Sex-differences in symptoms and functioning in > 5000 cancer survivors

Background: Previous reports highlight the greater number of side effects that women experience during cancer treatment, but little is known about sex differences in symptoms and functioning in long-term survivors. Methods: We investigated sex differences in the prevalence of physical (EORTC QLQ-C30) and emotional symptoms (Hospital Anxiety and Depression Scale) and loss of functioning (EORTC QLQ-C30) in 5339 cancer survivors (55% males). General linear models were computed to assess the differences in symptoms and functioning between female and male cancer survivors and between survivors and an age-matched reference population. Results: The direct comparison between female and male cancer survivors identiﬁed more symptoms, such as nausea and vomiting (M Z


Introduction
Almost all, not sex-specific, cancers affect men more than women, with the well-documented exception of thyroid cancer [1].Research points to the role of genetic differences in autosomes and sex chromosomes, the influence of circulating hormones and their receptors [2], as well as differences in the immune function and in exposure to risk factors [3].Sex differences also affect pharmacotherapy exposing female subjects to 1.5 times the risk of serious side effects compared to males [4].Gastrointestinal symptoms are more common in females; dermatological symptoms appear to be more common in males [5].Severe side effects up to premature death were also reported more frequent in females in a landmark study in the United States [6].
In addition to biological sex, gender might also impact the health of patients with cancer [7].Gender is a multilayered construct that takes different dimensions into account: e.g.gender identities, norms and relationships [8].Gender identity defines how an individual identifies, e.g. as a woman or a non-binary person; gender norms represent the expectations placed on individuals by their environment, e.g. the expectations that women should be more caring and men more assertive; and gender relationships represent the consequences of power imbalances in human interaction, e.g. the ability to negotiate time allocation within a partnership [8].Gender can impact health independently of sex [9] and is seldom taken into account in medical research.
Cancer therapy has evolved in the past decades, improving the prognosis for patients [10,11].Many survivors are classified as 'disease free' from a medical standpoint; however, the quality of their lives is still significantly impaired even in the absence of persistent malignancy.This can be due to physical and psychological symptoms, as well as social aspects and financial difficulties [12].Impaired Health-related Quality of Life (HRQoL) has been correlated with female sex and feminine gender.These findings are recapitulated in people living with cancer [13e15] and in their caregivers [16]; however, the available knowledge is limited.Importantly, most studies do not detail if biological sex or gender identity was considered.Although more common in female patients, reduced HRQoL is occasionally also reported in male patients, e.g. in stroke, lupus erythematosus, lumbar degenerative disease [17e19] and in some specific forms of cancer [20].These differences originate from a combination of the severity of disease, treatment, and presence of comorbidity, among other factors [17].Some authors have suggested the potential impact of gender roles on subjective quality of life [21].Gender d identity, norms and relations d is challenging to measure according to traditional medical methodology, yet it can significantly impact people's risk behaviour and their resources and resilience after disease [22].This variable could, thus, represent a meaningful source of information to improve patients' long-term journeys.
In the present study, we analyzed the impact of sex differences on long-term, i.e. up to 10 years, symptoms and functioning in cancer survivors and explored if cohort studies, not intentionally designed to investigate gender, could give any indication of its impact on the life of survivors.Comparing diminished functioning in male and female survivors to a reference population generates insight into the relative impact of these impairments, which are hidden in a simple comparison between female and male cancer survivors.Since the ageing process can naturally impact functioning domains, a direct comparison with cancer-free individuals of the same sex/ gender is necessary to reveal potential sex-specific and gender-specific consequences of living with a tumour.

Design and data-collection
This is a secondary cross-sectional analysis of data collected by the PROFILES (Patient-Reported Outcomes Following Initial treatment and Long-term Evaluation of Survivorship) registry [12].The PRO-FILES registry is an ongoing cohort study of cancer survivors initiated in 2008.A detailed description of the data collection has been previously published [12].Data from the PROFILES registry are freely available for noncommercial scientific research, subject to submission of a study question, privacy and confidentiality restrictions, and registration (www.profilesregistry.nl).

Patient population
The analysis covers four study cohorts from the PRO-FILES registry, only including nonsex specific cancer types: colorectal, hematologic cancers (chronic lymphocytic leukaemia, non-Hodgkin lymphoma, Hodgkin lymphoma and multiple myeloma), basal cell/squamous cell and thyroid cancer.Patients were recruited between January 2009 and June 2014.Detailed information about the inclusion of the study samples is published elsewhere [23e26].Participants were excluded if they were not able to complete a Dutch questionnaire according to their (former) attending specialist (due to, i.e. cognitive impairment, non-native speaker, too ill to participate).Individuals who died or emigrated prior to the start of the study, according to data from the Dutch municipal personal records database and/or data from the hospital of diagnosis, were also excluded.Ethical approval was obtained for all study samples separately from a local certified medical ethics committee (Maxima Medical Centre Veldhoven, the Netherlands).

Demographic, social and clinical measures
Sociodemographic information was obtained from the NCR (sex and age at diagnosis) or self-reported questionnaires (age at questionnaire, education, marital status and comorbidity).Sex was defined in a binary manner (female/male).The questionnaires did not include specific items to assess gender identity and norms.Comorbidity was assessed using the adapted Self-administered Comorbidity Questionnaire (SCQ) [27].
Clinical data (age at diagnosis, tumour type, primary treatments, date of diagnosis), were also obtained from the NCR.Tumour type was classified according to the third International Classification of Diseases for Oncology (ICDO-3) [28,29], and cancer stage was classified according to TNM or Ann Arbor Code (Non-Hodgkin lymphoma and Hodgkin lymphoma).Primary treatments received were classified into surgery, systemic therapy (chemotherapy, targeted therapy, immune therapy), radiation therapy (including brachytherapy), hormone therapy, no treatment/active surveillance or unknown.

Symptoms and functioning
The EORTC QLQ-C30 (version 3.0) was used to assess health-related quality of life (symptoms and functioning) [30].The questionnaire addresses five different functioning dimensions: physical, role, emotional, cognitive and social.Symptom scales included were fatigue, nausea/vomiting, pain, dyspnea, insomnia, appetite loss, constipation and diarrhoea.Answers were linearly transformed into a score ranging from 0 to 100 [31].The total number of symptoms was calculated by summing dichotomized symptom scales (a little, quite a bit or very much agreement with the proposed symptom).
We employed the Hospital Anxiety and Depression Scale (HADS) to assess symptoms of anxiety and depression.The HADS includes separate anxiety and depression scales, which both consist of seven items.All items were scored on a 0 to 3-point Likert scale, with higher scores indicating more symptoms.Total scores for anxiety and depression ranged from 0 to 21 [32].

Data of the reference population
Since 2009, the PROFILES registry annually collects information on symptoms, functioning, comorbidities and lifestyle in a cohort representative of the Dutchspeaking population in the Netherlands [33].A random selection of one cancer-free member per household (N Z 1408) was made to ensure the independence of observations and was randomly matched to the patient sample based on sex and age group (<40, 40e49, 50e59, 60e69, !70) based on the frequency distribution.By using frequency matching, a total of 389 panel members could be matched to the patients included in this study (ratio norm population: cancer patients is 1:0.07).For supplementary analyses, 365 and 629 panel members were matched to colorectal and hematologic cancer patients, respectively (ratio cancer patients: norm population is 0.14 and 0.36).
General linear models were computed to assess the differences in functioning and symptoms, between male and female patients, in the total sample and by tumour type, adjusted for predetermined covariates (i.e.age, tumour type, treatments received, cancer stage, number of comorbidities, the time between diagnosis and questionnaire, partner status, educational level, employment status).
General linear models were used to separately compare patients to a reference population for male and female patients.For supplementary analyses, we additionally assessed the two larger cohorts, colorectal and hematologic cancer patients, separately.We adjusted for covariates that were available in both patient and norm samples (i.e.age at questionnaire, number of comorbidities, partner status, employment status).Statistical analyses were conducted using SAS version 9.4 (SAS Institute, Cary, NC, 1999).
The female patients reported having a partner less often than the males (70% versus 83%, p < 0.01), were less likely to be employed (20% versus 24%, p < 0.01), and had received less formal education.Overall, female patients were more often diagnosed in stage I, a finding primarily attributable to the thyroid cancer cohort.No sex differences were detected at all other cancer stages.
Overall, female patients appeared more likely to obtain surgery as well as radiotherapy compared to males.Male patients were more likely to obtain systemic therapy.No significant sex differences could be identified in the time since diagnosis and invitation to participate in the study.

Discussion
To our knowledge, this study represents the first examination of sex differences in symptoms and functioning in long-term cancer survivors.Overall, female patients reported more comorbidities, more symptoms and lower functioning scores in direct comparison to male patients.However, comparison to an age-matched reference population without cancer reduced or inverted these differences.Contrary to our expectation, several symptoms such as fatigue and dyspnea, as well as anxiety and depression, were reported more frequently in the male patients than in the female ones when adjusted to an age-matched reference population.In addition to a higher level of physical and psychological symptoms, males also displayed a more significant net loss in the role and social functioning than females compared to a reference population.
Long-term symptoms can impact the physical, mental and functional domains of patients' lives.According to our results, female and male patients experience these effects at different rates.However, these differences are not consistently addressed when providing information, care and counselling to cancer patients.If female and male patients experience a different risk of long-term physical symptoms, this should be openly discussed when therapeutic decisions are made.Although knowledge of future symptoms will not modify their occurrence, awareness can increase patients' agency.More detailed and nuanced information, if desired, might support their resilience and coping abilities [22,36].
Long-term symptoms after cancer therapy can lead to a significant impairment of HRQoL and restrictions in individual functioning.In addition to the physical and psychological effects, broader domains of personal and professional life are affected.Females in the general population, as well as female patients, appear to experience lower levels of functioning and higher levels of symptoms compared to males [17e19].Most studies base their comparisons on patient populations without including reference controls.A recent publication by Nolte and colleagues [37] provided norm data on the EORTC QLQ-C30, which recapitulate the higher incidence of symptoms in females and the relatively higher functioning scales in males in the general population.Sex differences are, thus, present in the general population.In our sample, female patients reported significantly more physical symptoms.By using a sample of the general population as a reference, only insomnia and appetite loss were significantly worse in female cancer patients compared to males.The net prevalence of symptoms was much lower in female patients than expected by the direct comparison to male patients.In the male patient population, this effect was reversed.Although in a direct comparison with female patients, male patients seemed to experience fewer symptoms, they actually reported more symptoms than females when compared to their agematched peers without cancer.These findings should be taken into account when providing patients with information before starting therapy and in designing regimens to mitigate its long-term impact.If we are striving for more person-centred care, services should be sex-specific since the persistence of symptoms does not appear to be the same in both sexes.
In addition to the clinical relevance of our findings [38], there is also a theoretical one.In this study we also tried to explore if the available data would point towards a role of gender, not solely biological sex.The influence of gender has been reported as distinct from biological sex in predicting the recurrence of cardiovascular disease after an initial event [9,39].Although sex and gender influence each other throughout the life course [40], their impact on health might differ, as reported by Pilote and colleagues.The instrument used in their study includes multiple questions addressing different dimensions of gender, e.g.norms and relations [39].We propose that the role and social functioning scales of the EORTC QLQ-C30 questionnaires could be considered in this light.The cohorts included in our analysis do not offer any specific information about gender assessed with validated instruments, but the identified impact on functional domains highlights how the consequences of living with cancer cannot be solely attributed to biological differences.While the prevalence of physical symptoms might be primarily related to biological differences, role and social functioning transcend biology [41].The diminished role and social functioning identified in male patients compared to the reference population might, therefore, exemplify the impact of cancer on gender norms and relations [8,42].
Gender norms and relations define people's interaction with their environment and shape their identity and choices.Next to their role as potential predictors of recurrence [9] and coping behaviour [36], gender dimensions could have a wider impact on cancer survivors' long-term health.When informing patients about the long-term consequences of cancer therapy, we should not only include information about sex differences but also about the potential gender-specific impact of functionality changes.Male patients reported a higher level of depression and anxiety than females compared to the reference population.These effects could be related to loss of role and social functioning.Our current data does not allow investigating these dynamics, and future studies are needed to address these questions.
The current study includes a large, wellcharacterized population of cancer survivors and reference population, which allowed us to gain insight into previously unaddressed aspects of survivorship.Nevertheless, some limitations need to be considered.The included population is a selected group of patients willing to participate in a cohort study and might, thus, not fully represent the general population of cancer patients.Previous reports have described higher survival rates and increased male participation in this cohort [35].The female patients in the study reported lower education obtainment than the male patients, as well as living alone more frequently.Although these differences are statistically significant, they are representative of the sex differences in the general Dutch population and are also present in the matched reference population.The included cancer types also differ in their physical expression, and the timing of diagnosis might differ between the populations.Nevertheless, the trends identified in the whole group were clearly shared in the single subgroups, pointing towards a general trend.
In conclusion, we report significantly higher physical and psychological symptoms and lower functioning in male cancer survivors compared to a matched reference population of cancer-free individuals.Not only did males experience more pronounced levels of symptoms compared to females, but they also experienced a more significant loss of role and social functioning.These differences should be further explored, ideally employing mixed-methods approaches that can allow for an indepth exploration of the psychosocial, emotional and cultural aspects involved.Furthermore, we propose to specifically investigate the impact of gender norms and relations on the quality of life of cancer survivors, given the identified difference in role and social functioning.The present findings suggest that details on the impact of both sex and gender might be needed to provide adequate counselling and care for long-term cancer survivors.

5 )
Independent sample t-tests were used for continuous variables and Pearson's c 2 tests for categorical variables.Abbreviations: M Z mean; SDZStandard Deviation, *p < 0.05, **p < 0.01.Percentages may not always add up to 100 because they have been rounded off to whole numbers.a Includes indolent (including chronic lymphocytic leukemia) and aggressive B-cell non-Hodgkin lymphoma, Hodgkin lymphoma and multiple myeloma.b According to TNM.Ann Arbor Code was used for Hodgkin lymphoma and Non-Hodgkin lymphoma.For Chronic lymphocytic leukaemia and Multiple myeloma tumour stage was not determined or registered.c Systemic therapies were: chemotherapy, targeted therapy and immune therapy.S. Oertelt-Prigione et al. / European Journal of Cancer 156 (2021) 24e34

Table 1
Sociodemographic and clinical characteristics, by cancer type.

Table 1 (
continued ) ). Female patients described more nausea and vomiting, insomnia, appetite loss, constipation and diarrhoea.Most patients reported more than one symptom, females significantly more than males (mean 2.7 [95% CI Z 2.7; 2.8] versus.2.2 [95% CI Z 2.1; 2.3] respectively) except among thyroid cancer patients where no sex differences in the number of reported symptoms were identified.

Table 3
Comparison between cancer survivors and the age-matched reference population.
Analyses were adjusted for covariates available in both patient and norm samples (i.e.age at questionnaire, number of comorbidities, partner status, employment status).*p<0.05, **p < 0.01.aReferencepopulation was randomly matched to patients based on age (<40, 40e49, 50e59, 60e69, >70) and sex.bSum of symptoms fatigue, nausea/vomiting, pain, dyspnea, insomnia, appetite loss, constipation and diarrhoea, if a little, quite a bit or very much endorsement of the symptom-item or on any item of the symptom-scale.cHADS data was missing for basal/squamous cell survivors.