Abstract
Although the incidence for breast cancer in men is lower than for women, male breast cancer (MBC) patients are diagnosed at a later stage and have a higher mortality rate than women. This study examined male cases reported from 1988 through 2006 in the Surveillance, Epidemiology, and End Results Program of the National Cancer Institute for differences in cancer stage, tumor size at diagnosis, and survival rate between metropolitan and nonmetropolitan regions. Pearson’s chi-square was used to evaluate differences in stage and tumor size at diagnosis. Cox proportional hazards regression was used to assess survival differences after adjusting for confounders (race, marital status, median family income, age, and education). Regional differences in tumor grade size and stage at diagnosis were not statistically significant; however, survival differences were observed between metropolitan and nonmetropolitan regions. An interaction between nonmetropolitan area and regional stage MBC was a significant predictor of poorer survival. Raising awareness of MBC in nonmetropolitan areas could save the lives of many men and action should be taken to improve health care access, treatment, and thus prognosis in this population.
Introduction
The urban environment exerts an influence on many aspects of health such as what people eat and drink, housing conditions, and access and availability to health care (Galea, Freudenberg, & Vlahov, 2005). Researchers in urban health have theorized that people living in urban settings can be exposed to more unhealthy environments and increasing unemployment and have poorer health than people living in rural areas (Vlahov & Galea, 2002; Wasylenki, 2001). Vlahov, Galea, and Freudenberg (2005) noted that there are benefits associated with urban living, such as social support, accessibility to health care, and the ability to mobilize political and social movements, and that these benefits have been underrecognized in an urban setting. Additionally, an urban health advantage where the concentration of people is large enough is an economic advantage for health care facilities as they can offer specialized services to a large number of people (Rice & Smith, 2001; Vlahov et al., 2005).
Studies of geographic differences in screening practices, cancer stage, tumor grade size, and survival for female breast cancer in the United States have reported that the differences have been inconsistent (Barry & Breen, 2005; Blair et al., 2006; Bradley, Given, & Roberts, 2000; Campbell et al., 2001; Coughlin, Thompson, Hall, Logan, & Uhler, 2002; Hill, Khamis, Tyczynski, & Berkel, 2005; Liff, Chow, & Greenberg, 1991). Ormond, Zuckerman, and Lhila (2000), using the National Survey of America’s Families, observed that of the 44,000 families of individuals of the age 65 years and younger, 9.3% of the urban residents stated that their health status was poor compared with 13.1% of the rural residents (p < .05). Additionally, 71.8% of the rural population visited doctors compared with 77.3% of the urban population (p < .05) and 10.1% of the rural population lacked the certainty that they receive the needed care compared with 8.6% of the urban population.
In a study of rural male patients presenting with prostate cancer, Francis, Hegney, Bramston, and Warden (2001) used face-to-face interviews to understand the influence of rurality on diagnosis and treatment of prostate cancer. The authors made several essential observations that can be relevant to male breast cancer (MBC). For example, rural men did not recognize early symptoms of prostate cancer, and traveling great distances to access specialists was both a financial and an emotional challenge. The authors further stated that rural men’s health as an area of study has been neglected and should be a focus of further research.
As per survival statistics reported in the American Society of Clinical Oncology (n.d.) for men with early stages of breast cancer (Stage I or less), the 5-year survival after detection is approximately 98%; however, survival rates drop as the stage of disease increases. Males with breast cancer that has not metastasized beyond the breast have an estimated 84% 5-year relative survival rate, and for MBC that has spread to other parts of the body, the 5-year survival rate drops to 27% (American Society of Clinical Oncology, n.d.).
In general, MBC is diagnosed at a later stage, and late stage at diagnosis and advanced age at diagnosis are associated with a higher risk for advanced tumors (Giordano, 2005). Men with breast tumor size measuring 2 to 5 cm (20-50 mm) have a 40% higher risk of mortality than men with breast tumors that are less than 2 cm. The prognosis for men compared with women is reported to be worse primarily because males are diagnosed with breast cancer at an older age and present with an advanced stage of the disease (Giordano, 2005).
From 1997 to 2001, the lifetime risk of a woman being diagnosed with breast cancer has been improving (Ries et al., 2006). In 1997, the lifetime risk of a woman developing breast cancer was 13.4% (1 in 7), and in 2001, it was 12.7% (1 in 8). On the other hand, a man’s lifetime risk of developing breast cancer is about one tenth of 1%, or 1 in 1,000. Though statistics reported a decline in female breast cancer rates in Western society, the incidence of MBC in the 1970s was 1.0 per 100,000, and has increased to approximately 1.2 per 100,000 in 2008 (Speirs & Shaaban, 2008). Because cancer treatment is not standardized nationwide, survival for MBC following diagnosis may also vary.
Method and Data Source
The study used data from the Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute. SEER is a comprehensive source of population-based information in the United States and includes information on stage of cancer at diagnosis and patient survival data. Since its inception in 1973, the SEER program has been collecting information on cancer incidence and survival from population-based registries in geographic areas that make up approximately 26% of the U.S. population (http://seer.cancer.gov/). Since January 1, 1973, the program has been collecting information on cancer cases in Connecticut, Hawaii, Iowa, New Mexico, and Utah and in the metropolitan areas of Detroit and San Francisco–Oakland. In 1974-1975, the metropolitan area of Atlanta and the 13-county Seattle–Puget Sound areas were added to the program. In 1978, 10 predominantly Black rural counties in Georgia were added, and in 1980, American Indians residing in Arizona were also added. Three geographic areas participated in the SEER program prior to 1990: New Orleans, Louisiana (1974-1977, rejoined 2001); New Jersey (1979-1989, rejoined 2001); and Puerto Rico (1973-1989). In 1992, the program expanded to include minority populations, particularly Hispanics through the addition of Los Angeles County and four counties in the San Jose–Monterey area south of San Francisco. In 2001, the SEER Program further expanded coverage to include Kentucky, the remaining counties in California (Greater California), as well as New Jersey and Louisiana.
Data Analysis
Variables previously identified in the literature to be prognostic indicators were stage and tumor grade size at diagnosis and survival rate (Chan, Hart, & Goodman, 2006; Lengerich, Chase, Beiler, & Darnell, 2006; Liff et al., 1991; Monroe, Ricketts, & Savitz, 1992; Smailyte & Kurtinaitis, 2008). Male patients diagnosed with breast cancer from 1988 to 2006 served as the study population, which consisted of 4,222 male subjects of all racial/ethnic groups. There were 3,836 MBC cases reported in metropolitan counties and 384 reported in nonmetropolitan counties (Table 1).
MBC Cases by Region
Note. MBC = male breast cancer.
The SEER summary staging system (for stage at diagnosis) was used to classify MBC cases as follows: (a) localized, in which the cancer is confined to organ of origin, which in this case means the breast (American Cancer Society, n.d.) but may be invasive within the breast; (b) regional, in which the cancer has spread beyond the breast to surrounding tissue or organs via the lymphatic system (American Cancer Society, n.d.); and (c) distant, in which the cancer has spread to distant areas in the body (American Cancer Society, n.d.). Tumor size was categorized into three levels: (a) tumors <2 cm, (b) tumors 2 to 5 cm, and (c) tumors >5 cm. Five-year survival rate was defined as the percentage of males with breast cancer who are alive for 5 years after they were diagnosed with MBC.
The data were extracted and analyzed using SPSS (PASW Version 18). Pearson’s chi-square test was used to evaluate statistically significant differences in proportions of stage or tumor size at diagnosis between the groups (metropolitan and nonmetropolitan). A p value <.05 was used for rejecting the null hypothesis of no difference. Data analysis for survival rate differences between the groups was performed using Kaplan–Meier curves for unadjusted analysis. Cox proportional hazards regression was used to estimate hazard ratios and their 95% confidence intervals (CIs) after adjusting for age, race, marital status, education level, stage at diagnosis, tumor size at diagnosis, median family income, and histological type.
The SEER program collects socioeconomic measures such as education level or median family income at the county level but not on an individual case level. Group characteristics were attributed to individual cases by ascribing education and median family income measures using categorical levels similar to those previously described by Yu (2009). Briefly, each education level from the SEER database (percent education level up to ninth grade, percent education level up to high school, and percent education level of at least a bachelor’s degree) was assigned a numeric value according to the following algorithm: a value of 1 (low education value) was assigned to the percent education level up to ninth grade, a value of 2 (medium education value) was assigned to the percent education level up to high school, and a value of 3 (high education value) was assigned to the percent education level of at least a bachelor’s degree. These values were multiplied by the percent county level education in SEER and aggregated across each case. The education level was then divided into tertiles (dividing the ordered distribution into 3), with each tertile representing one third of the population. Categorizing education level this way resulted in a distribution of the value scores such that each tertile had similar number of cases. For the income levels, the county-level median family income from the SEER database was sorted and divided into tertiles. Characteristics were attributed to individual cases by ascribing median family income level measure using categorical levels as described here. One tertile represented the median family income of up to $50,000, the second tertile represented the median family income from $51,000 to $64,000, and the third tertile represented the median family income >$65,000. Categorizing median family income this way resulted in a distribution of median family income by tertiles such that each tertile had similar number of MBC cases.
Results
The descriptive characteristics of MBC cases diagnosed from 1988 through 2006 are presented in Table 2. Presented in Tables 3 and 4 are the cross-tabulation results of region by stage at diagnosis and tumor size at diagnosis, respectively. A significant difference was not found between region and stage at diagnosis, indicating that the stage at diagnosis was not associated with the region. A significant difference was not found between region and tumor size at diagnosis, indicating that the tumor grade size at diagnosis was also not associated with the region.
Sociodemographic and Cancer Characteristics of Male Breast Cancer Cases Diagnosed in the United States From 1988 to 2006
County level data.
Broad groupings.
Stage at Diagnosis by Region Cross-Tabulation
Note. 0 cells (.0%) have expected count less than 5. The minimum expected count is 29.01.
Tumor Size at Diagnosis by Region Cross-Tabulation
Note. 0 cells (.0%) have expected count less than 5. The minimum expected count is 12.97.
The difference in the survival distributions between the two regions was statistically significant (p = .041). The log-rank test of equality for the categorical variables demonstrated that race, age, marital status, education level, stage, and median family income were significant. Tumor grade size and histology type were not significant. Cox proportional hazards regression analysis for the adjusted hazard ratio (HR) was significantly higher for region (HR = 1.185, 95% CI = [1.01, 1.40], p = .042). The reference category for region was metropolitan, for race White, for marital status married, for age 51 to 70 years, for education medium score (percent <high school), for median family income <$50,000, and for stage localized disease. As noted, the Kaplan–Meier and Cox regression analyses for differences between regions were significant; however, in the Cox regression analysis region was not statistically significant (HR = 0.984, 95% CI = [0.816, 1.185], p = .862).
Results identified the following predictor variables as contributing to poorer survival. Being 80 years and older contributed greatly to poorer survival (HR = 3.657, 95% CI = [3.053, 4.382], p = .000). Being Black contributed to poorer survival (HR = 1.356, 95% CI = [1.164, 1.580], p = .000). Being divorced/separated (HR = 1.674, 95% CI = [1.381, 2.030], p = .000) contributed 12% to poorer survival than being widowed and 10% to poorer survival rate than being single. Distant stage disease contributed the greatest to poorer survival (HR = 6.762, 95% CI = [5.766, 7.929], p = .000). Interaction terms were used to test whether each of these covariates modified the region effect. Interactions using race, age, marital status, distant stage, regional stage disease, education level, and median family income were not significant, indicating that these covariates did not modify the region effect. Assessment for confounding showed that age and education level may play a confounding role for the region effect in this study population. Race, marital status, and median family income were not confounders in this model. The final model included the main effect covariates and their interactions. In a stepwise fashion, the nonsignificant interactions were removed and the model was rerun until there were no more nonsignificant interactions remaining in the final model. Results of the final model for significant interactions are seen in Table 5.
Multivariate Analysis of Predictors of Poorer Survival of Male Breast Cancer Cases With Interaction Terms—Final Model
Discussion
Sufficient evidence exists that demonstrates that lack of awareness and lack of recognition of symptoms are important risk factors for late stage presentation of breast cancer in men. For example, Lengerich et al. (2006) observed that female rural residents were more often diagnosed at a late stage of breast cancer, suggestive of decreased participation in screening activities and also a lack of access to diagnostic or treatment services. Macdonald, Macleod, Campbell, Weller, and Mitchell (2006) identified lack of awareness of the severity of symptoms as the primary risk factor for patient delay in presenting with cancer. Smith, Cokkinides, and Eyre (2006) commented on the survival benefits when physicians raise awareness and reinforce the important role that early recognition of symptoms plays in cancer prevention.
Gamm, Castillo, and Pittman (2010) observed that significantly fewer primary care physicians are practicing in nonmetropolitan versus metropolitan areas (156 per 100,000 population vs. 280 per 100,000 population), and though they did not find studies that explicitly linked primary care physician shortages and mortality rates, they commented that diagnosis and treatment delays, which might be a consequence of poor access to care, might lead to mortalities that otherwise could have been avoided. Schroen, Brenin, Kelly, Knaus, and Slingluff (2005) noted that distance to facilities that offer radiation therapy significantly influenced receipt of mastectomy versus lumpectomy, and similarly, Jacobs, Kelley, Rosson, Detrani, and Chang (2008) observed a relationship between access to radiation therapy facilities and the decision to select mastectomy over breast conservation. Jacobs et al. (2008) proposed that the density of radiation oncologists in urban and rural areas might have an effect on the likelihood for breast conserving surgery versus mastectomy. Coory and Baade (2005) observed a significant excess of mortality for prostate cancer among regional and rural areas and noted that radical prostatectomies were 29% lower and prostate-specific antigen screening was 16% less in regional and rural areas when compared with capital cities. Coory and Baade (2005) concluded that the management of prostate cancer depends on where one lives and may be related to access to urologists.
Results for this study population showed that metropolitan/nonmetropolitan differences in tumor grade size and cancer stage at diagnosis were not statistically significant. These results corroborate those observed by Mitchell et al. (2006), who found no rural–urban differences of median tumor size for breast cancer among women. However, other studies reported results suggesting that the greater the distance (defining the rural region) from a cancer center the more likely a person is to be diagnosed with distant stage colorectal and lung cancer, and this trend persisted after adjusting for significant variables (Campbell et al., 2001). Liff et al. (1991) noted a rural–urban disparity in stage for a number of cancers. Liff et al. (1991) and Campbell et al. (2001) examined the odds of nonlocalized stage cancer in rural versus urban areas, an appropriate analysis since screening tests exist for cancers such as colorectal or prostate cancer. Yet there is no screening test for breast cancer in males, and this could account for no differences observed in this study between stage and tumor size at diagnosis for nonmetropolitan and metropolitan regions.
Studies have identified rural living with advanced stage diagnosis in women (Coughlin, Thompson, Seeff, Richards, & Stallings, 2002; Schootman, Kinman, & Farria, 2003), and the survival disparity between metropolitan and nonmetropolitan MBC cases observed in this study suggest an association between awareness or education about MBC and survival. The combination of factors such as late stage at presentation (Anderson et al., 2006), lack of awareness of symptoms (Macdonald et al., 2006), challenges in recruiting and retaining professional staff in rural health centers (Rosenblatt, Andrilla, Curtin, & Hart, 2006), and treatment disparities between rural and urban areas may explain the difference in survival rate between MBC in nonmetropolitan and metropolitan areas.
This study identified that men who reside in nonmetropolitan regions and are diagnosed with regional stage disease have a poor risk of survival. The outcome for males diagnosed with distant stage breast cancer might be the same whether they reside in nonmetropolitan or metropolitan areas. In other words, prognosis may be poor for males with distant stage breast cancer irrespective of the region they reside. The findings from this study indicate that an interaction between nonmetropolitan area and regional stage MBC was a significant predictor of poorer survival (HR = 1.634, 95% CI = [1.108, 2.409], p = .013). Although this study did not examine if differences in care or treatment exist between metropolitan and nonmetropolitan MBC cases, it suggests that improvements can be made in postcancer detection access and care.
There were several limitations for this current research. First, it was anchored entirely in the SEER database, which limited the ability to examine the effect of potentially important confounders such as lifestyle factors (e.g., smoking or alcohol) or family history because this information was not available in the SEER database. Second, although other rural/urban definitions exist, this study assigned SEER definitions of metropolitan and nonmetropolitan regions. Third, because the scope of the study involved a 19-year time frame (1988-2006), the migration of patients in and out of SEER registry areas to regions with access to health care could have provided for earlier diagnosis, thus potentially affecting stage at diagnosis, tumor size at diagnosis, and survival rate. Fourth, the study design was nonexperimental such that a causal relationship of metropolitan and nonmetropolitan to stage at diagnosis, tumor size at diagnosis, and survival rate for MBC could not be made. Last, this study focused on differences in breast cancer stage at diagnosis, tumor size at diagnosis, and survival between metropolitan and nonmetropolitan MBC cases across SEER registries, and it was not meant to examine all known risk factors associated with the occurrence of MBC.
Raising awareness and understanding of MBC could save the lives of many men. MBC is not only a disease that occurs in the United States, but it is also acknowledged as a serious disease in many countries around the world. Action should be taken to improve access, care, and thus the prognosis for males with breast cancer.
Footnotes
The author(s) declared no potential conflicts of interests with respect to the authorship and/or publication of this article.
The author(s) received no financial support for the research and/or authorship of this article.