Key Points
Question
To what extent are clinical characteristics and treatment patterns associated with the survival difference between male and female patients with breast cancer?
Findings
In this large cohort study of 16 025 male and 1 800 708 female patients with breast cancer in the United States, the male patients had 19% higher fully adjusted overall mortality compared with their female counterparts. Clinical characteristics and undertreatments were associated with 63.3% of the excess mortality among male patients.
Meaning
These results suggest a need for further research into biological features of breast cancer and tailored treatments for men with the disease to mitigate the sex-based disparity in mortality.
This cohort study analyzes data from the National Cancer Database to evaluate the factors associated with higher mortality rates among men with breast cancer compared with women with breast cancer during the first few months after diagnosis.
Abstract
Importance
Survival differences between male and female patients with breast cancer have been reported, but the underlying factors associated with the disparity have not been fully studied. This understanding is fundamental to developing strategies for cancer treatment and survivorship care.
Objective
To compare mortality between male and female patients with breast cancer and quantitatively evaluate the factors associated with sex-based disparity in mortality.
Design, Setting, and Participants
This large, nationwide, registry-based cohort study used the National Cancer Database to identify and obtain data on patients who received a breast cancer diagnosis between January 1, 2004, and December 31, 2014. After exclusions, the final study population comprised 1 816 733 patients. Statistical analyses were conducted from September 1, 2018, to January 15, 2019.
Main Outcomes and Measures
The primary outcome was overall survival. Secondary outcomes were 3-year and 5-year mortality. Mortality differences were evaluated by Kaplan-Meier analysis. The roles of race/ethnicity, clinical characteristics, treatments, and access-to-care factors in the association between sex and mortality were estimated by nested Cox proportional hazards regression models with adjustment for age.
Results
In total, 16 025 male (mean [SD] age, 63.3 [13.0] years) and 1 800 708 female (mean [SD] age, 59.9 [13.3] years) patients with breast cancer were included in the study. Compared with female patients, male patients had higher mortality across all stages. For men, the overall survival rate was 45.8% (95% CI, 49.5-54.0; P < .001), the 3-year rate was 86.4% (95% CI, 85.9-87.0; P < .001), and the 5-year rate was 77.6% (95% CI, 76.8-78.3; P < .001). For women, the overall survival rate was 60.4% (95% CI, 58.7-62.0; P < .001), the 3-year rate was 91.7% (95% CI, 91.7-91.8; P < .001), and the 5-year rate was 86.4% (95% CI, 86.4-86.5; P < .001). Overall, clinical characteristics and undertreatments were associated with a 63.3% excess mortality rate for male patients. A higher proportion of excess deaths in men were explained by these factors in the first 3 years after breast cancer diagnosis (66.0%) and in all patients with early-stage cancer (30.5% for stage I and 13.6% for stage II). However, sex remained a significant factor associated with overall mortality (adjusted hazard ratio [HR], 1.19; 95% CI, 1.16-1.23) as well as mortality at 3-year (adjusted HR, 1.15; 95% CI, 1.10-1.21) and 5-year (adjusted HR, 1.19; 95% CI, 1.14-1.23) analyses, even after adjustment for clinical characteristics, treatment factors, age, race/ethnicity, and access to care.
Conclusions and Relevance
This study found that mortality after cancer diagnosis was higher among male patients with breast cancer compared with their female counterparts. Such disparity appeared to persist after accounting for clinical characteristics, treatment factors, and access to care, suggesting that other factors (particularly additional biological attributes, treatment compliance, and lifestyle factors) should be identified to help in eliminating this disparity.
Introduction
Although breast cancer is one of the most common malignant neoplasms worldwide, male breast cancer is rare, accounting for only 0.6% to 1% of all breast cancer cases1 and about 0.3% of all cancers in men globally.2 However, data from the Surveillance, Epidemiology, and End Results (SEER) program showed that age-adjusted incidence of male breast cancer across all races/ethnicities has steadily increased from 0.85 per 100 000 persons in 1975 to 1.19 per 100 000 persons in 2015 (40.0%), which was higher than the 24.7% increase in women.2 The American Cancer Society estimated that approximately 2670 new cases of invasive male breast cancer will be diagnosed in the United States in 2019, and approximately 500 of these patients will die from the disease.3
Studies have indicated that male patients with breast cancer had worse overall survival (OS) than their female counterparts, including those with early-stage disease,4,5 although results have been inconsistent.1,6 One study reported comparable 5-year and 10-year OS between male and female patients with breast cancer diagnosed from 1980 to 1998 after matching on age, date of diagnosis, disease stage, and primary histologic type.6 Another study included patients with breast cancer diagnosed between 1970 and 2007 and showed a greater survival advantage for male patients compared with female patients after accounting for demographic characteristics, disease stage, and treatment.1 Although survival differences may be attributed to older age, later stage at diagnosis, and shorter life expectancy in men,7 many studies suggest that breast cancer in men may have distinct biological characteristics compared with breast cancer in women.7,8 However, few studies have systematically investigated the factors associated with mortality in male patients with breast cancer or assessed whether breast cancer prognosis for men is congruent with that for women, accounting for the differences in clinical characteristics and treatment.
This cohort study used data from the National Cancer Database (NCDB) to examine the possible mortality disparity between male and female patients with breast cancer. We further investigated the potential factors associated with mortality.
Methods
Patients with a primary breast cancer diagnosis were identified from the NCDB,9 a registry that captures more than 70% of newly diagnosed cancers across the United States and is heavily audited through strict quality control.10 Because we used completely deidentified NCDB data, the institutional review board of Vanderbilt University approved the current study as exempt human research, for which no informed consent was needed. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
A total of 2 445 870 patients with breast cancer diagnosis between January 1, 2004, and December 31, 2014, were assessed for eligibility. Patients with a concomitant diagnosis or history of other malignant neoplasms (n = 423 653), those whose TNM stages were inconsistent with those of site-specific metastasis (n = 10 890) according to the American Joint Committee on Cancer Cancer Staging Manual,11 and those with missing follow-up information (n = 189 627), unknown follow-up times (n = 127), or invalid follow-up times (n = 4840) were excluded, resulting in a final study population of 1 816 733 patients (eFigure 1 in the Supplement).
Demographic characteristics, including sex, age at diagnosis, race/ethnicity, year of diagnosis, rural or urban residence, inferred annual household income and educational attainment, insurance status, and treating facility, were gathered from the NCDB. Sex was recorded according to the medical records. Household income and educational attainment were estimated by matching patient zip codes against American Community Survey data from 2000 or 2012, depending on time of diagnosis. Education was documented as a measure of the number of adults within patient zip codes who did not graduate from high school. Residence was estimated by matching patient Federal Information Processing Standards codes against US Department of Agriculture Economic Research Service records from 2003 or 2013, depending on time of diagnosis.12
Clinical characteristics obtained from the NCDB included tumor size, nodal status, TNM stage, estrogen receptor (ER) status, progesterone receptor (PR) status, human epidermal growth factor receptor 2 (HER2) status, histologic type, Nottingham combined histologic grade, lymphovascular invasion (LVI), Oncotype DX Breast Recurrence Score (RS; Genomic Health Inc), and Charlson/Deyo score. Information on HER2 status, histologic grade, LVI, and RS was not collected until 2010.
Available treatment information included breast operation type (eg, mastectomy or breast-conserving operation), chemotherapy, endocrine therapy, radiotherapy, and immunotherapy. Information on axillary staging approach was not collected until 2012, resulting in missing values for nearly 70% of all patients. Because axillary surgical type is more closely associated with quality of life than with survival, we included only nodal status in the current analysis. Detailed information on regimens of chemotherapy, endocrine therapy, and immunotherapy was unavailable in the NCDB, and these treatments were only coded as yes or no in this analysis.
Statistical Analysis
The primary study outcome was OS, defined as months from cancer diagnosis to death of any cause or to last contact. Patients lost to follow-up were censored at last contact. Causes of death and events of cancer recurrence or progression were unavailable in the NCDB; thus, we could not evaluate disease-free survival. Descriptive characteristics of male and female patients were compared using χ2 tests for categorical variables and 2-sided, unpaired t tests for continuous variables. Differences in OS between male and female patients were compared using the Kaplan-Meier method and log-rank test. Stratified analyses by stage, ER status, HER2 status, age, comorbidity, and calendar period were conducted. Analyses were carried out for OS and by follow-up time (3 years or 5 years after cancer diagnosis).
Cox proportional hazards regression models were used to estimate hazard ratios (HRs) and 95% CIs for overall, 3-year, and 5-year mortality associated with sex. First, we conducted analyses among male patients to identify the factors associated with mortality. Significant factors in male breast cancer survival analyses were included in the final analysis. Second, following the method of Breslin et al,13 we evaluated the extent to which these factors were associated with sex disparity in mortality by groups: race/ethnicity, clinical factors, treatments, and factors associated with access to care. We calculated relative attenuation in HR by including additional sets of covariates in the nested models: (HRR − HRROF) divided by (HRR − 1), in which HRR was the less adjusted HR for mortality comparing male with female patients, and HRROF was the HR with additional adjustment. Clinical factors included TNM stage, ER status, PR status, HER2 status, histologic type, histologic grade, LVI, and comorbidity as well as treatments, such as surgical procedure, chemotherapy, endocrine therapy, radiotherapy, and immunotherapy, and interval from diagnosis to first treatments. Factors associated with access to care included year of diagnosis, residence, annual household income, educational attainment, insurance status, treating facility, and distance to care. Missing data were treated as a category for discrete variables or were substituted with median values (distance to care) or stage-specific median values (interval from diagnosis to treatment) for continuous variables.
Considering the distinct relapse patterns of ER-positive and ER-negative cancers observed in female patients,14 we also performed time-dependent coefficient analyses.15 Schoenfeld residuals were plotted to test whether the association between sex and mortality was constant over time (ie, proportional hazard assumption).15 If significant time-varying effects were found (P < .05), results from analyses that used time-dependent coefficient Cox proportional hazards regression models were reported. Interactions of sex with ER status and disease stage were evaluated with stratified analyses and maximum likelihood tests.
All statistical tests were based on 2-sided probability with significance levels set at P < .05 and used R, version 3.5.1 (R Foundation for Statistical Computing). Adjustments for multiple comparisons were not made. Statistical analyses were conducted from September 1, 2018, to January 15, 2019 (eAppendix in the Supplement).
Results
In total, 16 025 male (mean [SD] age, 63.3 [13.0] years) and 1 800 708 female (mean [SD] age, 59.9 [13.3] years) patients with breast cancer were included in this study (Table 1). Male patients, compared with their female counterparts, were older at diagnosis (63.3 vs 59.9 years) and had higher proportions of late-stage breast cancer (stage III: 14.0% [n = 2240] vs 8.9% [n = 160 710]; stage IV: 5.8% [n = 928] vs 3.8% [n = 67 704]), ER-positive breast cancer (83.9% [n = 13 446] vs 74.7% [n = 1 345 226]), ductal carcinoma (75.5% [n = 12 100] vs 67.2% [n = 1 209 434]), and LVI (10.7% [n = 1712] vs 6.6% [n = 118 698]) as well as high and very low RS (RS≥31: 12.1% [n = 121] vs 10.0% [n = 12 625; RS≤10: 34.1% [n = 341] vs 23.0% [n = 28 857]). Endocrine therapy was less frequent among men with ER-positive or PR-positive cancers than among women (57.9% [n = 7838] vs 70.2% [n = 957 724]). Male, compared with female, patients were less likely to receive radiotherapy, including those who received a breast-conserving operation (2697 [64.9%] vs 810 620 [78.6%]; P < .001) or those with stage II or III cancer who received a mastectomy (2342 [38.6%] vs 146 040 [43.7%]; P < .001) (eTable 1 in the Supplement).
Table 1. Demographic Characteristics of Patients With Breast Cancer by Sex.
| Variable | Breast Cancer Case, No. (%) | P Value | |
|---|---|---|---|
| Male (n = 16 025) | Female (n = 1 800 708) | ||
| Age, mean (SD), y | 63.3 (13.0) | 59.9 (13.3) | <.001 |
| Race/ethnicity | <.001 | ||
| White | 13 221 (82.5) | 1 496 086 (83.1) | |
| Black | 2089 (13.0) | 205 207 (11.4) | |
| Hispanic | 487 (3.0) | 78 238 (4.3) | |
| Other | 228 (1.4) | 21 177 (1.2) | |
| Residencea | .53 | ||
| Metro | 13 307 (83.0) | 1 499 080 (83.2) | |
| Urban | 1950 (12.2) | 218 548 (12.1) | |
| Rural | 250 (1.6) | 28 442 (1.6) | |
| Unknown | 518 (3.2) | 54 638 (3.0) | |
| Annual household incomeb | <.001 | ||
| <$30 000 | 2045 (12.8) | 213 529 (11.9) | |
| $30 000-$34 999 | 2717 (17.0) | 301 127 (16.7) | |
| $35 000-$45 999 | 4136 (25.8) | 466 267 (25.9) | |
| ≥$46 000 | 6691 (41.8) | 775 108 (43.0) | |
| Unknown | 436 (2.7) | 44 677 (2.5) | |
| Educational attainment (no high school graduation)b | .01 | ||
| ≥29% | 2227 (13.9) | 253 487 (14.1) | |
| 20%-28.9% | 3527 (22.0) | 381 964 (21.2) | |
| 14%-19.9% | 4061 (25.3) | 456 751 (25.4) | |
| <14% | 5772 (36.0) | 663 910 (36.9) | |
| Unknown | 438 (2.7) | 44 596 (2.5) | |
| Distance to care, mean (SD), miles | 21.2 (84.6) | 22.2 (92.9) | .20 |
| Facility type | <.001 | ||
| Community | 1965 (12.3) | 176 873 (9.8) | |
| Comprehensive community | 7344 (45.8) | 825 466 (45.8) | |
| Academic/research program | 4437 (27.7) | 515 045 (28.6) | |
| Integrated network | 1768 (11.0) | 191 595 (10.6) | |
| Unknown | 511 (3.2) | 91 729 (5.1) | |
| Insurance | <.001 | ||
| No insurance | 454 (2.8) | 39 894 (2.2) | |
| Private insurance | 7459 (46.5) | 997 570 (55.4) | |
| Government insurance | 7761 (48.4) | 726 710 (40.4) | |
| Unknown | 351 (2.2) | 36 534 (2.0) | |
SI conversion factor: To convert miles to kilometers, multiply by 1.6.
Residence was estimated by matching patients’ Federal Information Processing Standards codes against US Department of Agriculture Economic Research Service records from 2003 or 2013, depending on the time of diagnosis.
Median annual household income and educational attainment were estimated by matching patients’ zip codes against American Community Survey data from 2000 or 2012, depending on the time of diagnosis.
With a median (interquartile range) follow-up of 54.0 (29.1-85.0) months for men and 60.5 (34.3-92.6) months for women (P < .001), 3988 deaths (24.9%) occurred among men and 288 989 (16.0%) among women. Male patients compared with their female counterparts had lower OS rates (45.8% [95% CI, 49.5-54.0; P < .001] vs 60.4% [95% CI, 58.7-62.0; P < .001]), 3-year survival rates (86.4% [95% CI, 85.9-87.0; P < .001] vs 91.7% [95% CI, 91.7-91.8; P < .001]), and 5-year survival rates (77.6% [95% CI, 76.8-78.3; P < .001] vs 86.4% [95% CI, 86.4-86.5; P < .001]; P < .001) (Figure, A) regardless of cancer stage, ER and/or HER2 status, age, year of diagnosis, and comorbidity status (eFigures 2-4 in the Supplement). The 5-year survival rates for men were 87.8% (95% CI, 86.7-88.8; P < .001) for stage I, 78.9% (95% CI, 77.7-80.2; P < .001) for stage II, 63.3% (95% CI, 61.0-65.5; P < .001) for stage III, and 21.4% (95% CI, 18.4-24.5; P = .007) for stage IV disease. The corresponding rates for women were 92.5% (95% CI, 92.4-92.5; P < .001) for stage I, 85.9% (95% CI, 85.8-86.0; P < .001) for stage II, 70.1% (95% CI, 69.8-70.3; P < .001) for stage III, and 25.1% (95% CI, 24.7-25.5; P = .007) for stage IV (Figure).
Figure. Crude Breast Cancer Overall Survival by Sex and Stage.
The number of patients (A) in the entire study population was 1 816 733, (B) with stage 0 breast cancer was 351 387, (C) with stage I breast cancer was 703 326, (D) with stage II breast cancer was 463 677, (E) with stage III breast cancer was 162 950, and (F) with stage IV breast cancer was 68 632. P values were from the log-rank test.
Age, clinical factors (histologic grade, disease stage, PR status, LVI, and comorbidity), treatments (surgical procedure, chemotherapy, radiotherapy, endocrine therapy, and immunotherapy), access to care (residence, income, insurance, and facility), and race/ethnicity were significantly associated with overall mortality for male patients. No significant associations with ER status, HER2 status, and histologic type were observed (eTable 2 in the Supplement).
For the entire cohort, time-dependent coefficient analyses showed constant associations between sex and total mortality over time (unadjusted HR, 1.72; 95% CI, 1.67-1.77; P for time-varying effect = .86). Age-adjusted HR for total mortality among male compared with female patients (model 1) was 1.49 (95% CI, 1.44-1.54). Further adjustments for clinical factors (model 2) resulted in 49.0% relative attenuation of the HR (adjusted HR, 1.25; 95% CI, 1.21-1.29), and adjustments for treatment factors (model 3) resulted in 57.1% relative attenuation (adjusted HR, 1.21; 95% CI, 1.18-1.25). Combined adjustment for both clinical and treatment factors (model 6) resulted in 63.3% relative attenuation (adjusted HR, 1.18; 95% CI, 1.15-1.22), and adjustment for all 5 factor groups (model 7) resulted in 61.2% relative attenuation (adjusted HR, 1.19; 95% CI, 1.16-1.23). Similar patterns were observed for 3-year (adjusted HR, 1.15; 95% CI, 1.10-1.21) and 5-year (adjusted HR, 1.19; 95% CI, 1.14-1.23) mortality analyses (Table 2).
Table 2. Factors in Sex-Based Disparity in Mortality Among Patients With Breast Cancera.
| Model | Total Mortality, Male vs Female, HR (95% CI) | Excess Mortality Associated With Additional Adjusted Covariates, %b | 3-y Mortality, Male vs Female, HR (95% CI) | Excess Mortality Associated With Additional Adjusted Covariates, %b | 5-y Mortality, Male vs Female, HR (95% CI) | Excess Mortality Associated With Additional Adjusted Covariates, %b |
|---|---|---|---|---|---|---|
| Unadjusted model | 1.72 (1.67-1.77) | NA | 1.69 (1.61-1.76) | NA | 1.73 (1.67-1.79) | NA |
| Model 1c | 1.49 (1.44-1.54) | NA | 1.47 (1.41-1.54) | NA | 1.51 (1.46-1.57) | NA |
| Model 2d | 1.25 (1.21-1.29) | 49.0 | 1.21 (1.16-1.27) | 55.3 | 1.25 (1.21-1.30) | 51.0 |
| Model 3e | 1.21 (1.18-1.25) | 57.1 | 1.18 (1.13-1.23) | 61.7 | 1.21 (1.17-1.26) | 58.8 |
| Model 4f | 1.48 (1.43-1.52) | 2.0 | 1.45 (1.39-1.52) | 4.3 | 1.50 (1.44-1.55) | 2.0 |
| Model 5g | 1.48 (1.44-1.53) | 2.1 | 1.45 (1.39-1.52) | 4.3 | 1.50 (1.44-1.55) | 2.0 |
| Model 6h | 1.18 (1.15-1.22) | 63.3 | 1.16 (1.11-1.21) | 66.0 | 1.18 (1.14-1.23) | 64.7 |
| Model 7i | 1.19 (1.16-1.23) | 61.2 | 1.15 (1.10-1.21) | 68.1 | 1.19 (1.14-1.23) | 62.7 |
Abbreviations: HR, hazard ratio; NA, not applicable.
Time-dependent coefficient analyses showed that the associations of sex with total mortality were constant over time for the entire cohort (P = .86), and results from a Cox proportional hazards regression model are reported in this table.
Proportion of excess mortality was relative to the age-adjusted model for multivariate factors (model 1): (HRR − HRROF)/(HRR − 1), where HRR was the HR for model 1, and HRROF was the HR with additional adjustment.
Adjusted for age.
Adjusted for age and clinical factors (progesterone receptor status, histologic grade, TNM stage, lymphovascular invasion status, and comorbidity).
Adjusted for age and treatment factors (surgical procedure, chemotherapy, endocrine therapy, radiotherapy, and immunotherapy).
Adjusted for age and race/ethnicity.
Adjusted for age and access to care (urban or rural residence, income, insurance status, and treatment facility).
Adjusted for age, clinical factors, and treatment factors.
Adjusted for age, clinical and treatment factors, race/ethnicity, and access to care.
As shown in Table 3, stage-specific subgroup analyses showed that sex differences in total mortality varied significantly over time after cancer diagnosis for patients with stage I or stage III breast cancer. For patients with stage I, after adjustment for clinical and treatment factors, the associations between sex and mortality were stronger within the first 3-year period (HR, 1.41; 95% CI, 1.26-1.58) compared with beyond the 3-year period (HR, 1.12; 95% CI, 1.01-1.23; P for time-varying effect = .002). For patients with stage III disease, the respective associations were opposite for within 3 years of cancer diagnosis (HR, 0.90; 95% CI, 0.82-0.99) and beyond 3 years after diagnosis (HR, 1.40; 95% CI, 1.28-1.53; P for time-varying effect <.001). Male patients with stage II or IV cancer had a higher mortality risk than their female counterparts in both periods, although the association was slightly higher and more broadly significant during the 3 years after diagnosis than the earlier period. Test of the multiplicative interaction between sex and stage was significant in the first 3-year period (P for interaction <.001) but was not significant thereafter (P for interaction = .67).
Table 3. Sex-Based Disparity in Mortality for Patients With Breast Cancer by Cancer Stagesa.
| Model | Within 3-y Period of Cancer Diagnosis | Excess Mortality Associated With Additional Adjusted Covariates, %b | After 3-y Period of Cancer Diagnosis | Excess Mortality Associated With Additional Adjusted Covariates, %b | P Valuec | ||||
|---|---|---|---|---|---|---|---|---|---|
| Deaths/Patients at Risk | HR (95% CI) | Deaths/Patients at Risk | HR (95% CI) | ||||||
| Male | Female | Male | Female | ||||||
| Stage I | 295/4817 | 23 797/698 509 | NR | NR | 419/3498 | 48 057/532 020 | NR | NR | .002 |
| Model 1d | 295/4817 | 23 797/698 509 | 1.59 (1.41-1.78) | NR | 419/3498 | 48 057/532 020 | 1.24 (1.12-1.36) | NR | |
| Model 2e | 295/4817 | 23 797/698 509 | 1.41 (1.26-1.58) | 30.5 | 419/3498 | 48 057/532 020 | 1.12 (1.01-1.23) | 50.0 | |
| Stage II | 569/5313 | 33 584/458 364 | NR | NR | 730/3655 | 44 081/339 640 | NR | NR | .06 |
| Model 1d | 569/5313 | 33 584/458 364 | 1.22 (1.12-1.32) | NR | 730/3655 | 44 081/339 640 | 1.34 (1.25-1.44) | NR | |
| Model 2e | 569/5313 | 33 584/458 364 | 1.19 (1.09-1.29) | 13.6 | 730/3655 | 44 081/339 640 | 1.30 (1.21-1.40) | 11.8 | |
| Stage III | 413/2240 | 28 864/160 710 | NR | NR | 480/1436 | 24 630/107 432 | NR | NR | <.001 |
| Model 1d | 413/2240 | 28 864/160 710 | 0.87 (0.79-0.96) | NR | 480/1436 | 24 630/107 432 | 1.36 (1.24-1.49) | NR | |
| Model 2e | 413/2240 | 28 864/160 710 | 0.90 (0.82-0.99) | NA | 480/1436 | 24 630/107 432 | 1.40 (1.28-1.53) | NA | |
| Stage IV | 549/928 | 37 878/67 704 | NR | NR | 150/275 | 10 502/21 248 | NR | NR | .06 |
| Model 1d | 549/928 | 37 878/67 704 | 1.05 (0.96-1.14) | NR | 150/275 | 10 502/21 248 | 1.21 (1.03-1.42) | NR | |
| Model 2e | 549/928 | 37 878/67 704 | 1.12 (1.03-1.22) | NA | 150/275 | 10 502/21 248 | 1.28 (1.09-1.50) | NA | |
Abbreviations: HR, hazard ratio; NA, not applicable; NR, not reported.
Interaction of sex with stage was significant for within the 3-year period (P < .001) but was not significant after the 3-year period of cancer diagnosis (P = .67).
Proportion of excess mortality was relative to the age-adjusted model for multivariate factors (model 1). If the HRs with additional adjustment were not attenuated compared with the less adjusted model, the extent to which the adjusted factors could explain the excessive mortality could not be calculated following this formula: (HRR − HRROF)/(HRR − 1) and was coded as NA. HRR was the HR for model 1, whereas HRROF was the HR for model 2.
P values were obtained from the assessment of proportional assumption for sex in the univariable Cox proportional hazards regression model. P < .05 indicates that a significant time-varying effect was found.
Adjusted for age.
Adjusted for age, clinical factors (progesterone receptor status, histologic grade, lymphovascular invasion status, and comorbidity), and treatment factors (surgical procedure, chemotherapy, endocrine therapy, radiation, and immunotherapy).
The time-varying associations between sex and mortality differed by ER status (P for interaction <.001, for both the first 3-year and after 3-year periods; eTable 3 in the Supplement). Male patients with ER-positive breast cancer had higher mortality than female patients across all stages. Male patients with stage III ER-negative breast cancer had lower mortality risk during the first 3 years after diagnosis and similar mortality risk afterward compared with their female counterparts. Male patients with ER-negative stage II or IV disease had higher 3-year mortality than female patients. Sex-mortality associations for patients with triple-negative breast cancer (ER-negative, PR-negative, and HER2-negative) were quite similar to patients with ER-negative cancer. However, sample sizes for triple-negative breast cancer subtype analyses were small, and the risk estimates were unstable.
Discussion
In this large-scale registry-based study, we found that male patients with breast cancer significantly differed from their female counterparts as characterized by older age at diagnosis, a higher proportion of ER-positive subtype or advanced disease, and less likelihood to receive conventional treatment. Men had higher mortality than women overall and across disease stages, particularly for ER-positive breast cancer. Clinical and treatment characteristics were the most common factors in sex-based disparity in mortality, but the differences remained even after adjustment for age, race/ethnicity, clinical and treatment characteristics, and access to care.
In this study, male patients accounted for 0.90% of all patients with breast cancer, consistent with SEER data,2 indicating no or minimal selection bias. Similar to previous reports,1,4,7,8,16 male compared with female patients were more likely to receive a diagnosis at a later age and have ER-positive disease and ductal histologic type. The proportion of ductal histologic type among male patients from this study (75.5%) was slightly lower than in other studies (>85.0%),8,17,18,19 possibly owing to the lack of centrally reviewed pathological data and potentially different diagnostic criteria applied across facilities in this study. Consistent with a recent report,20 we found that both high (≥31) and very low (<11) RS were more frequent in male than female patients. It has been speculated that breast cancer in men may have more similarities to breast cancer in postmenopausal women,16 but the differences found suggest a possible distinct biology for male breast cancer. Further studies on the mechanisms and clinical management (eg, endocrine therapy strategies) of male breast cancer are needed.
In this study, higher proportions of male patients had more advanced disease compared with female patients, which could be attributed to lack of awareness and screening of breast cancer in men.21 However, despite their aggressive clinical characteristics, male patients were less likely to receive conventional treatments. For example, despite higher proportions of ER-positive or PR-positive disease, fewer men received adjuvant endocrine therapy, suggesting either undertreatment or poor compliance. Unfortunately, we were unable to further examine the underlying factors owing to the unavailability of relevant information. Owing to a lack of clinical trials specific to male patients, treatment for male breast cancer typically follows guidelines developed for female patients. Nevertheless, previous reports indicate that compliance with these guidelines is generally lower for male patients.22,23 Research on current guidelines compliance and the development of sex-specific guidelines is warranted.
We found that OS for male breast cancer patients was substantially worse than for female patients, consistent with previous reports.1,4 However, results on stage-specific survival have been inconsistent. Some studies4,24,25 suggested the sex disparity in OS was only evident for patients with early-stage but not advanced-stage disease. In contrast, studies by El-Tamer et al6 and Miao et al1 showed no significant differences in stage-specific survival between male and female patients with breast cancer. In this study, the men had worse survival than the women across all cancer stages. This finding is supported by a SEER-based study26 that showed lower 5-year stage-by-stage OS for male and female cases. Discrepancies between the present study and the studies by El-Tamer et al6 and Miao et al1 may be explained by their coverage of earlier periods and longer time frames: 18 years (1980-1998) for El-Tamer et al6 and 37 years (1970–2007) for Miao et al1 compared with 10 years (2004-2014) for this study. Diagnostic approaches and treatment strategies are known to have remarkably improved in recent decades, although most advances were based on research findings about female patients. Although mortality rates have declined over time for male patients with breast cancer, improvement in survival for male cases may still lag behind that for female patients.16 For example, whereas a Greif et al4 study used NCDB data from 1998 to 2007, this study used 5-year survival rates, which included NCDB data from 2004 to 2014 and showed an improvement in survival by 7.5% (from 56.5%-63.3%) for men and by 13.2% (from 56.9%-70.1%) for women with stage III breast cancer. A delay in or underutilization of adjuvant therapy, especially tamoxifen citrate for hormone receptor–positive breast cancer, was associated with slower improvement of male breast cancer survival.27 In this study, we found that adjustment for cancer treatments attenuated HRs associated with sex for total mortality by 57.1%, which may support our hypothesis.
A unique finding of this study is that sex remained an independent factor associated with breast cancer prognosis after full adjustment, with male patients having 15% higher 3-year mortality and 19% higher 5-year mortality compared with their female counterparts. Men generally have higher mortality rates than women. However, we found that excess mortality among male patients cannot be completely explained by the general excessive mortality in men. For example, the stage- and ER status–specific analyses showed that HRs associated with sex were higher for the first 3 years after cancer diagnosis than for the entire study period for stage I or II ER-positive breast cancer. For ER-negative breast cancer, however, male patients showed no higher mortality than female cases except those with stage IV disease. Although this finding may be partly explained by the distinct relapse patterns of ER-positive and ER-negative breast cancers,14 we could not rule out the possibility of chance findings owing to multiple statistical testing. The subtype-specific associations may also indicate involvement of yet-to-be identified biological or treatment factors. In addition, ER status, HER2 status, and histologic type—the well-established prognosis factors associated with mortality in female patients—were not associated with mortality in male patients. Thus, low proportions of ER-negative, HER2-positive, and ductal histologic type among male patients may be an explanation. This sex disparity in mortality may also indicate distinct prognostic factors for male breast cancer.
Strengths and Limitations
Strengths of this study include a large number of patients over an 11-year time frame, which allowed a comprehensive evaluation of sex disparity with an adjustment of a wide range of covariates. The NCDB registry captures nearly 70% of all US cancer cases, and the proportion of male patients in the total population was consistent with that of SEER, indicating the generalizability of these findings.
The study also has some limitations. First, information on cancer recurrence and cause of death was unavailable in the NCDB. Although we carried out 3-year and 5-year survival analyses, assuming that death close to cancer diagnosis would be more likely associated with breast cancer, disease-free and cancer-specific survival could not be assessed. Second, patient records and associated pathological information were not centrally reviewed, and heterogeneity among facilities may have compromised the assessment of this information. Third, detailed pathologic and treatment information, genetic test information (eg, BRCA), patient compliance data, and lifestyle factor and comorbidity information were missing.
Conclusions
This study found that male patients with breast cancer had higher mortality rates than their female counterparts, particularly during the period right after diagnosis. This sex-based disparity in mortality appeared to be mostly associated with clinical and treatment characteristics, especially advanced TNM stages and undertreatments. Future research should focus on why and how clinical characteristics, as well as biological features, may have different implications for the survival of male and female patients with breast cancer. Additional factors, particularly compliance to treatment, biological attributes, and lifestyle factors (eg, smoking, drinking, and obesity), should be assessed to help in developing treatments tailored for men, which would mitigate this sex-based disparity.
eFigure 1. CONSORT Diagram
eFigure 2. Crude Overall Survival of Breast Cancer by Sex and ER/HER2 Status
eFigure 3. Crude Overall Survival of Breast Cancer by Sex and Therapy Provided
eFigure 4. Crude Overall Survival of Breast Cancer by Sex by Age, Calendar Period and Comorbidity
eTable 1. Clinical Characteristics and Treatment Patterns of Breast Cancer by Sex
eTable 2. Factors Associated With Total Mortality for Male Breast Cancer Patients
eTable 3. Sex Disparity in Mortality for Breast Cancer Patients by Estrogen Receptor Status and Cancer Stages
eAppendix. Log File of Statistical Analyses
References
- 1.Miao H, Verkooijen HM, Chia KS, et al. Incidence and outcome of male breast cancer: an international population-based study. J Clin Oncol. 2011;29(33):4381-4386. doi: 10.1200/JCO.2011.36.8902 [DOI] [PubMed] [Google Scholar]
- 2.National Cancer Institute. SEER Cancer Statistics Review (CSR), 1975-2015. https://seer.cancer.gov/csr/1975_2015/. Updated September 10, 2018. Accessed January 31, 2019.
- 3.Siegel RL, Miller KD, Jemal A. Cancer statistics, 2019. CA Cancer J Clin. 2019;69(1):7-34. doi: 10.3322/caac.21551 [DOI] [PubMed] [Google Scholar]
- 4.Greif JM, Pezzi CM, Klimberg VS, Bailey L, Zuraek M. Gender differences in breast cancer: analysis of 13,000 breast cancers in men from the National Cancer Data Base. Ann Surg Oncol. 2012;19(10):3199-3204. doi: 10.1245/s10434-012-2479-z [DOI] [PubMed] [Google Scholar]
- 5.Li X, Yang J, Krishnamurti U, et al. Hormone receptor-positive breast cancer has a worse prognosis in male than in female patients. Clin Breast Cancer. 2017;17(5):356-366. doi: 10.1016/j.clbc.2017.03.005 [DOI] [PubMed] [Google Scholar]
- 6.El-Tamer MB, Komenaka IK, Troxel A, et al. Men with breast cancer have better disease-specific survival than women. Arch Surg. 2004;139(10):1079-1082. doi: 10.1001/archsurg.139.10.1079 [DOI] [PubMed] [Google Scholar]
- 7.Giordano SH. Breast cancer in men. N Engl J Med. 2018;378(24):2311-2320. doi: 10.1056/NEJMra1707939 [DOI] [PubMed] [Google Scholar]
- 8.Cardoso F, Bartlett JMS, Slaets L, et al. Characterization of male breast cancer: results of the EORTC 10085/TBCRC/BIG/NABCG International Male Breast Cancer Program. Ann Oncol. 2018;29(2):405-417. doi: 10.1093/annonc/mdx651 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.American College of Surgeons. National Cancer Database https://www.facs.org/quality-programs/cancer/ncdb. Accessed September 1, 2018.
- 10.Winchester DP, Stewart AK, Phillips JL, Ward EE. The National Cancer Data Base: past, present, and future. Ann Surg Oncol. 2010;17(1):4-7. doi: 10.1245/s10434-009-0771-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A, eds. AJCC Cancer Staging Manual 7th ed. New York, NY: Springer; 2010. [Google Scholar]
- 12.National Cancer Data Base participant use data file (PUF) data dictionary Version: PUF 2014. Containing cases diagnosed in 2004-2014. https://m.facs.org/puf. Accessed September 1, 2018.
- 13.Breslin TM, Morris AM, Gu N, et al. Hospital factors and racial disparities in mortality after surgery for breast and colon cancer. J Clin Oncol. 2009;27(24):3945-3950. doi: 10.1200/JCO.2008.20.8546 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Gomis RR, Gawrzak S. Tumor cell dormancy. Mol Oncol. 2017;11(1):62-78. doi: 10.1016/j.molonc.2016.09.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Therneau T, Crowson C, Atkinson E Using time dependent covariates and time dependent coefficients in the Cox model. https://cran.r-project.org/web/packages/survival/vignettes/timedep.pdf. Published March 29, 2019. Accessed December 18, 2018. [Google Scholar]
- 16.Anderson WF, Jatoi I, Tse J, Rosenberg PS. Male breast cancer: a population-based comparison with female breast cancer. J Clin Oncol. 2010;28(2):232-239. doi: 10.1200/JCO.2009.23.8162 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Yu XF, Feng WL, Miao LL, Chen B, Yang HJ. The prognostic significance of molecular subtype for male breast cancer: a 10-year retrospective study. Breast. 2013;22(5):824-827. doi: 10.1016/j.breast.2013.02.005 [DOI] [PubMed] [Google Scholar]
- 18.Vermeulen MA, Slaets L, Cardoso F, et al. Pathological characterisation of male breast cancer: results of the EORTC 10085/TBCRC/BIG/NABCG International Male Breast Cancer Program. Eur J Cancer. 2017;82:219-227. doi: 10.1016/j.ejca.2017.01.034 [DOI] [PubMed] [Google Scholar]
- 19.Losurdo A, Rota S, Gullo G, et al. Controversies in clinicopathological characteristics and treatment strategies of male breast cancer: a review of the literature. Crit Rev Oncol Hematol. 2017;113:283-291. doi: 10.1016/j.critrevonc.2017.03.013 [DOI] [PubMed] [Google Scholar]
- 20.Massarweh SA, Sledge GW, Miller DP, McCullough D, Petkov VI, Shak S. Molecular characterization and mortality from breast cancer in men. J Clin Oncol. 2018;36(14):1396-1404. doi: 10.1200/JCO.2017.76.8861 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Giordano SH. A review of the diagnosis and management of male breast cancer. Oncologist. 2005;10(7):471-479. doi: 10.1634/theoncologist.10-7-471 [DOI] [PubMed] [Google Scholar]
- 22.Kiluk JV, Lee MC, Park CK, et al. Male breast cancer: management and follow-up recommendations. Breast J. 2011;17(5):503-509. doi: 10.1111/j.1524-4741.2011.01148.x [DOI] [PubMed] [Google Scholar]
- 23.Anelli TF, Anelli A, Tran KN, Lebwohl DE, Borgen PI. Tamoxifen administration is associated with a high rate of treatment-limiting symptoms in male breast cancer patients. Cancer. 1994;74(1):74-77. doi: 10.1002/1097-0142(19940701)74:1<74::AID-CNCR2820740113>3.0.CO;2-# [DOI] [PubMed] [Google Scholar]
- 24.Wu SG, Zhang WW, Liao XL, et al. Men and women show similar survival outcome in stage IV breast cancer. Breast. 2017;34:115-121. doi: 10.1016/j.breast.2017.05.012 [DOI] [PubMed] [Google Scholar]
- 25.Leone J, Zwenger AO, Leone BA, Vallejo CT, Leone JP. Overall survival of men and women with breast cancer according to tumor subtype: a population-based study. Am J Clin Oncol. 2019;42(2):215-220. doi: 10.1097/COC.0000000000000497 [DOI] [PubMed] [Google Scholar]
- 26.Giordano SH, Cohen DS, Buzdar AU, Perkins G, Hortobagyi GN. Breast carcinoma in men: a population-based study. Cancer. 2004;101(1):51-57. doi: 10.1002/cncr.20312 [DOI] [PubMed] [Google Scholar]
- 27.Ribeiro G, Swindell R. Adjuvant tamoxifen for male breast cancer (MBC). Br J Cancer. 1992;65(2):252-254. doi: 10.1038/bjc.1992.50 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eFigure 1. CONSORT Diagram
eFigure 2. Crude Overall Survival of Breast Cancer by Sex and ER/HER2 Status
eFigure 3. Crude Overall Survival of Breast Cancer by Sex and Therapy Provided
eFigure 4. Crude Overall Survival of Breast Cancer by Sex by Age, Calendar Period and Comorbidity
eTable 1. Clinical Characteristics and Treatment Patterns of Breast Cancer by Sex
eTable 2. Factors Associated With Total Mortality for Male Breast Cancer Patients
eTable 3. Sex Disparity in Mortality for Breast Cancer Patients by Estrogen Receptor Status and Cancer Stages
eAppendix. Log File of Statistical Analyses
