Marital Status and Overall Survival in Patients with Resectable Pancreatic Cancer: Results of an Ancillary Analysis of NRG Oncology/RTOG 9704

Abstract

Background

Several registry‐based analyses suggested a survival advantage for married versus single patients with pancreatic cancer. The mechanisms underlying the association of marital status and survival are likely multiple and complex and, therefore, may be obscured in analyses generated from large population‐based databases. The goal of this research was to characterize this potential association of marital status with outcomes in patients with resected pancreatic cancer who underwent combined modality adjuvant therapy on a prospective clinical trial.

Materials and Methods

This is an ancillary analysis of 367 patients with known marital status treated on NRG Oncology/RTOG 97‐04. Survival analysis was performed using the Kaplan‐Meier method and compared using the log‐rank test. Multivariate analysis was performed using the Cox proportional hazards regression model.

Results

Of 367 patients, 271 (74%) were married or partnered and 96 (26%) were single. Married or partnered patients were more likely to be male. There was no association between marital status and overall survival (OS) or disease‐free survival (DFS) on univariate (hazard ratio [HR], 1.09 and 1.01, respectively) or multivariate analyses (HR, 1.05 and 0.98, respectively). Married or partnered male patients did not have improved survival compared with female or single patients.

Conclusion

Ancillary analysis of data from NRG Oncology/RTOG 97‐04 demonstrated no association between marital and/or partner status and OS or DFS in patients with resected pancreatic cancer who received adjuvant postoperative chemotherapy followed by concurrent external beam radiation therapy and chemotherapy. Clinical trial identification number. NCT00003216.

Implications for Practice

Several population‐based studies have shown an epidemiological link between marital status and survival in patients with pancreatic cancer. A better understanding of this association could offer an opportunity to improve outcomes through psychosocial interventions designed to mitigate the negative effects of not being married. Based on the results of this analysis, patients who have undergone a resection and are receiving adjuvant therapy on a clinical trial are unlikely to benefit from such interventions. Further efforts to study the association between marital status and survival should be focused on less selected subgroups of patients with pancreatic cancer.

Short abstract

Evidence suggests a survival advantage for married compared with single patients with pancreatic cancer. This article examines the effect of marital status on survival for this cohort of patients with resected pancreatic cancer, based on data from NRG Oncology/RTOG 97‐04.

Background

Pancreatic cancer is the fourth leading cause of cancer death in the U.S., warranting continued efforts to identify and modulate factors that may contribute to survival 1. Population‐based studies have reported improved survival in married patients with pancreatic cancer 2, 3, 4, 5. Later stage at presentation and poor adherence to treatment recommendations among single or unmarried patients have been suggested as some of the potential reasons for this difference 4. However, inherent limitations of registry‐based studies, such as lack of detailed potential confounder information and variation in data collection, prohibit a more in‐depth analysis of this association 6, 7. Patients treated on protocol NRG Oncology/RTOG 97‐04, a phase III clinical trial of postoperative adjuvant chemotherapy and chemoradiation therapy for patients with resected pancreatic carcinoma, represent a defined group of patients with localized disease who underwent a surgical resection and received protocol‐dictated adjuvant therapy 8. Therefore, an ancillary analysis was performed of NRG/RTOG 97‐04 to examine the effect of marital and/or partner status on survival for this cohort of patients.

Materials and Methods

Study Population

Between July 20, 1998, and July 26, 2002, 538 patients were randomized on RTOG 97‐04. Eligibility included histologically confirmed adenocarcinoma of the pancreas and gross tumor resection, confirmed by central review, with no evidence of progression or persistent disease on postoperative imaging and stage T1‐4, N0‐1, M0 according to 1997 staging criteria. Of 538 patients enrolled, 451 patients were eligible for analysis as previously described 8, with known marital status in 367.

Statistical Analysis

In this trial, for marital status, patients chose from three options: married or other live‐in relationship; single, divorced, separated, or widowed; or prefers not to answer. In this analysis, marital status was treated as a dichotomized variable: married or other live‐in relationship (married or partnered) versus single, divorced, separated, or widowed (single). Patients who chose “prefers not to answer” were excluded from this analysis. Overall survival (OS) and disease‐free survival (DFS) were estimated using univariate analysis with the Kaplan‐Meier method 9, and marital status was compared using the log‐rank test. Cox proportional hazards models 10 were used to identify any associations of marital status or marital status and sex with OS and DFS. Variables included in the multivariable models to further evaluate the association of marital status with survival included the following: treatment arm (radiation therapy [RT] + 5‐fluorouracil vs. RT + gemcitabine), age, sex (male vs. female), race (white vs. other), tumor location (head vs. everything else), Karnofsky Performance Scale (KPS; 60–80 vs. 90–100), CA19‐9 (<90 vs. ≥90), margin status (negative vs. positive vs. unknown), tumor diameter (<3 cm vs. ≥3 cm), and any comorbidity (yes vs. no) or hypertension and diabetes grouping (neither vs. diabetes only vs. hypertension only vs. both). Whereas patient marital status was the factor of primary interest in the models, the independent effects of the other preselected factors also were evaluated. Statistical analyses were performed using SAS statistical software (version 9.4, SAS Institute Inc, Cary, NC).

Results

Patient Characteristics

There were 367 patients with known marital status enrolled on RTOG 97‐04 and eligible for analysis, including 271 married or partnered patients (74%) and 96 single patients (26%). At the time of the analysis, 298 (81%) of the patients had died, including 215 married or partnered patients (215/271 = 79%) and 83 single patients (83/96 = 86%). The median follow‐up was 1.53 and 6.99 years for all and for surviving married and partnered patients, respectively, and 1.44 and 7.17 years for all and for surviving single patients, respectively. Median age of subjects was 62 years (range, 33–82), and 201 (55%) were male. Married or partnered patients were more likely to be male (63% vs. 31%, p < .0001), with a trend to being white (90% vs. 83%, p = .079; Table 1). Although the groups did not statistically significantly differ with regard to number of patients with comorbid conditions, married or partnered patients were less likely to have hypertension (29% vs. 41%, p = .032) and diabetes (22% vs. 32%, p = .04). With regard to available socioeconomic factors, married or partnered patients were more likely to have postsecondary education, be retired by the time of diagnosis, and be less likely to be not working because of other reasons (supplemental online Table 1). Otherwise, there were no significant differences in pretreatment characteristics between the groups.

Table 1.

Patient characteristics

Characteristics	Married or partnered (n = 271)	Single (n = 96)	p valuea
Age, median (min–max)	62 (35–82)	62 (33–80)
Sex			<.0001
Male	171 (63.1)	30 (31.3)
Female	100 (36.9)	66 (68.8)
Race, n (%)			.079
White	244 (90.0)	80 (83.3)
Other	27 (10.0)	16 (16.7)
Pretreatment CA19‐9, n (%)	234	76	.53
<90b	203 (86.8)	68 (89.5)
≥90	31 (13.2)	8 (10.5)
Primary location, n (%)			.96
Head	232 (85.6)	82 (85.4)
Everything else	39 (14.4)	14 (14.6)
KPS, n (%)			.80
60–80	100 (36.9)	34 (35.4)
90–100	171 (63.1)	62 (64.6)
Largest tumor dimension of primary, n (%)			.90
<3 cm	115 (42.4)	40 (41.7)
≥3 cm	156 (57.6)	56 (58.3)
Comorbidity, n (%)
Any	152 (56.1)	63 (65.6)	.10
Cardiovascular	44 (16.2)	16 (16.7)	.92
Cirrhosis	2 (0.7)	0 (0.0)	1.00
Diabetes	59 (21.8)	31 (32.3)	.04
Hypertension	78 (28.8)	39 (40.6)	.032
Pancreatitis	8 (3.0)	2 (2.1)	1.00
Pulmonary	11 (4.1)	3 (3.1)	1.00
Renal	3 (1.1)	0 (0.0)	.57
Neither diabetes nor hypertension	160 (59.0)	46 (47.9)	.033
Diabetes only	33 (12.2)	11 (11.5)
Hypertension only	52 (19.2)	19 (19.8)
Both diabetes and hypertension	26 (9.6)	20 (20.8)
RX, n (%)			.35
RT + 5‐FU	145 (53.5)	46 (47.9)
RT + gemcitabine	126 (46.5)	50 (52.1)

^aIncludes Lewis antigen negative.

^b p value from chi‐square or Fisher's exact test.

Abbreviations: 5‐FU, 5‐fluorouracil; CA19‐9,; KPS, Karnofsky Performance Scale; RT, radiation therapy.

Missing Data

There were no statistically significant differences in baseline characteristics between the patients who were eligible and analyzable for marital status (n = 367) and those who were not analyzable because of unknown marital status (n = 84). There were no statistically significant differences in OS or DFS between those who were analyzable and those who were not analyzable (data not shown).

Marital Status and Survival

Median OS was 1.54 years (95% confidence interval [CI], 1.35–1.75) for married or partnered patients and 1.44 years (95% CI, 1.26–1.94) for single patients (Fig. 1). The 2‐ and 5‐year OS estimates were 39% (95% CI, 33–44) and 23% (95% CI, 18–28) for married or partnered patients and 39% (95% CI, 28–48) and 18% (95% CI, 11–27) for single patients, respectively (log‐rank p value = .52; hazard ratio [HR], 1.09; 95% CI, 0.84–1.40; Fig. 1). Median DFS was 0.78 years (95% CI, 0.72–0.90) for married or partnered patients and 0.81 years (95% CI, 0.62–1.17) for single patients. The 2‐ and 5‐year DFS estimates were 22% (95% CI, 18–27) and 13% (95% CI, 10–18) for married or partnered patients and 25% (95% CI, 17–34) and 14% (95% CI, 8–21) for single patients, respectively (log‐rank p value = .97; HR, 1.00; 95% CI, 0.79–1.28; Fig. 1). On Cox multivariable analysis, there were no statistically significant differences in OS or DFS whether CA19‐9 was included in the model (Table 2) or not (data not shown). Age, KPS, and presence of comorbidities were not statistically significantly associated with survival and were not included in the final model.

Figure 1.

Overall and disease‐free survival by marital status. (A): Overall survival. (B): Disease‐free survival.Abbreviations: CI, confidence interval; HR, hazard ratio; M/P, married/partnered; RL, reference level; S, single.

Table 2.

Multivariate analysis for OS and DFS (n = 310a/367)

Endpoint	Adjustment variables	Comparison	Adjusted HRb	95% C.I. LL	95% C.I. UL	p valuec
OS	Marital status	Married or partnered vs. single	1.04	0.78	1.37	.81
	Race	White vs. other	1.50	1.05	2.14	.027
	CA19‐9	<90 vs. ≥90	2.88	2.02	4.09	<.0001
DFS	Marital status	Married or partnered vs. single	0.94	0.716	1.234	.65
	Surgical margin	Negative	1.00
	Status	Positive	1.17	0.89	1.54	.25
		Unknown	0.70	0.51	0.95	.022
	CA19‐9	<90 vs. ≥90	2.45	1.73	3.46	<.0001

^a57 patients did not have analyzable postresection CA19‐9.

^bHazard ratio: a hazard ratio of 1 indicates no difference between the two subgroups. The variables were coded such that a HR >1 indicates an increased risk of death or failure for the second level of the variables listed.

^c p value from chi‐square test using the Cox proportional hazards model.

Abbreviations: C.I., confidence interval; DFS, disease‐free survival; HR, harzard ratio; LL, lower limit; OS, overall survival.

Marital Status and Sex

Given prior reports that male patients with cancer derive a greater survival benefit from marriage 3, 4, the interaction between marital status and sex was explored. Sex alone was not a significant predictor of survival on univariate analysis (data not shown). Therefore, OS and DFS in married or partnered men compared with all other patients were examined. Again, no statistically significant differences in OS or DFS were seen (Fig. 2).

Figure 2.

Overall and disease‐free survival by marital status and sex. (A): Overall survival. (B): Disease‐free survival.Abbreviations: CI, confidence interval; HR, hazard ratio; M/P, married/partnered; RL, reference level; S, single.

Discussion

Ancillary analysis of data from NRG/RTOG 97‐04 demonstrated no association between marital or partner status and OS or DFS in patients with resected pancreatic cancer who received adjuvant postoperative chemotherapy followed by concurrent external beam radiation therapy and chemotherapy. This is in contrast to several population‐based studies that have shown an epidemiological link between marital status and survival in this disease 2, 3, 4. Using the SEER database, Baine et al. have shown that among all adult patients diagnosed with pancreatic cancer between 1998 and 2003, median OS was 4 and 3 months for all married and unmarried patients, and 13 and 16 months for married and unmarried patients who underwent surgery 2. Another SEER analysis of 10 most common causes of cancer‐related deaths, including pancreatic cancer, among patients diagnosed between 2004 and 2008 found that married patients were less likely to present with metastatic disease, more likely to receive definitive therapy, and less likely to die of cancer‐related causes across all included malignancies 4. For patients with pancreatic cancer, the relative reduction in cancer‐associated death was 13%. When all unmarried patients with cancer were stratified into never married, separated, divorced, and widowed, all groups showed worse outcomes compared with married patients, although this analysis was not cancer type specific. In a later SEER analysis limited to patients who had undergone definitive surgery between 1988 to 2012, the association between marriage and survival was again confirmed, albeit not for all categories of unmarried individuals and in a stage‐dependent manner 5. Collectively, these findings suggest that the interaction of marriage with survival is greatly influenced by other factors, such as treatment and stage.

The positive effect of marriage on health outcomes has been attributed to better health habits, psychosocial support, and greater financial resources that may exist within a partnership 11. Consistent with these observations, the married or partnered patients in the current study had fewer select comorbidities and more attributes of higher socioeconomic status then single patients. The main difference between population‐based analyses and the current study is the highly selective nature of the patient population with regard to those attributes 12, 13. First, these are patients who have already undergone a surgical resection, which is the most difficult part of the treatment and likely selects for individuals who have good functional status and support network. Second, they met strict eligibility criteria further attesting to their overall health. Third, they were undergoing adjuvant therapy, a phase of the overall treatment plan, which is relatively easy to tolerate and may not require much support. Thus, it is not surprising that the potential effect of marriage at this particular juncture in the treatment course of these patients may be minimal or nonexistent. This selection undoubtedly narrowed the difference between those with plenty of support and those with little support, even despite a pre‐existing discrepancy in some measures of overall health and socioeconomic status favoring the married or partnered group.

The ultimate objective of studies to identify psychosocial factors that influence survival is, after all, development of improved support mechanisms and other interventions that could mitigate any identified negative risk factors. These results suggest that no such interventions need to be considered during the adjuvant aspect of treatment.

The positive effect of marriage on survival observed in population‐based studies may be more representative of the association between marriage and survival in the patients with pancreatic cancer as a group and, therefore, more generalizable than the findings presented here. Factors such as the failure to receive therapy including surgery and adjuvant treatments have been reported to be significantly more common among unmarried patients with pancreatic cancer diagnosed in 2009 in the U.S. 14. Failure to receive these treatments is much more likely to influence survival than supportive care during ongoing treatment. Therefore, it may be informative to focus future research into the access to health care and the initial interface with the medical system when the overall treatment plan is developed.

Potential limitations of this study include data from a single study and a highly selected patient population. The number of DFS and OS events provide sufficient statistical power for detecting associations for moderate differences between married or partnered and single persons, but less statistical power for small differences. Patients enrolled on a trial represent a select group with regard to clinical factors such as stage, functional status, and comorbidities as well as their attitudes toward medical interventions and socioeconomic status, all of which warrant some caution in generalizing these results to the population at large. Despite these potential limitations, this analysis adds to the current literature by providing a more detailed look at a selected subset of patients with pancreatic cancer who had completed and recovered from complex surgery. These findings warrant caution in using highly selected patient populations for evaluation of differences related to marital status.

Conclusion

These findings show that among patients with resected pancreatic cancer receiving adjuvant therapy on NRG/RTOG 97‐04, marital or partner status was not associated with improved survival. Future efforts to elucidate the potential association of marriage with survival in this disease should be focused on less selected populations and include patients prior to initiation of complex treatments such as surgical resection.

Author Contributions

Conception/design: Marsha Reyngold, Kathryn A. Winter, William F. Regine, Ross A. Abrams, Howard Safran, John P. Hoffman, David L. Sherr, Jennifer Moughan, Christopher H. Crane

Provision of study material or patients: Rex B. Mowat, John P. Hayes, Ivan L. Kessel, Thomas DiPetrillo, Samir Narayan, Yuhchyau Chen, Edgar Ben‐Josef, Guila Delouya, John H. Suh, Joshua Meyer, Michael G. Haddock, Marvin Feldman, Rakesh Gaur, Kathleen Yost, Richard A. Peterson

Collection and/or assembly of data: Kathryn A. Winter, Jennifer Moughan

Data analysis and interpretation: Marsha Reyngold, Kathryn A. Winter, Jennifer Moughan, David L. Sherr

Manuscript writing: Marsha Reyngold, Kathryn A. Winter, William F. Regine, Ross A. Abrams, Howard Safran, John P. Hoffman, Rex B. Mowat, John P. Hayes, Ivan L. Kessel, Thomas DiPetrillo, Samir Narayan, Yuhchyau Chen, Edgar Ben‐Josef, Guila Delouya, John H. Suh, Joshua Meyer, Michael G. Haddock, Marvin Feldman, Rakesh Gaur, Kathleen Yost, Richard A. Peterson, David L. Sherr, Jennifer Moughan, Christopher H. Crane

Final approval of manuscript: Marsha Reyngold, Kathryn A. Winter, William F. Regine, Ross A. Abrams, Howard Safran, John P. Hoffman, Rex B. Mowat, John P. Hayes, Ivan L. Kessel, Thomas DiPetrillo, Samir Narayan, Yuhchyau Chen, Edgar Ben‐Josef, Guila Delouya, John H. Suh, Joshua Meyer, Michael G. Haddock, Marvin Feldman, Rakesh Gaur, Kathleen Yost, Richard A. Peterson, David L. Sherr, Jennifer Moughan, Christopher H. Crane

Disclosures

William F. Regine: Xcision (IP–patent pending); Guila Delouya: Astellas, Bayer, Janssen, Ter Sera, Abbvie, Sanofi, Elekta (C/A, H), Janssen (RF); John H. Suh: Abbvie (C/A); Marvin Feldman: Boehringer Ingelheim, Eli Lilly & Co., Bristol‐Myers Squibb (other–personal fees); Christopher H. Crane: Celgene (H, C/A). The other authors indicated no financial relationships.

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board

Supporting information

See http://www.TheOncologist.com for supplemental material available online.

Supplemental Table S1: Education and Employment status