The course of health-related quality of life in the first 2 years after a diagnosis of head and neck cancer: the role of personal, clinical, psychological, physical, social, lifestyle, disease-related, and biological factors

There is convincing evidence that cancer patients have to deal with physical, psychological, and social side effects of the disease and its treatment, negatively affecting health-related quality of life (HRQOL) [1, 2]. In HNC patients, side effects such as oral dysfunction and related swallowing and speech impairment, also impact HRQOL [3‐5]. HRQOL of HNC patients deteriorates before, during, and shortly after treatment, and generally improves from 6 months after treatment [6‐12]. However, there is considerable variation between patients. Risk factors of poor HRQOL pertain to the disease and its treatment, and to personal, physical, psychological, social, and lifestyle factors [3‐12]. Biological factors related to aging (telomere), stress (cortisol), and inflammation (cytokines) may also be associated with HRQOL [13‐17]. Previous studies took only part of these factors into account [3‐12, 18].

The aim of this prospective cohort study was to estimate the relationship between the course of HRQOL in the first 2 years after diagnosis and treatment of HNC and personal, clinical, psychological, physical, social, lifestyle, HNC-related, and biological factors. The results are important to better understand which baseline factors (how HNC patient enter the cancer trajectory) and factors assessed at 6 months after treatment (how they overcome the acute phase of treatment) are associated with the subsequent course of HRQOL over time.

In this cohort of 638 HNC patients, there was a significant change of HRQOL over time with worse scores at baseline and 3 months after treatment. The absolute changes over time in QL (maximal improvement ≤ 7.8) were small, considering previous reported minimally important difference (MID) values of 8.64 on the EORTC QLQ-C30 QL scale among HNC patients [23]. There are no MID values available for SumSc but mean differences over time in the current cohort were also small (≤ 5.3 on a scale of 100). HRQOL of cancer patients in the Netherlands is generally higher compared to other countries [24]. Mean QL scores in this HNC cohort ranged from 71.7 to 79.5 and seem not to be different from mean scores of 77.4 and 77.9 as found in Dutch reference data [24, 25]. The same holds for SumSc: 86.8 to 88.9 in this HNC cohort versus 93.1 in a Dutch reference population [25]. It can be concluded that, in general, HRQOL of this HNC cohort was relatively good.

The baseline and post-treatment status of HNC patients were significantly associated with the course of HRQOL over time. HRQOL of patients who had more psychosocial problems (depressive symptoms, social contacts, social eating), oral pain or who were feeling more ill at baseline deteriorated relatively less and/or improved more from baseline to 24 months after treatment. In contrast, HRQOL of patients with a higher level of IL10, more stress, or more coughing at baseline deteriorated relatively more. As to post-treatment status, HNC patients with worse social contacts, and more speech, shoulder, and financial problems at 6 months after treatment, had a relatively larger improvement in HRQOL from 6 to 24 months after treatment, while patients with more stress and weight loss at 6 months after treatment had a relatively larger deterioration in HRQOL. It may be that HNC patients with more psychosocial and functional problems at baseline or post-treatment were referred to supportive care (including psychosocial care, pain medication, physiotherapy, speech therapy, and nutritional advice), which is effective to improve such problems and improve HRQOL [3‐5, 26‐28]. A poorer health condition due to coughing, stress, or weight loss may be related to underlying medical illness such as HNC recurrence, a second primary tumor (e.g., in the lungs), cachexia, or hypothyroidism [16, 29]. The role of the anti-inflammatory cytokine IL-10 (as measured at baseline) is interesting but also puzzling. Tumor and immune cells are sources of cytokines, which can lead to various symptoms negatively influencing HRQOL. Stress can corroborate the production of pro-inflammatory cytokines [14, 30]. However, IL-6, TNF-α, and CRP were not significantly associated with the course of HRQOL in the current study. It might be that the high level of IL-10 represents the remain of earlier inflammation before or at baseline, but more research is needed.

The strengths of this study are that we used the validated EORTC QLQ-C30 QL and SumSc which reflect overall HRQOL. To handle missing data at random which is common in longitudinal studies investigating HRQOL, we applied LMM analyses with maximum likelihood estimation which account for missing data. We did not account for missing data not at random. A limitation is the use of assumption of a compound symmetry covariance structure in the LMM analyses. Future researchers may explore other covariance structures. Another limitation of this study is that the NET-QUBIC cohort is not completely representative for the Dutch HNC population regarding age, sex, tumor subsite, and treatment modality [19, 20]. Retention rates were high at 2 years follow-up (80% among HNC patients alive) [20] but there was some selection in terms of tumor stage, physical performance, comorbidity and age, which might limit the representativeness of the results of this study. Another limitation is that the large number of variables may have induced co-incidental findings.

It is striking that so many variables were not significantly associated with the course of HRQOL over time, including none of the personal factors (age, sex, living status, level of education, personality, coping style, personal control, and self-efficacy), clinical (tumor stage, treatment modality, performance status, HPV, comorbidity), physical (daily living, blood pressure, heart rate, muscle strength, nutritional status), psychological (distress, anxiety, fear of cancer recurrence, fatigue, sleep, cognitive functioning adjustment to cancer), and lifestyle factors (smoking, alcohol use, BMI, physical activity). That does not mean that these factors are not important in HNC research, for example when developing prediction models or symptom clusters. It might be worthwhile, dependent on the research question, to stratify for HNC subsite in future research since (slightly) different courses of HRQOL were observed. Also, standardization of outcomes measurements is needed, preferably in consensus with (inter)national HNC organizations.

From a clinical point of view, understanding which factors impact an individual’s quality of life can help healthcare providers for instance by developing a risk assessment tool in clinical practice and targeted interventions. Also, it would be interesting to evaluate if changes in HRQoL over time are associated with patients’ need for supportive care.

Baseline clinical, psychological, social, lifestyle, HNC-related, and biological factors are associated with the course of HRQOL from baseline to 24 months after treatment. Post-treatment social, lifestyle, and HNC-related factors are associated with the course of HRQOL from 6 to 24 months after treatment.