Effects of marital status on overall and cancer-specific survival in laryngeal cancer patients: a population-based study

Marital status has long been recognized as an important prognostic factor for many cancers, however its’ prognostic effect for patients with laryngeal cancer has not been fully examined. We retrospectively analyzed 8834 laryngeal cancer patients in the Surveillance Epidemiology and End Results database from 2004 to 2010. Patients were divided into four groups: married, widowed, single, and divorced/separated. The difference in overall survival (OS) and cancer-specific survival (CSS) of the various marital subgroups were calculated using the Kaplan–Meier curve. Multivariate Cox regression analysis screened for independent prognostic factors. Propensity score matching (PSM) was also conducted to minimize selection bias. We included 8834 eligible patients (4817 married, 894 widowed, 1732 single and 1391 divorced/separated) with laryngeal cancer. The 5-year OS and CSS of married, widowed, single, and separated/divorced patients were examined. Univariate and multivariate analyses found marital status to be an independent predictor of survival. Subgroup survival analysis showed that the OS and CSS rates in widowed patients were always the lowest in the various American Joint Committee on Cancer stages, irrespective of sex. Widowed patients demonstrated worse OS and CSS in the 1:1 matched group analysis. Among patients with laryngeal cancer, widowed patients represented the highest-risk group, with the lowest OS and CSS.


Effect of marital status on cancer-specific survival (CSS) survival.
To explore the correlation between marital status and CSS, we performed Kaplan-Meier survival analysis on patients in the dataset. Figure 2B shows a significant difference in CSS among laryngeal cancer patients with various marital statuses. Male sex (P = 0.004), age < 65 years (P < 0.001), race (P < 0.001), grade I (P < 0.001), squamous cell carcinoma (P < 0.001), yes for surgery (P < 0.001), AJCC Stage I (P < 0.001), yes for radiotherapy (P = 0.001), no or unknown for chemotherapy (P = 0.001), and married state (P < 0.001) were associated with a higher 5-year CSS. (Table 3). To prevent possible interference between the variables, we used multivariate Cox regression analysis. As with the OS results, radiotherapy was not an independent predictor of CSS in patients with laryngeal cancer. In terms of marital status, married state remains a protective factor for the prognosis of laryngeal cancer.
Subgroup analysis by AJCC stage. We further evaluated the impact of marital status on OS and CSS of each AJCC stage. Interestingly, we obtained similar results in the various AJCC stage subgroups (Tables 4, 5, Fig. 3). First, marital status was an independent factor affecting the OS and CSS of each AJCC stage in univari- www.nature.com/scientificreports/ ate and multivariate analyses (P < 0.001). Second, the 5-year OS and CSS of patients in the widowed group were consistently lower than those of the other marital groups (Fig. 3). In terms of OS, among all AJCC stages, the survival rate of widowed patients in the AJCC stage II group was significantly lower than that of married patients (32.6% vs. 63.7%, P < 0.001; HR 2.06, 95% CI 1.70. 2.50, P < 0.001, Table 4). In the analysis of CSS, the most significant difference was between widowed patients and married patients in the AJCC stage III subgroup (32.3% vs. 58.6%, P < 0.001; HR 1.28, 95% CI 1.00.1.64, P = 0.042, Table 5). This phenomenon also occurred in the AJCC stage III subgroup. The risk of death was higher in the divorced/separated group than in the married group (HR 1.19; 95% CI, 0.97. 1.46), although the difference was not significant (P = 0.092, Table 5).
Subgroup analysis by sex. Subsequently Survival analysis in 1:1 matched group. To minimize the effect of possible confounding factors on the baseline features of the marriage subgroups and to verify the reliability of our results, we implemented a 1:1 matching cohort utilizing propensity score-matching (PSM) methods. We obtained three 1:1 matched cohorts, including a single and married cohort, a widowed and married cohort, and a divorced/separated and married cohort. The demographic and clinicopathological features of the matched cohort are presented in Table 8. As expected, the clinicopathological parameters were well balanced between the groups after PSM. Widowed patients showed worse OS and CSS in the divorce/separated-married cohort (Fig. 5A,B), the single-married cohort (Fig. 5C,D), and the widowed-married cohort (Fig. 5E,F).

Discussion
For the first time, a population analysis based on the SEER database was performed to assess the prognostic impact of marital status on the survival rate of patients with laryngeal cancer. This study found that marital status is an independent factor affecting the prognosis of laryngeal cancer. More specifically, married patients have the lowest risk of death, while widowed patients have the highest risk of death. This is like many previous research results [10][11][12] . We further confirmed after PSM that widowed patients had better OS and CSS than divorced, single, or married patients. In 1977, Engel proposed a new model of biological psychological medicine 13 . He believed that biological factors, psychological elements, and social factors influenced disease progression and outcome 14 . Since then, extensive research has been conducted on the relationship between biological psychological factors and disease [15][16][17] . The role of biopsychosocial factors in cancer patients has also gradually gained attention 18,19 . A study of women's marital status and mortality rates showed that single patients had higher mortality rates than divorced or widowed patients 20 . Another large-scale survey found that married patients with oral and laryngeal cancer are less likely to have metastases 21 . A Swedish study found that divorce and bereavement are risk factors for esophageal and gastric cancer 22 . The relationship between marital status and prognosis may be influenced by tumor stage, proportion of patients receiving treatment, and social support [23][24][25][26] . A higher percentage of married laryngeal cancer patients receive timely treatment, including surgery and adjuvant treatment, which may explain the high survival rate. However, it emphasizes the interrelationship between marital status and survival rather than causality. Therefore, it is necessary to explore how marital status affects the potential mechanism of survival to improve the outcome of patients with laryngeal cancer.
This study showed that marital status was associated with survival in patients with laryngeal cancer. We hypothesize the following reasons for the beneficial effect of married state on survival in patients with laryngeal cancer. First, a happy marriage may result in a well-balanced emotional state, and a wholesome family environment may ease work and social pressures. Second, married patients that have stable marriages usually are accompanied by appropriate family finances. A commonly accepted explanation of why married people have lower cancer mortality, was that it was related to better socioeconomic status. This was believed to buffer the effects of stressful events 27,28 . Chronic stress may cause long-term secretion of cortisol 29 , which leads to reverse regulation www.nature.com/scientificreports/ of leukocytes by down-regulating the cortisol receptor of leukocytes. This downregulation, in turn, reduces the ability of cells to respond to anti-inflammatory signals and leads to the vigorous development of cytokinemediated inflammatory processes 30 , which has been proven to be a poor prognostic factor for cancer 31,32 . Third, married patients have a wider social range than unmarried patients. They have a broader information base regarding medical equipment, experts, and treatment methods. This can help improve treatment outcomes 33 . Social networks influence patient compliance, and good compliance ultimately affects a patients' health outcomes 34 .
In addition, our results raise another intriguing question as to why widowed patients exhibit worse clinical outcomes. There is evidence to suggest that a widows' health is a problem before they are diagnosed with cancer 35 . Studies have shown that the recent death of a husband results in a significant decrease in the level of  38 . Compared with married patients, widowed patients had more psychological stress and less psychological support. This can cause disorders of the immune system and promote cancer progression 39 . Such an alteration affects the release of glucocorticoids and catecholamines, further affecting the tumor microenvironment 40,41 . An enhanced development of tumors results in a shortened survival time for widowed patients. Although this study is both instructive and relevant to clinical practice, it has some limitations. First, the marital status information provided in the SEER database is incomplete. It provides only marital status at the time of diagnosis, and some patients' marital status may change during follow-up. In the same way, it does not www.nature.com/scientificreports/ reflect whether marital status changes after diagnosis, which may also affect survival outcomes. In addition, patients aged 65 and above account for approximately 44% of the total. This means that it is possible to change from a married status to a widowed status during the follow-up period. Second, in addition to marital status, various other social factors are included in the bio-psycho-social medical model, such as education, income, and insurance. The SEER database lacks information regarding other societal factors. Our analysis cannot correct these factors, which may affect the survival outcomes. Third, the SEER database does not provide information on whether married patients are happy or not. Even in an equivalent married group of patients, the happiness of marriage may be questionable. For instance, prolonged marital discord can damage the immune system and have a negative effect on health 42 . Fourth, personal history such as smoking, drinking, and human papilloma virus (HPV) infections is not reflected in the SEER database. Some studies have shown that unmarried people may be more likely to engage in the bad habits of smoking and/or drinking 43,44 . Fifth, some patients may live with a partner without being married, despite the fact that the proportion of people in this circumstance may be small. In addition, we classify these patients as single patients, however their survival outcomes are expected to be better than those of other single people, which confounds our results. Sixth, the SEER database does not contain the number of patients with alternate sexual orientation. Finally, the AJCC stage provided in the SEER database is the sixth edition, and now the eighth edition is used in the clinic. The subgroup results of our AJCC stage need to be verified in further clinical practice. Despite these limitations, our findings are credible for the following reasons. First, our data is from the SEER database, which is composed of data from multiple centers across the country. Second, we chose the two outcomes of OS and CSS, which adds to the accuracy of our results. Last but not least, we also performed a stratified analysis on the AJCC stage and obtained high survival rates for married patients and low survival rates for the widowed patients. Due to a lack of social and psychological support, widowed patients should be advised to obtain psychological therapy during cancer treatment to improve their prognosis.

Conclusion
Marital status was associated with survival (OS and CSS) in patients with laryngeal cancer. It is clear that married patients had a better survival outcome, while widowed patients had a worse prognosis. If we are able to fully understand the impact of marital status on cancer patients, we will be able to provide individualized treatment and improve survival outcomes.  Description of covariates. The study variables included the following: sex (male, female), age (< 65 or ≥ 65 years), race (White, Black, or other), Grade (Grade I, Grade II ,Grade III/IV), histological type (squamous cell carcinoma, other), surgery (yes or no), AJCC stage (I, II, III, IV), radiotherapy (yes or no/unknown), chemotherapy (yes or no/unknown), marital status (married, widowed, single, divorced/separated). The laryngeal specific surgical code 10-90 was defined as invasive treatment and code 00 as non-surgical treatment. The primary outcome indicators were laryngeal cancer OS and CSS.

Statistical analysis. The clinical and pathological characteristics of patients with various marital statuses
were compared using the chi-square test (χ 2 ). Kaplan-Meier estimates and log-rank tests were used to estimate and compare survival functions between the different variables. Multivariate Cox proportional risk regression was used to estimate the risk ratio (HR) and 95% confidence interval (CI) for OS and CSS in the various marital statuses. To reduce potential baseline confounding factors, we employed propensity score matching (PSM) to reexamine the impact of marital status 45 . One-to-one PSM was conducted using the nearest-neighbor algorithm with a caliper width of 0.2. Data were analyzed using SPSS version 25 (IBM Statistics, New York, NY), and the survival curves were generated by R project version 3.6.1. P < 0.05 (2-sided) was considered significant.

Data availability
The data used in this study were provided by the SEER database.