Enucleation laterality (n = 89)
Unilateral	86	96.6
Bilateral	3	3.4
Treatment regime
Enucleation only	35	34.7
Chemotherapy only	9	8.9
Enucleation/chemotherapy	40	39.6
Enucleation/chemotherapy/radiation	14	13.9
Other	3	3.0
Anesthesia exposure (n = 95; Mdn/IQR)
No. of anesthetics per patient	14	8, 29
Total inhalational anesthesia (minutes)	693	402, 1,405
Chronic medical condition	19	19
Chronic psychiatric condition	5	5
Second malignant neoplasm/metastasis	4	4

Note. Mdn = median; IQR = interquartile range; RB = retinoblastoma.

HRQOL, Self-Esteem, and Social Support Between Groups

Mean PROMIS-29 physical function scores differed significantly between RB and controls (53.9 vs. 56.3, p < .001, d = −.24, respectively). RB survivors reported more depressive symptoms than controls (49.7 vs. 47.4, p = .051, d = .23). However, the other PROMIS®-29 domains and the physical and mental health summary scores showed no significant differences between RB and controls (Table 3).

Table 3.

Comparison of HRQOL Between RB Survivors and Healthy Controls.

Variables	RB (n = 101)		Controls (n = 97)		p
HRQOL (PROMIS®-29 Profile)	M	SD	Mdn	IQR
Symptomsa
Anxiety	53.3 53.8	8.7 48.2, 59.5	52.4 52.9	8.1 47.9, 57.5	.502c
Depression	49.7 49.1	8.9 42, 56.5	47.4 41	8.0 41, 52.4	.051c
Fatigue	44.5 45.3	9.2 33.7, 51	44.7 45.9	8.3 39.7, 51	.803c
Sleep disturbance	47.8 47.7	8.7 41.2, 53.6	46.7 46.4	6.8 41.8, 51.7	.311d
Pain interference	45.4 41.6	6.9 41.6, 50	44.7 41.6	5.6 41.6, 49.5	.665c
Functionb
Physical function	53.9 57	5.8 56.7, 57	56.3 57	2.5 57, 57	<.001c
Ability to participate in social roles and activities	58.0 58.5	7.3 52.9, 64.2	58. 6 63.5	6.5 53.5, 64.2	.731c
Pain intensity	n	%	n	%
0	62	61.4	56	57.7	.162c
1–3	25	24.8	32	33
4–6	10	9.9	9	9.3
7–10	4	4	0	0
Physical health summary score	54.81 57.5	5.9 54.9, 58.6	57.02 57.8	3.0 57.1, 58.6	.099c
Mental health summary score	54.5 55.3	7.4 50.3, 59.6	55.27 56.0	6.0 51.9, 60	.424d
PROMIS® Global Health
Question 1: General health rating	53.2 54	7.0 47, 62	55.1 54	5.5 54, 62	.05c
Question 2: Overall quality of life	50.6 51	7.9 44, 61	53.3 51	7.2 51, 61	.011c
Social support (n = 195)
MSPSS total	5.5	1.2	5.9	1.0	.013c
MSPSS-significant other	5.5	1.4	6.0	1.2	.003c
MSPSS-family	5.6	1.3	5.8	1.2	.224c
MSPSS-friends	5.3	1.5	5.8	1.2	.018c
Self-esteem
RSES	20.6	5.8	22.3	4.6	.028d

Note. SD = standard deviation; IQR = interquartile range; HRQOL = health-related quality of life; QOL = quality of life; PROMIS = patient-reported outcomes measurement information system; MSPSS = Multidimensional Scale of Perceived Social Support; RSES = Rosenberg Self-Esteem Scale.

^a Higher scores = greater symptom burden. ^bLower scores = worse functioning for physical function. ^cMann–Whitney test. ^dt-test.

On the single-item measures of general health and overall QOL, RB survivors reported worse mean general health (53.2 vs. 55.1, p = .05, d = −.30) and overall QOL than controls (50.56 vs. 53.28, p = .011, d = −.35). Statistically significant lower social support was found in RB survivors compared to controls on all domains of the MSPSS except the family domain (significant others 5.5 vs. 6.0, p = .003, d = −.38; friends 5.3 vs. 5.8, p = .018, d = −.33; and total score 5.5 vs. 5.9, p = .013, d = −.34). RB survivors reported statistically significantly worse self-esteem than controls (20.63 vs. 22.27, p = .028, d = −.31).

HRQOL in the RB Group

Mental health summary scores were lower in females than males (53.0 vs. 55.6, p = .042). Females also reported more depression (51.9 vs. 47.6, p = .001) and anxiety 56.0 versus 50.5, p = .001) than males. Of note, these differences were not seen within the control group except for anxiety which was also higher in healthy females than males (54.8 vs. 49.8, p = .002). Furthermore, within females only (RB vs. controls) no statistically significant differences in anxiety, depression, or mental health summary scores were found.

Physical health summary scores were statistically significant between visual acuity groups (normal, impaired, and blind) with the visually impaired reporting worse physical health (56.3 vs. 44.1 vs. 47.0, p < .001; Table 4). While unilateral survivors reported significantly worse physical function domains than bilateral survivors, the physical health summary scores were not significantly different. None of the other clinical variables (age at diagnosis [≤18 months/>18 months], laterality, heredity, chemotherapy, radiation, and enucleation) were significantly associated with HRQOL. There were no differences in HRQOL in adolescents (14–17 years) versus young adults (18–26 years).

Table 4.

Factors Affecting HRQOL in RB Survivors.

Variable	PROMIS® PH summary		PROMIS® MH summary
	M	SD	M	SD
Age at assessment
14–17 (n = 62)	55.36	5.39	54.71	8.35
18–26 (n = 39)	53.93	6.58	54.17	5.6
Gender
Male	55.56	5.02	55.6	5.89*
Female	54.07	6.6	53.02	8.4
Race/ethnicity
White/non-Hispanic	55.50	6.86	53.32	9.61
Latino/Hispanic	54.73	5.76	54.24	7.33
Black/African American	53.39	6.15	62.34	4.73
Asian/Pacific Islander	53.64	6.92	54.68	5.46
Other	56.47	3.13	54.73	2.06
SES
Lower status	55.12	6.48	55.82	7.59
Lower-middle status	54.32	6.07	52.27	8.05
Middle status	54.63	6.17	52.91	8.42
Upper-middle status	55.25	5.51	55.51	9.19
Upper status	54.58	6.12	57.67	4.95
Age at diagnosis
≤18 months (n = 64)	54.44	5.0	54.40	6.8
>18 months (n = 37)	55.45	5.74	54.67	8.4
Visual acuity
Normal vision	56.31	4.43**	54.68	7.21
Visual impairment	44.13	3.53	52.57	6.71
Blindness	47.04	6.92	54.33	11.7
Laterality
Unilateral	56.28	3.28	54.48	6.62
Bilateral	52.41	7.6	54.52	8.39
Heredity
No	54.73	5.85	54.47	7.43
Yes	55.35	6.37	54.65	7.34
Chemotherapy
No	56.59	3.08	54.62	6.37
Yes	53.78	6.83	54.43	7.96
Radiation
No	55.43	5.39	54.17	7.54
Yes	51.27	7.48	56.35	6.35
Enucleation
No	57.01	2.98	54.73	4.85
Yes	54.51	6.13	54.46	7.68
Chronic medical condition
No	54.93	5.6	54.63	6.8
Yes	54.13	7.23	54.02	9.85

Note. HRQOL = health-related quality of life; RB = retinoblastoma; PROMIS® = patient-reported outcomes measurement information system; PH = physical health; MH = mental health.

*p < .05; **p < .001.

Unique Associations of Other Variables With HRQOL

The list of significant variables associated with physical and mental health summary scores in the RB group are listed in Table 5. Visual acuity, laterality, RSES scores, and total inhalational anesthesia minutes were entered into the stepwise regression for physical health. Gender, MSPSS total scores, and RSES scores were entered for mental health. The final multivariate regression models for physical and mental health summary scores are listed in Table 6. Visual acuity and self-esteem were independent predictors of physical health, accounting for 52% (adjusted R²) of the variance (p < .001). Self-esteem was the only independent predictor of mental health and explained 38% of the variance (p < .001).

Table 5.

Covariates Associated With Physical and Mental Health Summary Scores.

Variable	r	p
Physical health summary score
RSES	.380	<.001
Laterality	−.157	.118
Visual acuity	−.485	<.001
Total inhalation anesthesia (minutes)	−.179	.082
Mental health summary score
RSES	.622	<.001
Gender	−.202	.042
MSPSS	.187	.062

Note. r = Spearman's rho for physical health summary and Pearson product–moment for mental health summary. RSES = Rosenberg Self-Esteem Scale; MSPSS = Multidimensional Scale of Perceived Social Support.

Table 6.

Final Ordinary Least Squares Regression Models for HRQOL in RB Survivors.

Scale	Variable	β	R 2	Adjusted R2	F	p
PROMIS ®-29 physical health summary score	Visual acuity RSES	−.669 .362	.529	.518	51.600	<.001
PROMIS ®-29 mental health summary score	RSES	.620	.384	.378	61.088	<.001

Note. HRQOL = health-related quality of life; RB = retinoblastoma; PROMIS® = patient-reported outcomes measurement information system; RSES = Rosenberg Self-Esteem Scale.

Discussion

Contrary to our central hypothesis, our findings demonstrate that AYA RB survivors perceive their HRQOL as similar to their healthy counterparts. Other studies have reported similar HRQOL between RB survivors and controls or normative samples (Feng et al., 2020; Mouw et al., 2017; van Dijk et al., 2007a; Weintraub et al., 2011). Despite these findings, physical function was significantly worse in RB survivors compared to controls. This finding was similar to one study that examined adult survivors of childhood cancer in which the RB group had worse physical component summary scores compared to other cancer types related to visual impairments (Rueegg et al., 2013). However, other studies with RB survivors did not identify physical function deficits (Batra et al., 2015; van Dijk et al., 2007b; Zhang et al., 2018) or worse physical functioning in the RB group (Feng et al., 2020), which could be due to differences in visual acuity or the lack of visual impairment within the study cohorts. This may explain our findings of similar physical HRQOL even with worse physical function as most of the sample had normal visual acuity in their unaffected eye. Another possible explanation is the lack of other physical functioning ailments associated with the condition. The other domains under physical health such as fatigue, ability to participate in social roles and activities, and pain were not significantly different between RB survivors and controls and contributed to the lack of difference in physical health summary scores. Furthermore, children with congenital birth defects or chronic medical conditions from birth often lack the ability to differentiate their state of wellness from “normal” because their only frame of reference is embedded within their chronic illness. This is often referred to as the “disability paradox” where those with disabilities perceive a higher QOL due to their ability to produce and maintain a sense of balance between their body, mind, and spirit and adapt and find meaning in their lives (Albrecht & Devlieger, 1999). Thus, AYA RB survivors may perceive their HRQOL as good since their current health status (e.g., decreased visual acuity or monocular vision due to enucleation) is what they have always known, compared to someone with a visual deficit due to disease which occurs later in life.

RB survivors reported worse general health and overall QOL than controls on the two PROMIS® general health items. While these single questions are highly correlated to the physical and mental health summary scores, respectively, reliability is lower and the standard error of measurement is higher than the summary scores (Hays, et al., 2015, 2017). Summary scores provide a more comprehensive assessment of physical and mental HRQOL. Despite the majority of survivors having good visual acuity and no chronic medical or psychiatric conditions, future qualitative research may help identify factors that contribute to this perception that cannot be captured on quantitative measures. These findings are pertinent to clinicians that provide follow-up care to RB survivors, as perceived physical health has been shown to decline over the years in other CCS (Reulen et al., 2007). Single-item self-rating scales are easy to administer in the clinical setting and can be useful in clinical decision-making or as a longitudinal measure of health status (Macias et al., 2015). With the possibility of declining visual acuity in RB survivors with age, it is vital to continue to measure perceived health status and overall QOL in this population.

Our findings also showed the RB group to have more depressive symptoms than controls, but this difference only trended toward significance. No other mental health scores were significantly different between these groups. Conversely, one study found significantly worse mental HRQOL in adult RB survivors who reported feeling different from others and had a history of being bullied about their facial appearance and/or visual impairment/blindness (van Dijk et al., 2007b). We did not assess bullying or satisfaction with facial appearance in our study, so it is unclear if that contributed to depressive symptoms in the RB group. Another potential explanation for our findings could be the effects of the COVID-19 global pandemic and civil protests on the mental health of the control group. In the general AYA population, studies have shown increased depression and anxiety due to the COVID-19 pandemic and social isolation (Hawes et al., 2021; Loades et al., 2020; Wang et al., 2020) with females being affected more than males (Hawes et al., 2021). Our study cohort reflects the ethnically diverse Los Angeles community, and the impact of world events may have contributed to the lack of significance between groups as more controls were recruited during this time. We did find higher anxiety and depression and lower mental health summary scores in female RB survivors than males. While increased anxiety and depression in females is not a new finding, these results indicate that female RB survivors may need more frequent screening or additional mental health support during follow-up care. Further research specific to the RB population is warranted to identify risk factors for female survivors.

Our findings differed from three European studies from India, the Netherlands, and China, which reported worse HRQOL in RB survivors associated with specific demographic or clinical factors such as disease severity and age at diagnosis (Batra et al., 2015), the age of participants and treatment received (van Dijk et al., 2007b), and the SES or quality of ocular prosthetics (Zhang et al., 2018). RB survivors in India were older at diagnosis contributing to more advanced stages of the disease, which limits treatment options and could have explained their findings of worse HRQOL (Batra et al., 2015). Adult RB survivors in the Netherlands had extensive radiation exposure, resulting in long-term effects related to the treatment era compared to current therapies. Radiation can have significant cosmetic effects on the eye resulting in orbital asymmetry and potential dissatisfaction with appearance (Mourits et al., 2018). Furthermore, the potential for dissatisfaction with facial appearance was found in RB survivors in China related to the families’ SES and ability to purchase quality ocular prosthetics (Zhang et al., 2018). Our findings reflect a younger age at diagnosis, the current treatment era with the use of minimal radiation, and predominately middle-class SES with insurance that covers ocular prosthetics. This may explain our findings of similar HRQOL compared to controls based on clinical and SES factors that possibly reflect a healthier RB cohort.

Self-esteem in RB survivors emerged as an independent predictor of physical and mental HRQOL. Other studies with CCS have found a moderate to strong correlation between self-esteem and HRQOL (Cantrell & Lupinacci, 2008; Langeveld et al., 2004, Li et al., 2013). Low self-esteem is associated with depression in young adult college students (Choi et al., 2019) and pediatric cancer survivors (Cheung et al., 2019), which can affect HRQOL. Higher self-esteem is associated with better physical health due to decreased health risk behaviors and coping mechanisms, which buffer stressors (Lu et al., 2018). Decreased self-esteem in the AYA RB population could be related to facial appearance and limitations imposed by visual impairment (e.g., sports and career choices; Banerjee et al., 2020).

Consistent with our hypothesis, visual acuity was found to be an independent predictor of physical HRQOL. Since no participants in the unilateral group had visual impairment, this finding is driven by the bilateral survivors. Other studies have also found that visual impairment affected physical HRQOL in RB survivors (Alessi et al., 2004; Rueegg et al., 2012; van Dijk et al., 2007a) with more than half of RB survivors reporting physical problems due to limited peripheral vision and depth perception (Banerjee et al., 2020). These results emphasize the importance of visual preservation in the treatment of RB. Newer treatment options such as ophthalmic artery chemosurgery have decreased enucleation rates and increased retinal function, showing potential for improved visual function and ultimately HRQOL in the RB population (Abramson et al., 2015, 2019; Brodie et al., 2009). Contrary to our hypotheses, younger age at diagnosis and hereditary status were not associated with worse HRQOL in RB survivors.

Social support, in particular support from friends and significant others, was significantly different between groups. However, social support from family members may reflect the age of the cohort, and most have normal visual acuity making them less dependent on family. Most AYAs are seeking independence from the family which leads to an increase in support-seeking behavior from peers and significant others (Szwedo et al., 2017). Lower perceived social support from friends and significant others in the RB group may be attributed to feelings of being different (e.g., eye prosthesis) and not being able to share with peers about their cancer experience (Lewis et al., 2013). In addition, one qualitative study of AYA CCS found that survivors have feelings of being different compared to their peers due to their cancer experience and associated consequences and make every effort to maintain normalcy and limit sharing of information among their peers regarding their cancer experience (Belpame et al., 2019).

Interestingly, we found no significant differences in HRQOL related to clinical variables such as treatment (chemotherapy, radiation, and/or enucleation) or genetic factors (RB1 gene positive). Adult hereditary RB survivors have reported impaired general health perception, which can be related to their increased treatment exposure and risk of developing secondary malignancy (van Dijk et al., 2007b). It is possible that our group of AYAs may not fully comprehend the later life effects of hereditary RB, as parents have reported difficulties in discussing these consequences with their children (van Dijk et al., 2007a). A previous study found that children with RB who received chemotherapy perceived their physical HRQOL to be lower than those who did not receive chemotherapy (Weintraub et al., 2011). Other studies with RB and other pediatric cancer survivors have not found treatment variables to be associated with or predictive of HRQOL (Batra et al., 2015; Halvorsen et al., 2018; Meeske et al., 2007, van Dijk et al., 2007b; Zhang et al., 2018). Parents of children diagnosed with RB often have concerns about enucleation and/or the addition of chemotherapy to their child's treatment regime. These results reinforce that regardless of treatment type AYA RB survivors with good visual acuity have similar HRQOL as their peers.

Implications for Clinical Practice

HRQOL is an important measure that can improve patient–provider communication and patient satisfaction, reveal unknown morbidities, facilitate clinical decision-making and improve healthcare outcomes over time (Cantrell & Kelly, 2015). Our findings that AYA RB survivors have overall similar HRQOL to their peers can provide reassurance to parents of young children diagnosed with RB. No differences were found between treatment regimens with only visual acuity predicting HRQOL. Thus, the continued emphasis on vision preservation during the treatment course is an important aspect of care.

Routine primary care, ophthalmology, or oncology surveillance visits with health care providers should include mental health and psychosocial screening with transitioning to adulthood as depression, self-esteem, and social support concerns were highlighted in this study. The use of peer-support groups or camps for the visually impaired to offer interaction with other RB survivors may be beneficial. The ability to meet someone with the same clinical condition provides the opportunity to share experiences and develop acceptance from peers, which is especially important during adolescence.

Directions for Future Research

Despite physical and mental HRQOL summary scores in RB survivors being comparable to their healthy counterparts, physical function was found to be worse. Self-esteem and social support were also lower in RB survivors. Future studies are needed to examine the effect of visual impairment on physical function/activities of daily living (e.g., exercise, driving, school, and job performance) in this population. Furthermore, interventions to improve modifiable factors such as self-esteem and social support may improve outcomes. Hence, the need for more quantitative research on body image and self-esteem, in particular female RB survivors, can provide additional information on the impact of childhood cancer survivorship. The effects of residual cosmetic defects (e.g., enucleation and prosthetic fit), and visual impairment on HRQOL may be better captured in qualitative studies to assess the “lived experience” of RB survivors.

As new chemotherapy treatments evolve (intraarterial and intravitreal), future studies will be needed to examine treatment effects on visual acuity which was one of the major predictors of HRQOL identified in this study. The impact of visual acuity deficits (e.g., loss of depth perception and peripheral fields common with monocular vision) and the ability for self-care is vital with the transition to adulthood. Lastly, longitudinal studies are needed to examine disease-specific visual deficits with aging and the possible impact on HRQOL in RB survivors.

Limitations

The strengths of this study include a relatively large sample size compared to previous studies of RB survivors, adequately powered to detect an effect size previously reported as significant, and the use of a matched control group. Furthermore, this study included a racially and ethnically diverse sample with a majority of participants identifying as Hispanic/Latino. No other HRQOL studies in RB survivors have included this subgroup of the population. However, this study also has some limitations. Although the linear regression models explained a quarter of the variance for physical and mental health, other factors not accounted for in this study contribute to HRQOL. Due to the cross-sectional design, it is not possible to determine any causal relationships between the independent variables and HRQOL. Since survivors were recruited from a single institution, generalizability to RB survivors from other settings is limited. Future multiinstitutional studies should be conducted to increase the generalizability of findings. In addition, selection bias cannot be ruled out as only survivors seeking active follow-up care and those who responded to letters or phone calls were recruited. Those lost to follow-up and nonparticipating survivors may have worse HRQOL. Since most survivors had normal visual acuity, this could have affected our results. The low reliability of the single-item measures must also be considered.

A significant limitation of this study was the occurrence of the COVID-19 pandemic mid-recruitment. The effects of the pandemic on the HRQOL of RB survivors and healthy controls are indeterminant. This could reflect the lack of statistical significance between groups in the mental summary scores considering more controls were recruited during the pandemic than RB survivors which may not be reflective of a normative sample.

Conclusions

The findings of this study show that a sample of AYA RB survivors in the U.S. has comparable HRQOL compared to healthy age-, gender-, and ethnicity-matched counterparts. Physical function and self-esteem were worse in the RB group compared to controls with only self-esteem and visual acuity explaining the majority of HRQOL. However, the greater part of our cohort had normal visual acuity. Continued assessment of visual acuity is warranted with aging as are interventions to improve self-esteem. Future use of vision or disease-specific measurements may identify more subtle changes or clinical factors that can contribute to HRQOL in RB survivors.