Pre-transplant marital status and hematopoietic cell transplantation outcomes

Background Evidence about the impact of marital status before hematopoietic cell transplantation ( hct ) on outcomes after hct is conflicting. Methods We identified patients 40 years of age and older within the Center for International Blood and Marrow Transplant Research registry who underwent hct between January 2008 and December 2015. Marital status before hct was declared as one of: married or living with a partner, single (never married), separated or divorced, and widowed. We performed a multivariable analysis to determine the association of marital status with outcomes after hct . Results We identified 10,226 allogeneic and 5714 autologous hct cases with, respectively, a median follow-up of 37 months (range: 1–102 months) and 40 months (range: 1–106 months). No association between marital status and overall survival was observed in either the allogeneic ( p = 0.58) or autologous ( p = 0.17) setting. However, marital status was associated with grades 2–4 acute graft-versus-host disease ( g v hd ), p < 0.001, and chronic g v hd , p = 0.04. The risk of grades 2–4 acute g v hd was increased in separated compared with married patients [hazard ratio ( hr ): 1.13; 95% con-fidence interval ( ci ): 1.03 to 1.24], and single patients had a reduced risk of grades 2–4 acute g v hd ( hr : 0.87; 95% ci : 0.77 to 0.98). The risk of chronic g v hd was lower in widowed compared with married patients ( hr : 0.82; 95% ci : 0.67 to 0.99). between of caregiver commitment, and to assess health-related quality of life together with health care utilization.


INTRODUCTION
The outcomes of patients undergoing hematopoietic cell transplantation (hct) depend on a multitude of variables such as hct type, underlying disease, stability of the underlying disease, and patient sociodemographic variables 1,2 . Interest has been increasing in evaluating the potential impact of psychosocial variables-for example, marital status-on hct outcomes. The results of observational single-centre and registry studies evaluating the association between marital status and outcomes of hematologic malignancies, including hct outcomes, have been inconsistent [3][4][5][6][7][8][9][10] .
Marital status could be considered a surrogate for the presence of a caregiver or social support, where caregivers MARITAL STATUS AND HEMATOPOIETIC CELL TRANSPLANTATION OUTCOMES, Tay et al. are an important source of both instrumental and emotional support 11 . Moreover, in hct programs that advocate for outpatient-based programs, reliance on the hct recipient's social support systems-predominantly a partnered caregiver or spouse-has been increasing 11,12 . Indeed, a systematic review about the influence of social support on hct demonstrated an association of social support with hct outcomes, but the conclusions were limited by smaller studies and important covariates variably considered in hct survival analyses 11 . Additionally, studies in general oncology would further suggest that outcomes are better for married patients, with a larger benefit accruing to male patients 3 . A better understanding of how marital status contributes to hct outcomes would allow and advocate for a bolstering of supportive resources for the hct recipient.
We hypothesized that marital status is associated with improved outcomes after hct, such that patients who are married or living with a partner, compared with those who are not, will experience superior post-hct overall survival (os) in the autologous and allogeneic settings alike and superior acute and chronic graft-versus-host disease (gvhd) outcomes in the allogeneic setting. Further, we hypothesized that sex would mediate the foregoing potential associations. Using data from the Center for International Blood and Marrow Transplant Research (cibmtr), we examined the potential influence of marital status at the time of hct on os in the autologous and allogeneic settings, and on acute and chronic gvhd outcomes in the allogeneic setting.

Data Source
The cibmtr is an observational database that captures hct data from more than 420 hct centres worldwide. Data are submitted to a statistical centre at the Medical College of Wisconsin, Milwaukee, Wisconsin, U.S.A. Participating centres are required to report all hcts consecutively; patients are followed longitudinally, and compliance is monitored by on-site audits. Computerized checks for discrepancies, physician review of submitted data, and on-site audits of participating centres ensure data quality. Observational studies conducted by the cibmtr are performed in compliance with all applicable federal regulations pertaining to the protection of human research participants. Protected health information used in the performance of such research is collected and maintained in cibmtr's capacity as a public health authority under the U.S. Health Insurance Portability and Accountability Act of 1996 Privacy Rule.

Patient Population and Study Design
We identified patients 40 years of age and older who underwent either autologous or allogeneic hct for a hematologic malignancy between 1 January 2008 and 31 December 2015 (to best reflect current hct practices). Given the completeness of the U.S. data within the cibmtr hct registry, patients were exclusively from the United States. Additionally, we expected the autologous hct cohort to be smaller than the allogeneic cohort because cibmtr data collection for autologous hct does not necessarily include the marital status variable.

Exposures
Marital status was defined as one of married or living with a partner, single (never married), separated or divorced, and widowed. We selected patients 40 years of age and older to avoid confounding and to ensure balance between the 4 marital strata, given that younger age is associated with being single.

Outcomes and Definitions
Overall survival was defined as death from any cause. Surviving patients were censored at last follow-up; cases of acute and chronic gvhd were diagnosed and graded according to consensus criteria 13,14 .
Marital status is declared at the level of each participating centre at a single time point before hct (upon enrolment within the cibmtr database) as one of married or living with a partner, single (never married), separated or divorced, or widowed.

Statistical Analysis
Baseline patient, clinical, and sociodemographic variables were summarized using descriptive statistics. The primary outcome was os, with grades 2-4 acute gvhd and chronic gvhd being secondary outcomes. The os probabilities were estimated using Kaplan-Meier curves. To accommodate competing risks, the probabilities of acute and chronic gvhd were calculated using the cumulative incidence estimator. Multivariable analysis using Cox proportional hazards models was performed to determine the association of marital status with the primary and secondary outcomes, while adjusting for patient, clinical, and sociodemographic variables. In addition to marital status, the variables considered in the models included age, sex, race, performance status, education, smoking status, income and insurance status, distance to the hct centre, employment status, disease type, disease risk and status at hct, recipient cytomegalovirus serostatus, donor or graft type, donor-recipient sex and ABO match, conditioning regimen intensity, comorbidity index, gvhd prophylaxis for allogeneic hct recipients, and hct period. The assumption of proportional hazards was tested for each variable. A stepwise model-building approach was used to develop models for os and for acute and chronic gvhd. A p value less than 0.05 was considered statistically significant. The analysis was performed using the SAS software application (version 9: SAS Institute, Cary, NC, U.S.A.).
In general, we observed no appreciable differences in baseline patient, clinical, or sociodemographic variables for the patients in the 4 marital status categories. However, a few notable minor imbalances were evident, with widowed patients being more likely than non-married patients to be female, older, and retired.

Marital Status and Allogeneic HCT Outcomes
Based on the results of the multivariable analysis, os was not statistically different in the 4 marital status categories for patients receiving allogeneic hct (p = 0.58). Table ii summarizes the results of the regression analyses. When compared with patients who were married or living with a partner, those who were single (never married), separated or divorced, and widowed were not at increased risk of death (respectively, hr: 1.06; 95% ci: 0.95 to 1.17; hr: 0.99; 95% ci: 0.91 to 1.08; hr: 1.07; 95% ci: 0.92 to 1.24). Figure 1 shows the probabilities of os by marital status adjusted for age, performance status, hct comorbidity index, disease risk index, and other factors associated with mortality risk. The 5-year adjusted os probabilities were 37% (95% ci: 36% to 39%) for patients who were married or living with a partner and 39% (95% ci: 35% to 43%) for those who were single (never married), 39% (95% ci: 35% to 42%) for those who were separated or divorced, and 35% (95% ci: 29% to 42%) for those who were widowed.
In contrast, marital status was associated with grades 2-4 acute (p < 0.001) and chronic gvhd (p = 0.04). The risk of grades 2-4 acute gvhd was greater in patients who were separated or divorced compared with those who were married or living with a partner (hr: 1.13; 95% ci: 1.03 to 1.24; p = 0.01). However, the risk of grades 2-4 acute gvhd appeared to be lower for patients who were single (never married) than for those who were married or partnered (hr: 0.87; 95% ci: 0.77 to 0.98; p = 0.03). The risk of chronic gvhd was lower in patients who were widowed than in those who were married or living with a partner (hr: 0.82; 95% ci: 0.67 to 0.99; p = 0.03).
Table ii summarizes the multivariable analyses. Figures 2 and 3 show the probabilities of grades 2-4 acute gvhd and chronic gvhd by marital status, adjusted for disease, conditioning, employment, distance to the hct center, gvhd prophylaxis, and other factors associated with the development of gvhd. There was no interaction between marital status and sex.

Marital Status and Autologous HCT Outcomes
We observed no statistical difference in os between the 4 marital status categories for patients receiving autologous hct (Figure 4, p = 0.17). Table ii summarizes the analyses.
Compared with patients who were married or living with a partner, single (never married), separated or divorced, and widowed patients were not at an increased risk of death (respectively, hr: 1.10; 95% ci: 0.92 to 1.33; hr: 1.17;     Figure 4 shows the probabilities of os by marital status adjusted for age, performance status, hct comorbidity index, and disease risk index-the other factors associated with mortality risk. The 5-year adjusted survival probabilities were 65% (95% ci: 62% to 67%) for patients who were married or living with a partner and 62% (95% ci: 56% to 68%) for those who were single (never married), 59% (95% ci: 54% to 64%) for those who were separated or divorced, and 61% (95% ci: 53% to 69%) for those who were widowed. As for the allogeneic population, no interaction of marital status and sex with survival was evident.

DISCUSSION
Our study suggests that marital status is not associated with os after hct in either the allogeneic or autologous setting. However, marital status appears to influence the post-hct outcomes of grades 2-4 acute gvhd and chronic gvhd alike. In particular, the evidence demonstrates that, compared with patients undergoing allogeneic hct who are married or living with a partner, those who are separated or divorced are at a higher risk of acute gvhd (hr: 1.13; p = 0.008), and those who are single appears to be protected (hr: 0.87; p = 0.03). Likewise, being widowed appears to be protective against chronic gvhd (hr: 0.82; p = 0.03). We believe that those results are compelling given our multicentre data, the large sample size, and the inclusion of a comprehensive set of patient, disease, and psychosocial variables as covariates (Table i). Taken together, the evidence demonstrates a relationship between marital status and post-hct gvhd outcomes. It is difficult to reconcile the counterintuitive results in the allogeneic setting, where, compared with being married or living with a partner, being single is associated with less acute gvhd and being widowed is associated with less chronic gvhd. Given our large sample size of more than 10,000 recipients of allogeneic hct and the observed hrs close to 1, it is possible that some associations are statistically significant, but possibly not clinically meaningful. Moreover, it remains unclear how marital status might exert its effects. Marital status might influence hct outcomes through some combination of instrumental, emotional, or informational social support frameworks, where a partnered caregiver or the married state might be considered to be the optimal "intervention" that embraces all of those framework aspects 11 . Further, Foster et al. 15 suggest that "general" social support lacks "the interpersonal resonance and the interactive empathy characteristic of partnered relationships." Indeed, the quality of social support is associated with post-hct outcomes: Frick et al. 16 suggested that positive social support does not affect hct survival, but that the presence of problematic social support is associated with inferior survival. In contrast, Ehrlich et al. 17 recently suggested that pre-hct emotional support was significantly associated with better outcomes after allogeneic hct. Additionally, socioeconomic support has also been associated with superior hct outcomes 18,19 . Taking those data together, marital status might be an imperfect surrogate for social support, given that the persistence, quality, and strength of the marital relationship is not assessed, potentially explaining our incongruent results.
Is there a biologic basis or biomarker that might help in gaining insights? It has been suggested that behaviour within social relationships can modulate the responsivity of the immune system to stress and the depressive-reactive pathways, with depression potentially being a central pathway to immune dysfunction, leading to poor biophysical outcomes 20,21 . Further, spousal similarities noted in gene expression, immune profiles, and gut microbiota might offer additional insight into potential biologic or biomarker understandings within the larger construct of social support 20,22 . In the hct setting, the "conserved transcriptional response to adversity" (ctra) gene expression profile of cytokines in recipients of hct might be a potential stress biomarker that links socioeconomic status with post-hct biophysical outcomes 23,24 . Meaningful differences in ctra expression profiles between hct recipients of low and high socioeconomic groups has been demonstrated, with ctra expression being associated with upregulation of creb activity, inhibition of interferon response factor signalling, and desensitization of glucocorticoid receptor activity 25 . Untangling various aspects of socioeconomics (including social support and marital status) and its relative influence on ctra undoubtedly remains to be elucidated. However, is intriguing to ponder that both the quantity and quality of social support might lead to changes of the stress biomarker ctra in hct recipients, which might in turn influence the development of gvhd and disease relapse. Still, it is unclear how such potential biomarkers or surrogate markers for social support might influence post-hct outcomes or whether they are modifiable.
Other studies have examined marital status in the general oncology setting. For instance, data from the U.S. Surveillance, Epidemiology, and End Results program evaluating more than 1.2 million cases of cancer between 2004 and 2008 suggest that "married patients were less likely to present with metastatic disease (adjusted odds ratio [or], 0.83; 95% ci, 0.82 to 0.84; p < .001), more likely to receive definitive therapy (adjusted or, 1.53; 95% ci, 1.51 to 1.56; p < .001), and less likely to die as a result of their cancer after adjusting for demographics, stage, and treatment (adjusted hazard ratio, 0.80; 95% ci, 0.79 to 0.81; p < .001) than unmarried patients," where married men benefitted more than married women 3 . In contrast, data from studies evaluating individual malignancies have mixed results, with positive associations being found in patients with myeloma 9 , Hodgkin lymphoma 10 , and hematologic malignancies in general 8 , and no associations being noted in acute lymphoblastic leukemia 26 and historical studies 6,27 . Interestingly, a systematic review of eighteen studies assessing the influence of marital status and stage of cancer at diagnosis suggests that being unmarried increases the odds of having a later stage of breast cancer (odds ratio: 1.297; 95% ci: 1.035 to 1.627) or melanoma (odds ratio: 1.35; 95% ci: 1.16 to 1.57) at diagnosis 7 . To our knowledge, all reported studies in general oncology and specific malignancies have been based on U.S. Surveillance, Epidemiology, and End Results or state cancer registry data, with methodologic differences between the studies in how the data are analyzed and in the covariates considered or available for analysis.
In contrast to studies in general oncology and specific malignancies, two published studies have assessed marital status with respect to hct outcomes, and both demonstrated the lack of an association. Gerull et al. 5 examined 715 patients who received allogeneic hct between 2009 and February 2015 in the Swiss Transplant Cohort Study. The authors classified marital status as either single (encompassing single, divorced or separated, and widowed) or in a stable partnership. No differences in os, progression-free survival, non-relapse mortality, relapse, acute gvhd, or chronic gvhd were observed for the groups with and without a stable partnership. However, patients with missing information about their relationship status experienced significantly worse os and progression-free survival than did their counterparts whose records had that information. Similarly, Sato et al. 4 evaluated 309 Japanese patients who, between January 2000 and January 2017, underwent allogeneic hct and were classified as either married or unmarried. No differences in 5-year os, relapse, transplantationrelated mortality, and acute or chronic gvhd were observed between the married and unmarried recipients of allogeneic hct. Limited by small numbers, both studies variably considered important allogeneic hct covariates that might have influenced the results of their study. However, both studies suggest that, in the hct setting, other disease and hct factors remain highly integral to predicting post-hct outcomes, with marital status having unclear effects. In contrast, an abstract by Foley et al. 28 , reporting on data for 269 recipients of allogeneic hct from the University of California-San Francisco between January 2012 and January 2016, suggests that decreased os is associated with being divorced compared with being single or married (p = 0.025). Interestingly, a recent systematic review of recipients of solid-organ grafts suggested that neither social support nor marital status is predictive of medication adherence or post-transplantation outcomes 29 .

Limitations
Our study has limitations beyond the traditional biases associated with registry studies. First, it is possible that same-sex unions might not have been considered as married or living with a partner. Additionally, patients might be single (never married) but might still have children who act as caregivers and provide social support. Further, marital status was declared before the hct without further ascertainment of possible changes in marital status over the longer hct trajectory. Second, our data from the cibmtr reflects the U.S. environment, and it might not reflect circumstances in other geographic locations and cultures. However, our results would mirror the experience of both the Swiss and the Japanese cohorts of patients who received allogeneic hct, whose data suggested the lack of an association between marital status and hct survival outcomes. Third, an inherent selection bias might be present, given that hct centres might allow hct to proceed only in the presence of adequate social support, negating the potential influence of marital status. For instance, hct centres might assume that married patients have good social support, but might conduct a more rigorous assessment of social support for unmarried patients before proceeding with hct. Finally, caregiver burden has been recognized to potentially indirectly affect patient care and outcomes 30,31 . Unfortunately, data concerning caregiver or spousal burden, where the quality of caregiving might be affected by competing life circumstances such as work and young children, are unavailable. In the absence of additional data concerning social support, it is impossible to disentangle the overlapping concepts of marital status and social support.

CONCLUSIONS
We suggest that the influence of marital status on the outcomes of os (in both the autologous and allogeneic hct settings) and gvhd (in the allogeneic setting) are clinically negligible. Future research should consider measuring social support using validated scales such as those proposed by promis 32 or the patient and caregiver report of caregiver commitment, and should assess health-related quality of life together with health care utilization outcomes to better appreciate the potential effect of marital status and social support.