The age-related impact of surviving sarcoma on health-related quality of life: data from the SURVSARC study

Background Health-related quality of life (HRQoL) data of sarcoma survivors are scarce and the impact of age remains unclear. The aims of this population-based study were to (i) compare HRQoL scores amongst three age-groups [adolescents and young adults (AYA, aged 18-39 years), older adults (OA, aged 40-69 years) and elderly (aged ≥70 years)]; (ii) compare HRQoL of each sarcoma survivor age group with an age- and sex-matched normative population sample; (iii) determine factors associated with low HRQoL per age group. Methods Dutch sarcoma survivors, who were 2-10 years after diagnosis, were invited to complete the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-Core 30-questions questionnaire on HRQoL. Results In total, 1099 survivors (58% response rate) completed the questionnaire: 186 AYAs, 748 OAs and 165 elderly. The median time since diagnosis for all patients was 5.2 years. Bone sarcomas were seen in 41% of AYAs, 22% of OAs and in 16% of elderly survivors (P < 0.01). AYA and OA survivors reported statistically significant and clinically meaningful worse physical, role, cognitive, emotional and social functioning compared with a matched norm population, which was not the case for elderly survivors. AYAs reported significantly worse scores on emotional and cognitive functioning compared with OA and elderly survivors. Malignant peripheral nerve sheath tumour, osteosarcoma and chordoma were the subtypes of which survivors reported the lowest HRQoL scores in comparison with the norm. For all age groups, chemotherapy, having a bone sarcoma and having comorbidities were most frequently associated with low scores on HRQoL subscales, whereas a shorter time since diagnosis was not. Conclusion In this nationwide sarcoma survivorship study, the disease and its treatment had relatively more impact on the HRQoL of AYA and OA survivors than on elderly survivors. These results emphasise the need for personalised follow-up care that not only includes risk-adjusted care related to disease relapse, but also age-adjusted care.


INTRODUCTION
Sarcomas are rare tumours originating from mesenchymal stem cells and account for approximately 1% of adult malignancies. With more than 70 histological subtypes, sarcomas are extremely heterogenous tumours that affect people of all ages and can occur at any anatomical site. 1 Broadly, a histological distinction can be made between soft tissue sarcomas (STS) and bone sarcomas (BS), with a 5year survival rate of 55%-60% and 50%-55%, respectively. [2][3][4] These poor outcomes are at least partially ascribed to delay in diagnosis, advanced disease at presentation, biological aggressiveness and, due to its heterogeneity, a relative lack of clinical trials. 1,[5][6][7][8][9] Sarcomas may present with complex symptoms and the clinical picture can vary from indolent to a highly aggressive phenotype. 1 The latter consequentially leads to a high burden of symptoms and requires intensive treatment, potentially resulting in long-term side-effects and disabilities in survivors, [10][11][12] which in turn can result in a decreased health-related quality of life (HRQoL). 13 HRQoL is a patient reported outcome and encompasses physical, social, psychological, cognitive and spiritual aspects of well-being. Both HRQoL and survivorship are increasingly considered important components of patient-centred care and research suggests that cancer patients consider HRQoL an important treatment outcome. [14][15][16] In personalising the care of sarcoma patients, much attention is paid to the various histological subtypes, their treatment and involved specialists. However, less attention has been paid to age aspects that are linked to sarcoma, even though sarcomas demonstrate an extremely diverse nature across the age spectrum both in terms of subtypes and incidence. 17 In addition, the challenges that cancer patients in general face vary greatly at different ages. Adolescent and young adult (AYA) sarcoma patients are diagnosed at an emotionally, cognitively and socially challenging time in their lives, which may interfere with the acquisition of regular developmental milestones. 18 In contrast, key issues that must be considered in treatment of elderly patients are physiological changes associated with aging, functional and role functioning, comorbidities, cognitive status and polypharmacy. 19 Lack of representation in clinical trials is seen in both AYA and elderly patients. 20,21 A recent study identified significant age-related differences in the sarcoma patient journey that are not only related to variations in sarcoma subtypes, such as more incorrect diagnoses and higher burden of treatment in younger patients and less referral to rehabilitation services in the elderly. 22 HRQoL in sarcoma survivors has been marginally researched. This is a nationwide study with the purpose of assessing HRQoL in sarcoma survivors. The aim of this study is to (i) assess age-related differences in HRQoL between AYA (aged 18-39 years), older adult (OA, aged 40-69 years) and elderly (aged !70 years) sarcoma survivors. In addition, (ii) general HRQoL will be compared between sarcoma survivors and an age-and sex-matched normative sample and (iii) patient, tumour and treatment characteristics associated with low HRQoL score will be determined.

Study design and participants
This SURVSARC study is an exploratory population-based cross-sectional questionnaire study among adult (!18 years of age) sarcoma survivors, registered in the Netherlands Cancer Registry (NCR). Sarcoma survivors diagnosed between 1 January 2008 and 31 December 2016 at one of the six participating sarcoma expertise centres (Radboud University Medical Centre, Antoni van Leeuwenhoek/The Netherlands Cancer Institute, University Medical Centre Groningen, Leiden University Medical Centre, Erasmus Medical Centre Rotterdam, Maastricht University Medical Centre) were eligible. Exclusion criteria were cognitive impairment and physical condition too poor to participate. Survivors with desmoid fibromatosis, grade I chondrosarcoma, gastrointestinal stromal tumours, atypical lipomatous tumours or giant-cell tumours were also excluded considering the indolent clinical behaviour and less aggressive treatment strategies. Ethical approval was obtained from the medical ethical committee of the Radboud University Medical Centre (2017-3944) and the study was registered in the Dutch Trial Registry (NTR-7253).

Data collection
Eligible sarcoma survivors were invited by their (former) treating physician and if informed consent was obtained, participants were able to complete the questionnaire. The NCR contains data on patient, tumour and treatment characteristics of all newly diagnosed cancer patients in the Netherlands. Demographic, clinical and treatment characteristics were obtained from this database. All histology was verified through electronic patient records. Completion of the questionnaire was conducted between October 2018 and June 2019 within the Patient Reported Outcomes Following Initial treatment and Long-term Evaluation of Survivorship (PROFILES) data management system; patients were therefore 2-10 years after diagnosis. 23

Study measures
Marital status, educational level and employment status were self-reported by the participants. Treatment characteristics were patient-reported and any missing data were supplemented with treatment data from the NCR. In order to report on the age-related HRQoL in sarcoma survivors, participants were divided into three age categories according to their age at diagnosis; adults and young adolescents (aged 18-39 years), OAs (aged 40-69 years) and elderly survivors (aged !70 years).

HRQoL
In order to assess cancer-generic HRQoL, sarcoma survivors completed the Dutch version of the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-Core 30-questions (EORTC-QLQ-C30). 24 It was designed for cancer patients in general and consists of 30 items. The questionnaire contains five functional scales, a global QoL scale, three symptom scales and a number of single symptom items. A linear transformation of all scales and single item measures was conducted and scores ranged from 0 to 100.

Normative sample
An age-matched and sex-matched normative sample without cancer was obtained from CentERdata, a research institute at Tilburg University, using a household panel representative of the population in the Netherlands. The panel members were randomly matched to each age group of sarcoma survivors separately based on sex and age at the time of questionnaire completion. 25

Statistical analyses
An anonymous comparative analysis between responders and non-responders was conducted by an NCR employee and non-responder data were not shared with the research team. When comparing groups, chi-square tests were used for categorical variables, whereas independent samples ttests and one-way analysis of variance were used for continuous variables.
Analysis of covariance (ANCOVA) was conducted to compare HRQoL functioning scales and symptom scales between survivors and the normative population by age group. These analyses were corrected only for age at the time of questionnaire completion and not for other potential confounding factors (e.g. education, partner, comorbidity) since variance between survivors and the norm could be caused by having cancer and the subsequent treatment. In addition to the statistical significance, the clinical relevance of the difference in scores was determined according to the Evidence-Based Guidelines for Determination of Sample Size and Interpretation of EORTC QLQ-C30 as determined by Cocks et al. 26 ANCOVA was conducted to compare the HRQoL between the three age groups, correcting for time since diagnosis, comorbidity and whether patients had received chemotherapy. In case of significant effects, Bonferroni post hoc tests were carried out.
Univariate, age-stratified logistic regression analyses were used to determine the association between patient, tumour and treatment characteristics and clinically relevant low scores on functioning and symptom scales.
All statistical analyses were carried out using SPSS Statistics (IBM Corporation, version 26.0, Armonk, NY) and P values < 0.05 were considered statistically significant.

Responders versus non-responders
In total, 1887 sarcoma survivors were approached, of whom 1099 provided informed consent and completed the questionnaire. The response rate for AYAs was 41%, for OAs 66% and for elderly 53%. Elderly responders were more often male than non-responders (P < 0.01) and AYA and OA responders were older than non-responders, whereas elderly responders were younger than non-responders (P < 0.01). No statistically significant differences were observed for STS versus BS between responders and non-responders, stratified by age group.

Sociodemographic characteristics of survivors and the normative sample
An age-stratified comparative analysis of all 1099 responding sarcoma survivors compared with the 1319 individuals from the normative sample was conducted. Due to the younger age of the normative population compared with the study participants, the age matching based on age categories for the elderly population was imperfect. The age at the time of the survey was significantly higher in elderly survivors compared with their norm (P < 0.01). OA survivors had a partner more frequently than the OA norm (P < 0.01). Compared with elderly sarcoma survivors, individuals in the elderly norm were more often highly educated (P < 0.01). Both the AYA and elderly norm were more frequently reported to have either no comorbidities or two or more comorbidities compared with the survivors in their corresponding age groups.

Comparison of HRQoL between survivors and the norm according to age
In comparison with an age-and sex-matched normative sample, both AYA and OA sarcoma survivors reported significantly lower scores on five of the six functional scales ( Figure 1A and B). In the elderly, no functional scales differed significantly between survivors and the norm after correction for age was applied ( Figure 1C).
On the nine symptom scores, AYA survivors only reported significantly worse scores on financial difficulties which were of small clinical importance ( Figure 1D). In OA, six symptom scores were significantly worse for survivors compared with their norm ( Figure 1E). Within the elderly age group, survivors reported no statistically significant higher scores in comparison with the elderly norm population ( Figure 1F).

Comparison of HRQoL between age groups
Lower emotional and cognitive functioning in AYA survivors was observed in comparison with OAs and the elderly, which was also statistically significant after correcting for time since diagnosis, comorbidity and whether patients had received chemotherapy ( Table 2). Physical and role functioning was significantly worst for elderly survivors after correction. AYAs and OAs scored significantly higher on financial difficulties compared with elderly survivors.

HRQoL scores stratified by histological subtype
In Table 3, the scores for several functioning and symptom scales are depicted stratified by age group and by histological subtype. Especially malignant peripheral nerve sheath tumour, chordoma and osteosarcoma showed large clinically relevant differences between survivors and the norm population.
Patients, tumour and treatment characteristics associated with low HRQoL scores Extremity localisation had an odds ratio of 2.35 (95% confidence interval 1.25-4.39) for low physical functioning in  The clinical relevance of the emotional functioning score was determined according to the role functioning interpretation guideline. *P < 0.05 and trivial effect. **P < 0.01 and small effect. ***P < 0.01 and medium effect. AYAs and not in OAs and elderly survivors (Table 4). Chemotherapy as part of the treatment strategy, having a BS and having multiple comorbidities were most often associated with low scores on subscales. A shorter time since diagnosis did not seem to be associated with impaired HRQoL.

DISCUSSION
This population-based, cross-sectional study assessing HRQoL amongst sarcoma survivors showed that AYA and OA survivors reported statistically significant and clinically meaningful worse physical, role, emotional, cognitive and social functioning in comparison with the normative population, which was not the case for elderly survivors. This suggests that being diagnosed and treated for sarcoma has a greater impact on the functional status of younger survivors than older survivors. A possible explanation for the lesser impact of sarcoma on elderly survivors could be that due to the natural course of aging, a decline in functional status occurs also in elderly without cancer and therefore the difference in HRQoL is not significant. Explanations for the greater impact on younger survivors might be that they have higher work-related and social demands, are emotionally more vulnerable as they are still establishing an identity and lack effective coping strategies since they have never experienced severe illness. Interestingly, AYA survivors reported significantly worse scores on the functional scales compared with their norm, however they did not report worse scores on the symptom scales. The same phenomenon was seen in AYA patients with thyroid cancer. 27 A possible explanation might be that younger survivors have more physical resilience than older patients and thus experience symptoms to a lesser extent. 28 Notably, AYAs received chemotherapy most often and were most often diagnosed with BS in comparison with OAs and elderly. This might play a part in the low functioning scores in AYA survivors, but apparently had no effect on symptom scores. AYA survivors reported significantly worse scores on emotional and cognitive functioning compared with OA and elderly survivors and worse social functioning compared with elderly survivors. Considering the generally assumed better HRQoL in the young, these striking results emphasise the impact of sarcoma on HRQoL in AYA survivors once again. The time in their lives at which AYA patients are diagnosed is an important time for establishing an identity and making significant life decisions.
Concerning specific sarcoma subtypes, the difference in HRQoL between the normative population and chordoma survivors was striking and once again emphasises the severity of this sarcoma subtype, that is mainly localised in the axial skeleton. 29 With these small numbers, however, caution has to be taken with generalisations.
Interestingly, survivors who were >5 years since diagnosis did not have significantly better HRQoL scores in comparison with survivors who were <5 years since diagnosis. This suggests that any improvement in HRQoL occurs in the first years after diagnosis and after 5 years no further improvement can be expected in general. A longitudinal investigation of HRQoL in AYA survivors in the first 2 years after diagnosis found that the most improvement in HRQoL occurs within the first year after diagnosis. Improvement in HRQoL was also seen in the second year after diagnosis, although this was far less than in the first year. 30 The results from this study are in line with studies on HRQoL in thyroid cancer survivors and diffuse large B-cell lymphoma survivors, which both reported that the difference between survivors and the norm population was Table 4. Univariate age-stratified logistic regression analyses amongst sarcoma survivors for the odds of (1) having a clinically relevant worse score on the concerning subscale compared with the normative population versus (0) having no clinically relevant worse score compared with the normative population. The threshold used was for a small clinical importance as determined by Cocks et al. 26  The demographic and clinical factors that were statistically significant in the univariate logistic regression analysis were then put into a multivariate logistic regression analysis. If the odds ratio remained statistically significant in the multivariate regression analysis, the cells of the table displaying this odds ratio is grey.
largest for the younger age group. 27,31 A similar study on HRQoL in multiple myeloma survivors did not find agerelated differences. 32 A pooled analysis on the effects of age on HRQoL in cancer populations concluded that HRQoL is generally impaired in cancer patients, however the impact on specific domains varies with age. Young cancer patients had more financial problems and worse social and role functioning in comparison with the general population, which is largely in line with the results from this study. 33 Limitations of this study include a possible selection bias, since it is unknown whether survivors did not participate due to either a poor health or an absence of symptoms. 34 A challenge in interpreting these age-stratified results remains the extreme heterogeneity of sarcoma across the age spectrum. This study attempted to take the heterogeneity of subtypes and treatment into account by combining HRQoL outcomes with clinical data. Furthermore, survivorship bias might impede the generalisation of these results to all sarcoma patients. Strengths of this study are that it is population-based and includes a large number of patients. Considering the limited amount of data available on HRQoL in sarcoma survivors, this study is a valuable addition and provides more insight into the impact of sarcoma on HRQoL and the role of age.
In order to meet the needs of sarcoma survivors on physical, emotional, role, cognitive and social functioning, a specific measurement strategy for HRQoL in the extremely heterogenous sarcoma survivor population is essential. 35 The incorporation of age-specific questions is crucial in such a measurement strategy to meet the age-specific needs of sarcoma survivors. The benefits of addressing the psychological and social issues in sarcoma patients may be lifelong, especially in younger patients who are establishing an identity and making important life choices that will affect the rest of their lives. The results from this study emphasise the need for the development of appropriate psychological and social interventions. Ideally, the psychological and social needs and issues in sarcoma patients should at least always be addressed at some point throughout the disease trajectory.

CONCLUSION
In conclusion, this nationwide study shows that in a very heterogeneous sarcoma survivor population, there is also a wide variety in HRQoL scores. Nevertheless, clear agerelated patterns in HRQoL are observed showing sarcoma has a greater impact on younger survivors. These results emphasise the need for personalised follow-up care that not only includes risk-adjusted care related to disease relapse, but also age-adjusted care to address the needs that impact important aspects of HRQoL.