The association between depression and esophageal cancer in China: a multicentre population-based study

Background Esophageal cancer (EC) is one of the leading contributors to the global burden of cancer, and the underlying mechanism is still unknown. Recently, there has been a growing interest in understanding modifiable psychosocial risk factors, particularly depression, to prevent EC and reduce morbidity and mortality. However, related research is sparse and has been ignored. The study was designed to assess the association between depression and EC in China. Methods From 2017 to 2019, a population-based multicenter study was conducted in high-risk regions of EC. Participants underwent a free endoscopy screening. If the endoscopic results were suspicious, a pathological biopsy was applied to confirm. Depression was measured with Patient Health Questionnaire-9 (PHQ-9). In addition, information on demographic characteristics and risk factors was collected from participants by trained interviewers using uniform questionnaires. Results After Endoscopy and pathologic diagnosis, 15,936 participants in high-risk regions of EC (ECHRRs) were enrolled, 10,907 (68.44%) of which were diagnosed health, 4048 (25.40%) with esophagitis, 769 (4.83%) with low-grade intraepithelial neoplasia (LGIN), 157 (0.99%) with high-grade intraepithelial neoplasia (HGIN), and 55 (0.35%) with EC, respectively. The overall prevalence of depression symptoms of participants was 4.16% (health: 4.63%, esophagitis: 2.99%, LGIN: 2.99%, HGIN: 5.73%, and EC: 9.09%). Multiple logistic regression analyses revealed that the unadjusted OR (95% CI) between depression and each esophageal pathology grades were esophagitis 0.93 (0.92-0.95), LGIN 0.97 (0.94-0.99), HGIN 1.05 (1.00-1.10), and EC 1.04 (0.97-1.14), respectively. However, after adjustment for potential confounders (age, gender, region, alcohol consumption, BMI), no statistically significant associations between depression and EC (adjusted OR = 1.10, 0.99-1.21) and esophageal lesions (esophagitis: adjusted OR = 1.02, 0.99-1.04; LGIN: adjusted OR = 0.98, 0.95-1.01; HGIN: adjusted OR = 1.04, 0.98-1.11) were observed in this study. Conclusions No significant association was observed between depression and EC in the study. Future prospective cohort studies are needed to verify this preliminary finding.


Background
Esophageal cancer (EC) is one of the most prevalent malignancies with high mortality and increasing incidence [1,2]. In 2015, the incidence and mortality of EC in China were 17.9 per 100,000 and 13.7 per 100,000 [3,4]. One-half of new cases occur in China, imposing a heavy economic burden and mental stress on families and society [3][4][5]. Like most common malignant tumors, EC is a complex disease with multifactorial etiology. Both genetic and environmental factors influence the risk of developing the disease [6][7][8][9][10]. In the past, most etiological studies of EC focused on biology, and social or psychological factors were easily ignored [11][12][13][14].
Growing evidence has shown that depression may exert an etiologic role in cancer [15][16][17][18][19]. A recent meta-analysis of 51 prospective studies showed that depression and anxiety disorders could cause a significant 13% increase in cancer risk and a 21% increase in cancer-specific mortality [20]. Evidence associated with depression and cancer indicated an increased cancer risk in individuals with depression [21][22][23]. However, many previous studies on depression and cancer have primarily focused on breast, lung, colorectal cancers [21,22]. Only a few studies have explored the relationship between depression and ECspecific risk. A meta-analysis reviewing depression and anxiety concerning cancer incidence and mortality covered 21 common tumors [20], but a recent meta-analysis found that only one study concerned depression and the risk of esophageal cancer [24]. The evidence of which was still insufficient. Therefore, a population-based, multicenter study was implemented to estimate the status of depression in high-risk regions of the EC (ECHRRs) and evaluate the association between depression and EC, aimed to provide clues for preliminary screening and prevention of EC and fill the gap in this field.

Study design
The national cohort of esophageal cancer (NCEC) is a multi-center prospective cohort study of EC and precancerous lesions based on high-risk populations in China [25]. Details on the cohort have already been published [25]. This study is based on the NCEC cohort and provided free gastroscopy screening services for upper gastrointestinal tumors to residents aged 40 to 69 years in five ECHRRs (Linzhou, Henan; Cixian, Hebei; Feicheng, Shandong; Yangzhong, Jiangsu; Yanting, Sichuan) from May 2017 to November 2019 (Fig. 1). All participants were recruited and interviewed face-to-face by trained staff. A uniform questionnaire was used to collect their basic information, including living and eating habits, disease history, family tumor history, and other exposure factors. Then, eligible participants received an upper gastrointestinal endoscopy examination. If an endoscopy detects suspicious esophageal lesions, esophageal pathology would be applied to confirm clinical health, esophagitis, low-grade intraepithelial neoplasia (LGIN), high-grade intraepithelial neoplasia (HGIN), and EC. The details related to the design of the NCEC are described on the website (http://www.ncec-China.cn/ cmmct.html) and elsewhere [25].

Study participants
The inclusion criteria included: (a) residents aged 40-69 years old; (b) no severe vision or hearing problems; (c) be able to comprehend the survey process properly; and (d) being competent to provide written informed consent. The exclusion criteria included: (a) previous diagnosis of EC or other cancers and (b) being contraindications for endoscopic examinations (e.g., acute perforation of the upper digestive tract, severe heart, lung, kidney, brain dysfunction, and multi-organ failure).

Instruments
Before the esophageal endoscopy examination, all depression surveys were conducted using the Chinese version of the Patient Health Questionnaire-9 (PHQ-9). The PHQ-9 is a self-reported questionnaire assessing the presence and severity of depression with good test-retest reliability and good validity in primary medical care and clinical practice [26,27].
PHQ-9 consists of 9 items (anhedonia, depressed mood, sleep problems, fatigue, weight/appetite change, feelings of worthlessness/guilt, poor concentration, psychomotor retardation/agitation, thoughts of self-harm/ suicidal ideations) that are based directly on the nine diagnostic criteria for major depression in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-V) [28]. PHQ-9 has 4 response scores representing the frequency of each depressive symptom in the past 2 weeks (0 = "not at all", 1 = "several days", 2 = "more than half the days", and 3 = "nearly every day"). The depression score was the total of each item from 0 to 27. A higher depression score indicates severe depression. Depression score is also used as a binary variable: cutoff score at 7, meaning 7 or above is depressive [29].

Covariate variables
Based on the results of face-to-face interviews and a comprehensive literature review. The following covariates were included in the study: age (≤50y, 51-60y, >60y), gender, region, alcohol consumption [30,31], and body mass index (BMI) [32][33][34][35][36]. Alcohol consumption was used to evaluate the frequency of drinking in the past year, divided into never (or almost never), only occasionally, most days (or almost every day). Additionally, both weight and height were measured on the endoscopy examination day, BMI was calculated as the weight in kilograms divided by the square of the height in meters (kg/m 2 ), and classified according to the World Health Organization (WHO) guidelines: underweight: < 18.5, normal ≥18.5 and < 25, overweight ≥25 and < 30, and obese ≥30.

Statistics
SAS version is 9.4 statistical programs (SAS Institute, Cary, NC) was used for data management and analyses. The continuous and qualitative variables were summarized as mean ± standard deviation (SD) or median with the first and third quartile (Q1-Q3) and frequency (%), respectively. We used the Student's t-test for normally distributed variables and Kruskal-Wallis for non-parametric variables to analyze the continuous data. Categorical variables were compared using Chi-square or Fisher's exact test. Multiple linear regression models (stepwise) were performed to test for correlation between covariates before covariate modeling. Collinearity diagnostics using the variance inflation factor (VIF) showed no evidence of collinearity among covariates (VIF < 10.0) [37]. Finally, we performed multivariable-adjusted logistic regressions to determine the association between depression (PHQ-9 scores) and different esophageal lesions. Specifically, model 1 was non-adjusted that estimated the raw contribution of depression to the prediction of esophageal lesions; model 2 was adjusted for age, gender, and region; model 3 was adjusted for age, gender, region, alcohol, and BMI. All tests of significance were two-tailed, and P < 0.05 was examined statistically significant.

Ethics statement
The study was performed in accordance with the institutional research ethics guidelines and the Helsinki declaration. Formal ethics approval was granted by the Ethics Committee of Cancer Institute and Hospital, Chinese Academy of Medical Sciences (No.16-171/1250). All the participants were informed about the purpose of the study and provided written informed consent.

Baseline characteristics of participants with different esophageal lesions
The baseline characteristics of participants are shown in Table 1. The mean age of all participants was (55.44 ± 7.74) years. Healthy people are younger than others (P < 0.001). Over half of the participants (58.56%) were women, while the HGIN and EC group included more males patients (50.32, 56.36%). Moreover, the differences in baseline characteristics between the healthy group and the other esophageal lesions groups were statistically significant in terms of region, marital status, highest education level, occupation, household income, smoking status, alcohol and tea consumption, physical activities, life satisfaction status, self-rated health status, and BMI (all P < 0.05).
Multiple linear regression analysis of depression (PHQ-9 scores) and baseline characteristics Table 3 shows the regression equation results by multiple linear regression analysis (forward stepwise selection method). Depression (PHQ-9 score as dependent variable) is related to occupation, life satisfaction status, region, household income, physical activities, self-rated health status, tea consumption, BMI, alcohol consumption, marital status, highest education level. The results of collinearity diagnostics showed that all VIF values were below 10, which indicated that no severe multicollinearities exist between the independent variables in this study.

Discussion
With the transition from a biological medical model to a biopsychosocial model, the impact of psychosocial factors on cancer progression has attracted much attention. This study is the first to focus on depression symptoms and esophageal lesions in a large-scale multi-center population on a global scale, which filled the gap in this field. Findings from this population-based study supported the view that the prevalence of depression in ECHRRs was high, especially for EC. Nevertheless, we found no evidence of an association between depression and the risk of either esophageal lesions or EC.
Depression is among the most prevalent and disabling psychological disorders worldwide and affects 350 million people [38][39][40]. Evidence from the China Kadoorie Biobank (CKB) study of 0.5 million adults indicated that depression could not be ignored in China [41]. The latest National Health Survey (NHS) in 2019 reported that the lifetime prevalence of depressive disorders was 6.8% in China [42]. In our study, the overall prevalence of depression in ECHRRs was 4.16%, participants with EC was 9.09%, which was higher than the national level. The possible explanation is that EC is one of the most human malignant tumors, with high mortality and poor survival [2,3], threatening the health of people living in ECHRRs. Residents have a higher risk of EC and suffer higher stress and anxiety, prioritizing those under limited psychological health resources.
Interestingly, the prevalence of depression was lower in patients with esophagitis and LGNI than health in our study. A possible reason was that the study was carried out in screening populations instead of clinical medical records, so most screening participants were healthy people. It indicated that the healthy people had strong health awareness on cancer screening, paid much attention to their body, and even worry about their health, which may lead to a high level of depressive disorders. The second explanation was that patients with  esophagitis and LGIN have almost no physical discomfort and do not need special treatment and surgery, which would bring little psychological burden. Besides, patients with esophagitis and LGIN maybe feel lucky and glad that they do not have cancer.
Our study found no evidence of an association between depression and the risk of either esophageal lesions or EC. The following factors may explain the observed non-significant association. 1) There is a lack of solid evidence of a positive association between cancer and depression, and the existing epidemiological studies have yielded conflicting results. Several metaanalyses and systematic reviews have been published on the topic but have reported mixed results. More recently, a meta-analysis published in 2007 suggested a small and modestly significant relationship between depression and the risk of cancer incidence (RR 1.13; 95% CI:1.06-1.19) [20]. However, a meta-analysis found that clinically diagnosed depressive disorder people do not have an elevated risk for cancer incidence (OR 1.15; 95% CI:0.85-1.56) [43]. So far, the research on the relationship between esophageal cancer and depression is sparse, and our findings filled the gap in the field. 2) Although the total screening participants is enough, the HGNI and EC from the screening population were relatively more minor, which means that related results carry uncertainty to some extent and need to be interpreted cautiously. 3) We tried to put many covariates in modeling regression before, and the results showed a positive association between depression and esophageal cancer. On second consideration, considering that the smallest group only includes 55 people and the possibility of overfitting, we made a priori selection for the current confounding factors, and the results turned out to change from positive correlation to non-significant correlation. Despite our negative results between depression and EC-related diseases, depression must not be ignored because increasing evidence found that depression may be influencing the progression of cancer [15,16,44,45]. Reverse causation may exist between depression and EC because cancer diagnosis could influence the mental health status or rise to depression [46]. Unlike most previous studies, our research evaluated the association of depression and EC in a population and excluded people with a cancer history, minimizing the influence of reverse causality. Furthermore, the results indicated that distress symptoms alone appear to be relatively less harmful to cancer development [43]. Considering that most human cancers have a long latency period and are difficult to detect during the early stages of cancer, an association between depression and EC was not identified in this cross-sectional study. In addition, lifestyle and behavioral changes may influence the association between depression and cancer indirectly. People with depression are more likely to have unhealthy lifestyle habits [47]. For instance, according to Watts, most individuals with depressive disorders abuse alcohol in search of disinhibition or reduce emotional and behavioral symptoms of depression [48]. Obesity and BMI were associated with depression [49,50]. In this cross-sectional study, the ratio of depression differed according to regions, education level, occupation, alcohol drinking habit, and BMI level. However, risk factors of depression and EC are in part shared, further complicating causal interpretations. Among those, socioeconomic status (SES), education level may play mediating roles in the  association between depression and EC. Intermediary factor analysis will be taken into account in further study. Several limitations to this pilot study have to be acknowledged. First, due to the study's cross-sectional design, causal inferences cannot be shown, and the long-term effects of depression on EC progression are also unavailable. Second, volunteer bias may exist. Residents were willing to participate in the endoscopy program actively because of free. Third, even though we try to control the confounding factors as much as possible, the smallest group only included 55 persons; we do not have the correct number of participants to include more covariables. Therefore, we cannot control the confounding factors completely. Finally, the results in ECHRRs may not be generalization to the general population, which should be interpreted with caution.

Conclusion
Our study took the lead in investigating the association between depression and EC in China. Findings from this population-based study supported the view that the prevalence of depression in ECHRRs was high. There is no clear evidence that depression may be a contributing factor to EC and precancerous lesions. The results should be interpreted with caution.

Implications
Depression, causing a significant psychological burden, has long been underestimated seriously worldwide. Confronted with a lack of awareness of the psychological health of Chinese people, and there is a considerable gap in psychological services between China and developed counties. The government in China is actively promoting improving residents' mental health literacy to 30% by 2030. The priorities of psychological health resources should be provided to high-risk populations, such as residents in ECHRRs and people screened to HGIN and EC.