Impact of marital status on survival in patients with stage 1A NSCLC

Introduction

Socioeconomic factors have been actively investigated for their impact on cancer prognosis and potential to facilitate novel public health interventions. Among these factors, marital status has been found to be related to survival outcome in breast cancer [1], renal cell carcinoma [2], bladder urothelial carcinoma [3], esophageal cancer [4], rectal cancer [5] and prostate cancer [6]. However, the effect of marital status on survival of NSCLC patients remains equivocal: some researchers found no association between marital status and survival in NSCLC [7, 8], while others concluded that marital status affected overall survival (OS) of stage IV lung cancer [9] and being married was associated with lower risk of death [10].

Possible explanations of inconsistent findings in previous studies include varied sample size, age, social background, tumor malignancy and treatment models. Compared to patients with advanced cancer, Stage IA NSCLC patients usually have longer survival times and are less likely to undergo therapies other than surgical resection, e.g., chemotherapy and radiation, with debilitating side effects. Therefore, non-medical factors, such as marriage, may play more important roles in their prognosis.

Thus far, the effect of marital status on OS of stage IA NSCLC remains unclear. Meanwhile, there is neither literature comparing the effect of marital status on OS at different time periods nor in different age strata. Therefore, our primary aim was to study the impact of marital status on OS of stage IA NSCLCs patients of different ages and during different time periods in the SEER database.

Results

Impact of marital status on OS

Kaplan-Meier analysis and log-rank test revealed that impact of marital status on OS differs between the age-strata and may differ in different time period.

Regardless of age, the widows had the worst OS, while patients being MR/W.P had the highest OS (in group1, 2, p < 0.05; Figure 1A, 1B). Besides, OS of the widows was similar to the DV/SP after stratification (in group 1, 2; Figure 1C–1H).

Figure 1. Kaplan-Meier analysis and log-rank test revealed that impact of marital status on OS varies in different time period (A, B) and in different age-strata (C–H).

In patients younger than 60-years old, OS in the MA/W.P was statistically longer than other 3 marital statuses (group 1, 2; Table 3 and Figure 1C, 1F).

Table 3. Survival over marital status in different age-group within different time period.

		N	Median OS (Months)	Median age (Years)	Log-rank test for equality of survival function (p)
					DV/SP	MA/W.P	Single	Widowed
age < 60 years	group1
	DV/SP	695	121.0	54(50,57)		0.000	0.021	0.456
	MA/W.P	2598	191.0	54(50,57)	0.000		0.000	0.002
	Single	653	147.0	52(47,56)	0.021	0.000		0.443
	Widowed	176	128.0	56(54,58)	0.456	0.002	0.443
	group2
	DV/SP	1084	NR	55(52,58)		0.000	0.200	0.296
	MA/W.P	3754	NR	55(50,57)	0.000		0.000	0.036
	Single	1333	NR	54(49,57)	0.200	0.000		0.732
	Widowed	233	NR	57(53,58)	0.296	0.036	0.732
60 ≤ age < 75 years	group1
	DV/SP	1367	81.0	66(63,70)		0.000	0.738	0.685
	MA/W.P	6832	91.0	68(64,71)	0.000		0.005	0.000
	Single	868	79.0	67(63,70)	0.738	0.005		0.977
	Widowed	1819	84.0	70(66,72)	0.685	0.000	0.977	0.685
	group2
	DV/SP	2746	90.0	67(64,70)		0.001	0.108	0.721
	MA/W.P	11729	102.0	68(64,71)	0.001		0.471	0.000
	Single	2009	109.0	67(63,70)	0.108	0.471		0.045
	Widowed	2502	89.0	69(66,72)	0.721	0.000	0.045
age ≥ 75 years	group1
	DV/SP	324	57.0	78(76,80)		0.828	0.377	0.906
	MA/W.P	2840	59.0	78(76,80)	0.828		0.161	0.482
	Single	266	52.0	78(76,81)	0.377	0.161		0.272
	Widowed	1785	55.0	79(77,82)	0.906	0.482	0.272
	group2
	DV/SP	850	62.0	78(76,80)		0.061	0.063	0.384
	MA/W.P	5176	66.0	78(76,81)	0.061		0.513	0.118
	Single	555	72.0	79(76,82)	0.063	0.513		0.167
	Widowed	3013	62.0	79(77,83)	0.384	0.118	0.167
MR/W.P, married or had unmarried or domestic partner; DV/SP, divorced/separated.

From 60- to 75-year-old, OS of the MA/W.P was better than those with marriage loss (DV/SP, widows) in 1995-2005 and in 2006-2015. OS of the singles was shorter than MA/W.P in 2006-2015 (group 1), however, was improved in 2006-2015(group 2) and became similar to OS of the MA/W.P.

Starting at age 75 years, the OS was not distinct among older patients from different marital statuses (Table 3 and Figure 1E, 1H).

Discussion

Several studies have found that cancer patients in marriage are more likely to live longer [11]. Other researchers have reported greater benefits of marriage on conditional relative survival in cancer at early stages [12]. However, the association between marital status and OS of NSCLC patients remains controversial [8, 13–15]. Such controversy may partially arise from treatment complexity in advanced-stage cancer. In order to minimize confounding issues such as differing chemo- and radiotherapy regimens, we focused on stage IA NSCLCs, for which surgical resection alone is the only recommended treatment. As a result, the impact of marital status on OS can be explored while minimizing confounding medical factors.

In the last two decades, cancer screening methods, surgical strategies, population characteristics, social paradigms, and economic factors have evolved significantly in the United States. Hence, analysis was performed separately in two consecutive time periods to validate the result. Another important factor is age, which can be both a confounder and an effect modifier. Theoretically, people are more capable of recovering from marriage failure when they are young. Psychosocial interventions may have an important effect on survival and these interventions appear to be more effective in patients who are unmarried, older, attend Cognitive-Behavioral Therapy and have early-stage cancer 21. Since the World Health Organization categorizes age beyond 60-year-old as elderly, median age of the widows was 74-year old, patients beyond 74 had similar median age in different marital status, and using 60, 65 or 70 as boundary of stratified analysis would yield similar result (impact of marital status on OS), we decided to separate patients into three subgroups (age ≤60, 60 ≤ age < 75, age ≥ 75 years).

Although some of our findings corroborated previous publications, the following new findings were identified in our study.

First, we revealed a significantly prolonged OS during the last two decades except for the DV/SP patients: Median OS was prolonged significantly in single patients (from 89 to 109 months, p <0.001) and in widowed patients (from 70 months to 72months, p = 0.018). This can be explained mainly by progress in medical science. However, median OS of DV/SP patients did not show a statistically significant increase over the 20 years. Many factors, including lifestyle and weak social support systems, may contribute to this phenomenon. These socioeconomic factors [16] may impact cancer prognosis and even outweigh the survival benefits from medical progress. As the global population ages, further studies are needed to explain and address this problem.

Nevertheless, this is the first-time failed marriage (DV/SP) was shown to potentially offset the survival benefits from medical progress in the last decade in stage IA NSCLC patients, which can contribute to current literatures.

Second, this study not only verified the conclusion that being married is associated with lower relative risk of death in cancer patients [4, 7, 15, 17–19] but also validated protective effect of MA/W.P in cohort from different time periods and in different age group. We reported, for the first-time, possible protective role of being single (Table 4). which seems only present in patients ≥ 60 years and received diagnosis after 2005. As time goes by, populations of different marital status may have different characteristics. And how marital status affects cancer survival cloud also change. Therefore, further studies are required to understand the mechanisms of the evolving phenomenon and to provide updated reference to personalized care.

Finally, this is the first study revealing that the impact of marital status on patient survival varies significantly by age-group. Previous studies on the impact of widowhood on survival have had conflicting results. One study of Floridian patients with lung cancer revealed longer survival in married and widowed patients than in DV/SP and single patients [20]. Another Japanese study of 1,230 NSCLC patients demonstrated that widowed patients had the worst survival among married, separated, divorced and single male patients [14]. One study of 5,898 NSCLC patients from Mayo Clinic Lung Cancer Cohort found no statistically significant difference in survival among widowed, married, single and divorced patients, while subgroup analysis found stage IA widowed patients had a shorter survival [8]. Age difference may contribute to such controversy, for instance median age of the widows is 6-12 years elder than other marital status. In our study, after adjusting by age and other factors, being widowed and the DV/SP patients had similar OS shorter than the MA/W.P. One possible explanation is their social networks and may rely more on their families for support. The passing away of a loved one or marriage loss may lead to psychological trauma, less emotional and financial support.

Survival disparities based on marital status could also be attributed to the influence of family on the patient’s treatment decisions. Previous studies demonstrated that the family was involved in the decision-making process for most patients first-time diagnosed with lung cancer [21]. Unmarried patients, when diagnosed with staged IA NSCLC, more commonly refused surgery and came from lower socioeconomic backgrounds [22]. In addition, widowed and divorced patients preferred less aggressive cancer therapy [8] and were less likely to receive first-line surgical treatment [23].

This retrospective study has limitations: first, the potential for selection bias exists, as the sample population is mainly U.S based. Although robust data from other countries was not accessible, validation was performed using another group of patients with a different year of diagnosis. Moreover, the large sample size supports these findings as extrapolatable to the general population. Second, although both regression and stratification were used to control for confounding variables, unknown confounders may still exist, including comorbidities and other social variables, especially retirement. Patients were stratified by age older or younger than 60 years, which is a common age of retirement. However, retirement could not directly be controlled for, as this information was absent in the records.

In summary, this is the first age-stratified study about marital status and survival on early-stage NSCLC using historical cohort for validation. Being married or having an unmarried partner was associated with a significant lower relative risk. In contrast, marriage loss (widowed, DV/SP) were at a significantly higher risk of death. The progress of medical science in the preceding decades did not benefit survival of patients with marriage loss. Future studies could be designed with input from multidisciplinary specialists, e.g., psychologists and geriatric physicians, to further elucidate the underlying mechanisms and establish tangible interventions to address these survival disparities among individuals with different marital statuses.

Materials and Methods