Depressive symptoms in relation to overall survival in people with head and neck cancer: A longitudinal cohort study

Abstract Objective The objective of the study is to investigate the relation between pretreatment depressive symptoms (DS) and the course of DS during the first year after cancer diagnosis, and overall survival among people with head and neck cancer (HNC). Methods Data from the Head and Neck 5000 prospective clinical cohort study were used. Depressive symptoms were measured using the Hospital Anxiety and Depression Scale (HADS) pretreatment, at 4 and 12‐month follow‐up. Also, socio‐demographic, clinical, lifestyle, and mortality data were collected. The association between before start of treatment DS (HADS‐depression > 7) and course (never DS, recovered from DS, or persistent/recurrent/late DS at 12‐month follow‐up) and survival was investigated using Cox regression. Unadjusted and adjusted analyses were performed. Results In total, 384 of the 2144 persons (18%) reported pretreatment DS. Regarding DS course, 63% never had DS, 16% recovered, and 20% had persistent/recurrent/late DS. People with pretreatment DS had a higher risk of earlier death than people without DS (hazard ratio (HR) = 1.65; 95% confidence interval (CI) 1.33‐2.05), but this decreased after correcting for socio‐demographic, clinical, and lifestyle‐related factors (HR = 1.21; 95% CI 0.97‐1.52). Regarding the course of DS, people with persistent/recurrent/late DS had a higher risk of earlier death (HR = 2.04; 95% CI 1.36‐3.05), while people who recovered had a comparable risk (HR = 1.12; 95% CI 0.66‐1.90) as the reference group who never experienced DS. After correcting for socio‐demographic and clinical factors, people with persistent/recurrent/late DS still had a higher risk of earlier death (HR = 1.66; 95% CI 1.09‐2.53). Conclusions Pretreatment DS and persistent/recurrent/late DS were associated with worse survival among people with HNC.


| BACKGROUND
Clinical depression as well as depressive symptoms (DS) have been reported to increase mortality and reduce survival in different populations. [1][2][3] Among people with different types of cancer, those with a clinical diagnosis of minor or major depression have a 39% higher risk of dying during the follow-up period than people without depression. 1 People with increased levels of DS, as measured using validated patient-reported outcome measures, have a 25% higher risk of dying during the follow-up period. 1 People diagnosed with head and neck cancer (HNC) are prone to depression or DS. 4,5 Previous studies on the association between clinical depression 6 or DS [7][8][9][10][11][12][13] and survival in people with HNC reported mixed results. Some studies reported no association, 7,8 while others reported worse survival or higher mortality in people with depression or DS. 6,[9][10][11][12][13] Half of these studies were, however, limited by small number of events (eg, disease-related or overall deaths), 7,[9][10][11][12][13] hampering the ability to account for different covariates in the survival analyses.
In addition, most studies were limited to a single measurement of clinical depression or DS, 7,8,[10][11][12][13] mostly prior to treatment. 7,10-13 As previously reported, 14 pretreatment DS may result from the short-term response to cancer diagnosis and may not necessarily reflect a person's long-term course of DS and, therefore, may be a less important associated factor of survival than DS at follow-up.
A previous study reports that, in 40% of people with HNC, DS level indeed changed between the pretreatment and posttreatment measurement. 15 Four different courses of DS were identified: people without DS, people who developed DS (33%), people who recovered from DS (7%), and people with persistent DS (4%). A recent study comparing survival outcomes of people with lung cancer reported on 4 comparable courses of DS. 16 They found that people who developed or had persistent DS had an increased risk of earlier death, while people who recovered had the same risk as the reference group of people who never reported DS.
A recent large longitudinal study that measured depression more than once in people with HNC in relation to survival found that depression in the 2 years before HNC diagnosis as well as depression in the year after diagnosis was associated with worsened cancer-specific and overall survival. 6 In that study, however, no distinction was made between people who recovered from their depression during follow-up and those who did not. In addition, depression was defined as a registered clinical depression diagnosis based on Medicare claims data. The generalizability of these findings to people with DS or undiagnosed depression is unclear.
This study, therefore, aimed to investigate the relation between pretreatment DS as well as the course of DS during the first year after cancer diagnosis and overall survival among people with HNC.

| Design and study population
In this study, data from the Head and Neck 5000 prospective clinical cohort study was used (dataset version 2.1), 17

| Measures
The English version of the Hospital Anxiety and Depression Scale (HADS) was used to assess psychological distress (HADS-total), level of DS (HADS-D), and level of anxiety symptoms (HADS-A) before the start of treatment, and at 4 and 12-month follow-up. 19,20 A HADS-D > 7 was used as a cutoff for identifying persons with DS. 21 Internal consistency of the HADS-D in this study was α = .851.
Study-specific questions were used to measure pretreatment tobacco use and alcohol consumption. Tobacco use was categorized as current smoker, former smoker, or never smoked. 22 For alcohol consumption, people were categorized as nondrinkers, moderate drinkers (1-14 drinks per week), hazardous drinkers (14-35 drinks/ week for women and 14-50 drinks/week for men), or harmful drinkers (>35 drinks/week for women and >50 drinks/week for men). 22 In addition, age, gender, marital status, education level, annual household income, and deprivation status were measured. Deprivation status was measured using the Index of Multiple Deprivation (IMD) 2010. 23

| Clinical information
Clinical information was abstracted from the hospital information  24 At the start of the study, participants were flagged with the United Kingdom Health and Social Care Information Centre so that the study team was provided with information on overall mortality (mortality and mortality date).

| Statistical analyses
All analyses were performed using the IBM Statistical Package for the Social Science (SPSS) version 23 (IBM Corp., Armonk, NY USA).
Chi-square tests and independent samples t-test analyses were used to analyze differences between groups.
To assess the association between pretreatment DS and overall survival, a series of Cox regression analyses were performed. At first, minimally adjusted analyses adjusted for age and gender were performed. Analyses were performed in the total population as well as in people with oral cavity, HPV-positive oropharyngeal, and HPV-negative oropharyngeal and laryngeal cancer separately. Survival time was defined as days from date of consent to censoring or date of death.
Besides these minimally adjusted analyses, we investigated whether potential associations remained after adjusting for socio-demographic and clinical factors. Also, Cox regression analyses adjusted for lifestyle-related factors were performed. Previous literature hypothesized that lifestyle may mediate the association between depression or DS and survival. 3,25 However, other studies added lifestyle as a potential confounder to the model. 7,8 Results can, therefore, be interpreted either as the direct effect after taking the potential mediating role of lifestyle into account or as the association that remains after adjusting for lifestyle as a potential confounder. Finally, post hoc analyses were performed by including each factor 1 by 1 to the minimally adjusted model, to investigate which factors had a strong influence on the association between DS and survival (defined as >10% change in hazard ratio (HR)). All categorical variables adhered to the proportional hazard assumption. Multicollinearity was not found.

| RESULTS
The HADS-D score of the total study population (n = 2144) was on

| DISCUSSION
Using data from the Head and Neck 5000 study, 17,18 it was found that 13% to 18% of people with HNC experience pretreatment DS. During the first year after diagnosis, 63% of people with HNC never had DS, 16% recovered from DS, and 20% had persistent/recurrent/late DS. Another pathway may be that untreated depression can cause suicide. 25 Although suicide is, compared to other diseases, relatively common among people with HNC, 28 in absolute terms, it is a rare event. Also, tumor-related and patient-related biomarkers of endocrine, immune, and autonomic (dys)function or other clinical variables may explain the association between depression and survival. 25 This might explain why we found a potential association between pretreatment DS and overall survival in people with HPV-negative oropharyngeal cancer and not in HPV-positive oropharyngeal cancer.
However, future research is warranted to replicate these findings and to explore the specific role of HPV status and other biomarkers.

| Clinical implications
People with pretreatment DS as well as persistent/recurrent/late DS are at increased risk of earlier death. Previous studies have hypothesized that lifestyle and suicide may explain (part of) this association.
Also, tumor-related or patient-related biomarkers are hypothesized to mediate this association.   b HADS-D above threshold at baseline and/or 4-month follow-up, but recovered at 12-month follow-up. c HADS-D above threshold at 12-month follow-up, regardless of outcome at baseline and 4-month follow-up.