Hypertension is an independent predictor of survival disparity between African-American and white breast cancer patients

Authors


Abstract

The objective of this study was to determine whether comorbidity, or pre-existing conditions, can account for some of the disparity in survival between African-American and white breast cancer patients. A historical cohort study was conducted of 416 African-American and 838 white women diagnosed with breast cancer between 1973 and 1986, and followed through 1999 in the Kaiser Permanente Northern California Medical Care Program. Information on comorbidity, tumor characteristics and breast cancer treatment was obtained from medical records, and Surveillance, Epidemiology and End Results, Northern California Cancer Center Registry. Associations between comorbidity and survival were analyzed with multiple Cox proportional hazards regression. Over a mean follow-up of 9 years, African Americans had higher overall crude mortality than whites: 165 (39.7%) versus 279 (33.3%), respectively. When age, race, tumor characteristics and breast cancer treatment were controlled, the presence of hypertension was associated with all cause survival [hazard ratio (HR) = 1.33, 95% confidence intervals (CI) 1.07–1.67] and it accounted for 30% of racial disparity in this outcome. Hypertension-augmented Charlson Comorbidity Index was a significant predictor of survival from all causes (HR = 1.32, 95%CI 1.18–1.49), competing causes (HR = 1.52, 95%CI 1.32–1.76) and breast cancer specific causes (HR = 1.18, 95%CI 1.03–1.35). In conclusion, hypertension has prognostic significance in relation to survival disparity between African-American and white breast cancer patients. If our findings are replicated in contemporary cohorts, it may be necessary to include hypertension in the Charlson Comorbidity Index and other comorbidity measures. © 2008 Wiley-Liss, Inc.

African-American breast cancer patients experience shorter survival than their white counterparts, and on average present with more advanced stage and hormone receptor-negative disease at younger ages.1 African-American ethnicity seems to be an independent predictor of poor breast cancer outcome, even after accounting for socioeconomic factors and inadequate health care access.2 Importantly, African-American women also tend to have poorer overall health, as reflected in greater prevalence of comorbidity.3 Furthermore, inadequately controlled comorbidities tend to adversely affect cancer treatment, as exemplified in the National Cancer Institute's Black White Cancer Survival Study, with breast cancer patients presenting with disproportionately increased rates of hypertension.4 A recent study by Tammemagi et al.5 reporting on outcomes among breast cancer patients followed for a median of 10 years at the Henry Ford Health System in Michigan showed that comorbidities such as hypertension and diabetes accounted for almost half of the overall survival disparity among African-American and white breast cancer patients. Thus, optimizing comorbidity measurement may help not only to uncover predictors of survival disparity between African-American and white breast cancer patients, but also opportunities to intervene for improved survival.

Several methods of comorbidity measurement have been developed and validated to some extent.6–9 Of the comorbidity indices developed to date, the Charlson Comorbidity Index (CCI) has been most extensively studied and deemed a valid and reliable method of measuring comorbidity for cancer research.10 Developed among patients admitted to an emergency department with respect to mortality at 1 year of follow-up as a function of comorbidity, the CCI was subsequently validated in a cohort of hypertensive patients (n = 685).11 Each condition included in the original CCI conferred an independent relative risk of death of 1.2. The comorbid conditions in the original CCI were weighted so that those leading to relative risks between 1.2 and <1.5 were scored as 1; between 1.5 and <2.5 as 2; between 2.5 and <3.5 as 3; and 2 conditions with a relative risk of 6 or more were scored as 6. The total scores calculated by tallying these weighted scores range from 1 to 6 (0 if the comorbidity is absent), are then collapsible into 4 summary categories: 0, 1–2, 3–4 and 5 points.

Recent evidence suggests that the CCI omits several important prognostic comorbidities for cancer including hypertension.5, 12 High blood pressure, while unrepresented in the CCI, is a highly prevalent condition, particularly among African Americans, and it has been found to affect mortality.5 These pieces of evidence thus provoke the question: Can high blood pressure account for some of the disparity in survival between African-American and white breast cancer patients? Of note also is the finding that hypertension has been clearly related to cancer mortality.13 In the present study, therefore, we first sought to evaluate the prognostic impact of hypertension. As a second step, we examined the value of augmenting the CCI with hypertension, i.e., creating the hypertension-augmented Charlson Comorbidity Index (hCCI), both in terms of its prognostic ability compared with the CCI, as well as its ability to account for survival disparity between African-American and white patients.

Material and methods

Study design and population

A historical cohort study was conducted of 416 African-American and 838 white women as part of the Kaiser Permanente Northern California Medical Care Program (KPMCP) Comorbidity Study.14 The patients were diagnosed with histologically confirmed invasive breast cancer between 1973 and 1986. All women were residents of the San Francisco Bay Area, had a known stage of disease and course of treatment. KPMCP, a group practice, prepaid health plan, is one of the largest health care providers in northern California. Non-Hispanic whites at KPMCP make up 71% of the enrollees versus 66% in the Bay Area Metropolitan Statistical Area, Hispanics 10% versus 14%, African Americans 7% versus 9% and Asians 8% versus 11%. KPMCP enrollees are also representative of the general population for many other demographic and socioeconomic categories, with the exception of the very high and very low ends of the economic spectrum.15

Case identification

Cases were identified by the Northern California Cancer Center's Surveillance, Epidemiology and End Results (SEER) Registry, a population-based cancer surveillance system funded by the National Cancer Institute.

Outcome variables

There were three measures of survival: overall survival, breast cancer specific survival and competing cause survival. Cause of death data, classified by International Classification of Diseases (ICD) codes version 9, were obtained from the SEER Registry. ICD codes 174.0–174.9 represented breast cancer deaths and other ICD codes represented competing causes of death. Additionally, we evaluated breast cancer recurrence (new breast cancer events), which include local/regional cancer recurrence, distant recurrence/metastasis and development of a contralateral breast primary.

Predictor variables

Demographic and clinical variables, taken from the KPMCP and SEER Registry files, include age at breast cancer diagnosis, race (non-Hispanic African American or white), tumor stage, histology, tumor size, estrogen receptor status, type of surgery, type of adjuvant treatment and information on breast cancer recurrence. Breast cancer stage at diagnosis was defined as localized, regional, distant or unknown; in situ cases were excluded. Additional treatment information was obtained from medical records and coded as a categorical variable (no surgery, partial mastectomy only, partial mastectomy plus adjuvant therapy, modified radical therapy only, modified radical therapy only plus adjuvant therapy). Adjuvant therapy included radiation, chemotherapy, and hormonal therapy; radiation, chemotherapy and hormonal therapy were treated as separate variables. Comorbid conditions were abstracted directly from the medical charts. Comorbidity data evaluated in this study included data collected from 1 year before diagnosis to the time of diagnosis. Ten percent of patients' records were re-abstracted and evaluated for reliability. All data were double entered into a database and reviewed for outliers by range and logic edits. Medical abstractors from KPMCP reviewed the medical records of all breast cancer cases in the cohort to code comorbidity and other data as well as the presence and the time of occurrence of 32 nonbreast cancer medical conditions, which consist of the CCI conditions and those suggested to be important in a previous study by Satariano and Ragland16 (Table AI shows the CCI conditions and their weights). The CCI supplemented with hypertension, which was assigned the weight of 1, is hereafter referred to as the hypertension-augmented Charlson Comorbidity Index (hCCI). It is important to note that the CCI conditions and weights in this study were in keeping with recommendations of Charlson and colleagues,10 resulting in the assignment of a comorbidity score to each patient. Therefore, the only difference between CCI and hCCI was that hCCI included hypertension alongside the CCI conditions.

Institutional Review Board approvals from the University of California, San Francisco were obtained for the use of human subject records in this study.

Statistical analysis

Descriptive statistics, including patient characteristics by race, were evaluated using contingency table analyses, Fisher exact test and nonparametric tests for trend for ordinal data17 and t-tests for continuous data. Logistic regression odds ratios (ORs) and 95% confidence intervals (CIs) were used to evaluate associations between predictors and dichotomous outcomes including factors predicting loss to follow-up. Survival to death from all causes, breast cancer and competing causes were examined using Cox proportional hazard ratios (HRs) and 95% CIs, guided by a priori considerations18 and were aided by backward stepwise elimination. Kaplan-Meier plots were used to graphically examine associations between study variables and survival or breast cancer recurrence. The effects of race, selected comorbid conditions, age, stage, size, estrogen receptor status and breast cancer treatment were examined in relation to survival from all causes, breast cancer and competing causes. In survival regression analyses, proportional hazard assumptions were tested graphically and statistically and were met for all presented models. In multivariate models, interaction terms for hypertension and race were considered. The alpha error was set at 0.05 and all reported p values are two-sided. Stata version 9.0 software (Stata Corporation, College Station, TX) was used to prepare statistics and figures.

Results

Baseline characteristics and loss to follow-up

The distributions of baseline characteristics for selected variables by race are presented in Table I. African Americans were significantly younger than the whites (means 53.0 vs. 58.4 years, p < 0.0001). African Americans were also significantly more likely to be unmarried compared with whites: 168 (40.4%) vs. 282 (33.7%) (p = 0.03). They tended to be diagnosed at a higher tumor stage compared with whites, as reflected by 176 (44.2%) versus 284 (35.2%), respectively, presenting with regional disease (p < 0.0001). Furthermore, estrogen receptor-positive tumors were less frequent among African-American patients than among their white counterparts: 108 (26.1%) versus 274 (32.7%); p < 0.01. In addition, African Americans had a higher body mass index (BMI) than whites (averages were 30.3 vs. 27.3, respectively, p < 0.0001).

Table I. Sociodemographic, Tumor and Clinical Characteristics of the Study Cohort, by Race
 African American (n = 416) n (%)White (n = 838) n (%)P*
  • *

    P values were calculated using the chi-squared test.

Age, years   
 ≤4079 (19.0)78 (9.3)<0.0001
 >40 to ≤50106 (25.5)165 (19.7) 
 >50 to ≤60106 (25.5)211 (25.2) 
 >60 to ≤7090 (21.6)238 (28.4) 
 >70 to ≤8029 (7.0)108 (12.9) 
 >806 (1.4)38 (4.5) 
Unmarried168 (40.4)282 (33.7)0.03
Tumor stage  <0.001
 Localized200 (50.3)493 (61.1) 
 Regional176 (44.2)284 (35.2) 
 Distant22 (5.5)27 (3.4) 
Breast cancer treatment  0.09
 Breast conserving surgery18 (4.4)19 (2.3) 
 Lumpectomy with node removal334 (80.9)650 (78.8) 
 Modified radical mastectomy21 (5.1)56 (6.8) 
 Other surgery or unknown40 (9.7)100 (12.1) 
 Chemotherapy178 (43.2)277 (33.1)<0.0001
 Radiation therapy157 (37.9)311 (37.1)0.78
 Hormone therapy134 (32.3)256 (30.6)0.53
Estrogen receptor status   
 Positive108 (26.1)274 (32.7)<0.01
 Negative118 (28.5)181 (21.6) 
 Unknown188 (45.4)382 (45.6) 
Body mass index30 (6.7)27 (5.5)<0.0001

Patients were followed for a mean of 8.9 years (median 8.4 years; range 0.01–21.1 years). Of 1254 patients included in the study, 21 (2%) were lost to follow-up.

Survival and recurrence data

Significantly more African Americans than whites died during the follow-up period: 165 (39.7%) versus 279 (33.3%, p = 0.03; Table II). Of African Americans and whites, respectively, 119 (28.6%) and 181 (21.6%) died of breast cancer and 46 (11.1%) and 98 (11.7%) died from competing causes, indicating higher overall crude mortality for African Americans. Tumor recurrence was also observed more frequently among African Americans than whites: 207 (59.1%) versus 143 (40.9%), p < 0.0001. Although there were no significant racial differences in the distribution of the Charlson Comorbidity Index (p = 0.9), proportionately more African Americans than whites experienced elevated blood pressure, specifically 175 (43.3%) versus 233 (28.3%, p < 0.0001), respectively. However, the interaction between hypertension and race in the regression models was not statistically significant, which justifies the augmentation of the CCI with hypertension.

Table II. Distribution of Comorbidity and Survival by Race
 African American (n = 416)White (n = 838)P*
  • *

    P values were calculated using the chi-squared test.

Charlson Comorbidity score   
 0274 (65.9)540 (64.4)0.9
 192 (22.1)195 (23.3) 
 230 (7.2)62 (7.4) 
 315 (3.6)29 (3.5) 
 4–95 (1.2)12 (1.4) 
Specific non-CCI comorbidity   
 Hypertension175 (43.3)233 (28.3)<0.0001
Survival   
 All-cause deaths165 (39.7)279 (33.3)0.03
 Breast cancer-specific deaths119 (28.6)181 (21.6)<0.01
 Competing causes deaths46 (11.1)98 (11.7)0.7
Breast cancer recurrence207 (59.1)143 (40.9)<0.0001

Since African Americans were significantly younger than whites, age was controlled in all analyses. In multivariate models, additional adjustments were also performed for tumor biology (tumor stage, tumor size, estrogen receptor status) and treatment (surgery, chemotherapy, hormonal and radiation therapy). The most complex multivariate model that included race, age, comorbidity, tumor biology and treatment as predictors was based on the total of 1071 subjects, who had complete data on variables of interest. Completeness of data was comparable between the two racial groups: 85.2% (714 of 838) among whites and 85.8% (357 of 416) among African Americans.

Hypertension

All cause survival

Hypertension was associated with all cause survival not only after controlling for age and race, but also after taking tumor characteristics and breast cancer treatment into consideration (HR = 1.33, 95%CI 1.07–1.67; Table III). Estimates were not markedly changed when additional adjustments were performed for the CCI, BMI and marital status (data not shown). Comparing the multivariate HRs for race unadjusted and adjusted for hypertension (HR = 1.33, 95%CI 1.07–1.66 vs. HR = 1.23, 95%CI 0.98–1.54, respectively), we found that this single comorbidity alone explained 30.3% of racial disparity in all cause survival.

Table III. Prediction and Explanation of Disparity in Survival by Hypertension
 HR (95%CI)
  • HR, hazard ratio; CI, confidence interval.

  • 1

    Adjustments were made for tumor characteristics (stage, size, estrogen receptor status), breast cancer treatment (chemotherapy, surgery, radiation therapy, hormonal therapy); further additional adjustment for the Charlson comorbidity did not significantly change the estimates; additional adjustment for body mass index did not significantly change the estimates; additional adjustment for marital status did not significantly change the estimates.

All cause survival (n = 1071; deathsn = 372)HR (95%CI)
Hazard Ratio for hypertension adjusted for age and race1.48 (1.17–1.85)
Hazard ratio for hypertension adjusted for age, race and other covariates11.33 (1.06–1.67)
Hazard ratio for race adjusted for age and other covariates11.33 (1.07–1.66)
Hazard ratio for race adjusted for hypertension, age and other covariates11.23 (0.98–1.54)
Proportion of racial disparity explained by hypertension, %30.3%
Breast cancer survival (n = 1071; deathsn = 254)
Hazard ratio for hypertension adjusted for age and race1.43 (1.08–1.91)
Hazard ratio for hypertension adjusted for age, race and other covariates11.22 (0.91–1.62)
Hazard ratio for race adjusted for age and other covariates11.40 (1.08–1.82)
Hazard ratio for race adjusted for hypertension, age and other covariates11.32 (1.00–1.73)
Proportion of racial disparity explained by hypertension, %20.0%
Competing cause survival (n = 1071; deaths = 118) 
Hazard ratio for hypertension adjusted for age and race1.68 (1.14–2.47)
Hazard ratio for hypertension adjusted for age, race and other covariates11.68 (1.14–2.48)
Hazard ratio for race adjusted for age and other covariates11.10 (0.74–1.67)
Hazard ratio for race adjusted for hypertension, age and other covariates10.98 (0.64–1.48)

Breast cancer specific survival

After taking age and race into account, hypertension was associated with breast cancer specific survival (HR = 1.43, 95%CI 1.08–1.91; Table III), but after further adjustment for tumor characteristics and breast cancer treatment, the effect was no longer statistically significant (HR = 1.22, 95%CI 0.91–1.62). Further adjustment for the CCI did not result in significant estimate changes. We compared the multivariate HRs for race unadjusted and adjusted for hypertension (HR = 1.40, 95%CI 1.08–1.82, vs. 1.32, 95%CI 1.00–1.73, respectively), which indicates that elevated blood pressure accounted for 20.0% racial disparity in breast cancer specific survival.

Competing cause survival

Hypertension was associated with competing cause survival in multivariate models (HR = 1.68, 95%CI 1.14–2.48; Table III). However, as shown in Table III, African Americans were not at a significantly increased risk of competing cause death compared with whites.

Recurrence

Recurrence was associated with increased risk of death from breast cancer (HR = 2.99; 95%CI 2.27–3.93). We next examined the possibility of differential recurrence estimates between African Americans and whites who suffered from elevated blood pressure versus those who did not (Table IV). Importantly, when age, tumor characteristics and breast cancer treatment were controlled, elevated blood pressure was associated with recurrence among African Americans (HR = 1.60, 95%CI 1.07–2.40) but not among their white counterparts (HR = 1.22, 95%CI 0.86–1.73; Table IV, Fig. 1). Estimates were not markedly changed when additional adjustments were performed for the CCI, nor when we estimated hypertension effects in a subgroup of women whose CCI was zero. Similar trends were found in relation to breast cancer deaths: HRs were 1.49, 95%CI 0.95–2.34 among African Americans versus 1.24, 95%CI 0.85–1.80 among whites.

Figure 1.

Kaplan-Meier plot of recurrence by hypertension. (a) African-American patients (n = 357)* and (b) white patients (n = 714)*. *see Table IV for HRs and 95%CIs estimates.

Table IV. Race Stratified Hazard Ratios for the Association Between Hypertension and Breast Cancer Outcomes
 African American, HR (95%CI)White, HR (95%CI)
  • HR, hazard ratio; CI, confidence interval.

  • 1

    Tumor characteristics (stage, size, estrogen receptor status), breast cancer treatment (chemotherapy, surgery, radiation therapy, hormonal therapy); additional adjustment for the Charlson comorbidity did not significantly change the estimates; additional adjustment for body mass index did not significantly change the estimates; additional adjustment for marital status did not significantly change the estimates

Breast cancer recurrence(n = 354; recurrences = 125)(n = 714; recurrences = 177)
Hazard ratio for hypertension adjusted for age2.04 (1.36–3.05)1.24 (0.88–1.76)
Hazard ratio for hypertension adjusted for age and other covariates11.60 (1.07–2.40)1.22 (0.86–1.73)
Breast cancer specific survival(n = 354; deaths = 103)(n = 714; deaths = 148)
Hazard ratio for hypertension adjusted for age1.80 (1.16–2.81)1.26 (0.86–1.83)
Hazard ratio for hypertension adjusted for age and other covariates11.49 (0.95–2.34)1.24 (0.85–1.80)

Charlson comorbidity index versus hypertension-augmented Charlson comorbidity index

We next evaluated the predictive validity of the hCCI compared with the CCI.

All cause survival

The all cause survival estimates by CCI and hCCI for African-American and white breast cancer patients are presented in Table V. Adjusted for age, the all-cause survival HRs for both hCCI and CCI were identical (HR = 1.32 per 1 of 4 levels of change, 95%CI 1.18–1.46 vs. 1.18–1.49, respectively), which after further adjustment for race, tumor characteristics and breast cancer treatment remained significant at 1.34 (95%CI 1.22–1.49) and 1.39 (95%CI 1.24–1.56), respectively. We next evaluated racial disparity in all cause survival. African Americans were 1.40 times more at risk of all cause death than whites (95%CI 1.13–1.73). With the CCI in the model, African Americans still showed a significantly increased risk of all cause mortality (HR = 1.24, 95%CI 1.00–1.54). However, when the hypertension-augmented CCI was modeled instead of the original CCI, the disparity was no longer significant (HR = 1.17, 95%CI 0.94–1.46).

Table V. Prediction and Explanation of Disparity in Survival by Charlson Comorbidity Index (CCI) versus Hypertension Augmented Charlson Comorbidity Index (hCCI)
 CCI, HR (95%CI)hCCI, HR (95%CI)
  • HR, hazard ratio; CI, confidence interval.

  • 1–3, 1–4, 1

    Tumor characteristics (stage, size, estrogen receptor status), breast cancer treatment (chemotherapy, surgery, radiation therapy, hormonal therapy).

  • 1–3, 1–4, 2

    Additional adjustment for body mass index did not significantly change the estimates.

  • 1–3, 1–4, 3

    Additional adjustment for marital status did not significantly change the estimates.

  • 1–4

    4Compared with the age-adjusted African American versus white hazard ratio = 1.40 (95%CI 1.13–1.73).

  • 5

    Compared with the age-adjusted African American versus white hazard ratio = 1.48 (95%CI 1.15–1.90).

  • 6

    Compared with the age-adjusted African American versus white hazard ratio = 1.18 (95%CI .79–1.77).

All cause survival (n = 1071; deaths, n = 372)  
Hazard ratio for comorbidity per unit increase adjusted for age1.32 (1.18–1.49)1.32 (1.18–1.46)
Hazard ratio for comorbidity adjusted for age, race and other covariates1–31.39 (1.24–1.56)1.34 (1.22–1.49)
Hazard ratio for race adjusted for age, CCI vs. hCCI and other covariates1–41.24 (1.00–1.54)1.17 (0.94–1.46)
Breast cancer survival (n = 1071; deaths, n = 254)  
Hazard ratio for comorbidity per unit increase adjusted for age1.11 (0.94–1.32)1.18 (1.03–1.35)
Hazard ratio for comorbidity adjusted for age, race and other covariates1–31.17 (0.98–1.39)1.17 (1.02–1.35)
Hazard ratio for race adjusted for age, CCI vs. hCCI and other covariates1,2,3,51.38 (1.06–1.79)1.33 (1.02–1.74)
Competing cause survival (n = 1071; deaths, n = 118)  
Hazard ratio for comorbidity per unit increase adjusted for age1.57 (1.33–1.85)1.52 (1.32–1.76)
Hazard ratio for comorbidity adjusted for age, race and other covariates1–31.61 (1.36–1.90)1.56 (1.35–1.81)
Hazard ratio for race adjusted for age, CCI vs. hCCI and other covariates1,2,3,61.10 (0.73–1.66)0.98 (0.65–1.48)

Breast cancer specific survival

Age-adjusted breast cancer survival HR for hCCI was significant at 1.18 (95%CI 1.03–1.35) whereas the corresponding HR for CCI was slightly increased but not statistically significant (HR = 1.11; 95%CI 0.94–1.32); further adjustments for race, stage, size, estrogen receptor status and treatment did not significantly change the estimates (Table V).

African Americans were more likely to die from breast cancer than their white counterparts (age-adjusted HR = 1.48, 95%CI 1.15–1.90). The HR for race with CCI in the model was 1.38 (95%CI 1.06–1.79). When hCCI was modeled instead of CCI, the HR for race dropped but remained statistically significant (HR = 1.33; 95% CI 1.07–1.75).

Competing cause survival

Neither hCCI nor CCI were associated with significantly worse age-adjusted survival from competing causes (HR = 0.98, 95%CI 0.65–1.48 vs. HR = 1.10, 95%CI 0.73–1.66, respectively; Table V).

Discussion

In the present study, African-American breast cancer patients exhibited shorter breast cancer specific survival and increased risk of breast cancer recurrence than their white counterparts, but there was little or no African American versus white disparity in relation to competing cause survival. Hypertension-augmented CCI was better than the original CCI at explaining disparity in overall survival, after taking into account age, tumor characteristics and breast cancer treatment. Further, hypertension was an independent predictor of survival from all causes and it accounted for almost a third of racial disparity in this outcome, as shown in Table III. Importantly, when other potential confounders such as age, tumor biology and breast cancer treatment were taken into consideration, hypertensive African Americans still showed a 1.60 times higher recurrence estimates than their nonhypertensive counterparts (Table IV). Yet, such a relationship was not observed among whites.

It is notable that a related study conducted in the 1970s has indeed linked hypertension with cancer mortality in a sample of largely white women,13 but, to our knowledge, our report may be the first to report the link with recurrence in a new population. Our ability to adjust for tumor biology, breast cancer treatment and comorbidity in the analysis was a strength; however, we were unable to evaluate other potential confounders, particularly the effect of antihypertensive treatment.19 The adverse effect of hypertension in our study was confined to African-American patients. If hypertension is truly causally associated with breast cancer, the next important step will be to elucidate the biologic mechanisms linking hypertension and malignancy. For example, increased expression of inositol triphosphate and cytosolic calcium have been hypothesized to be involved in the pathogenesis of hypertension and in the early events of cell proliferation that are activated by endogenous oncogenes.20 Another study has pointed to aberrant carcinogen binding to deoxyribonucleic acid in lymphocytes of hypertensive patients.21 Cell death via apoptosis can also influence the growth of vascular smooth muscle cells, and related aberrations have also been identified in hypertension.22 Finally, neurohormones such as angiotensin II, catecholamines, vasopressin, insulin and growth hormone not only regulate blood pressure but also have a mitogenic effect.20

The CCI's failure to capture hypertension had a stronger effect on all cause survival among African-American patients than on their white counterparts, as shown in Table V. Importantly, the disparity in all cause survival appeared to stem from African American-white differentials in breast cancer specific survival rather than those from other causes. Furthermore, at least partially, our findings lend support to the possibility that, in certain populations, comorbidity also affects cancer mortality.23, 24 Although preliminary, our results nevertheless suggest that hypertension is an important comorbidity to consider in the context of research on racial disparity in breast cancer and our findings warrant its inclusion in comorbidity measures like the CCI. It is notable that the original CCI, while also associated with survival from all cause and competing causes, was not associated with breast cancer specific survival. Given that the CCI was developed in a predominantly white population,7 in which a lower prevalence of hypertension is well established, it is understandable why hypertension did not become part of the CCI. Furthermore, use of a short follow-up in that study to evaluate mortality meant that hypertension may not have had a chance to demonstrate its long-term adverse effects.

The current study had design strengths and limitations. It included a large cohort of breast cancer patients from a Northern Californian health maintenance organization (whose members are representative of the general population with respect to most socioeconomic categories15) with a long follow-up. Comorbidity data were collected through medical chart reviews, which are deemed more reliable than administrative databases.25 As the data indicated that African-American breast cancer patients were significantly younger than their white counterparts, age was controlled in the analysis, in addition to tumor and treatment characteristics. However, specific data on socioeconomic circumstances were not available in the charts and, therefore, were not evaluated or controlled in our analysis. Indeed, poverty is associated with unfavorable breast cancer outcomes independent of race,3 and may lie on the pathway between adverse socioeconomic circumstances, comorbidity and malignancy.26 In addition, our study sample consisted of women who resided in a single geographic location (San Francisco Bay Area), and thus, our findings may lack external validity. Another potential limitation is that the present data are dated. It is possible that secular trends in blood pressure management could modify the relations observed in more contemporary cohorts. In addition, medications that may address (control) the comorbid conditions we examined were not abstracted.

In conclusion, the unique finding of our study is that hypertension has prognostic significance in relation to survival among breast cancer patients. In addition, hypertension accounts for approximately 30% of the survival disparity between African-American and white breast cancer patients. Currently, hypertension is not part of the widely used Charlson Comorbidity Index. Our findings suggest hypertension may influence prognosis and survival. If these results are validated in contemporary cohorts, it may be necessary to include hypertension in the Charlson Comorbidity Index and other comorbidity measures.

Acknowledgements

The authors are grateful to Dr. Hyman Muss, for thoughtful comments on an earlier draft of this article. This study was funded by the Windber Research Institute (Executive Director: Dr. Michael Liebman) and US Department of Defense (Center of Excellence in Breast Cancer Care; PI Dr. Laura Esserman). Data were derived from a previous study performed at the Division of Research Kaiser Permanente Medical Care Program in Oakland, California, where Dr. Robert Hiatt was Assistant Director for Epidemiology. The funding sources did not influence the design and conduct of the study.

Appendix

AI

Table AI. Comorbidities Abstracted the KPMCP Comorbidity Study1
ComorbidityCharlson Comorbidity Index weight
  • 1

    CCI conditions were included in the present study.

  • 2

    Not part of the CCI but included in the present study; assigned weight 1.

  • 3

    Not part of the Charlson Comorbidity Index and not included in the analysis.

Hypertension02
Osteoarthritis1
Endocrine disease (not diabetes)03
Peripheral vascular disease1
Diabetes (mild)1
Diabetes (moderate)1
Diabetes (severe)2
Renal disease (mild)0
Renal disease (moderate)2
Renal disease (severe)2
Liver disease (mild)1
Liver disease (moderate)3
Liver disease (severe)3
Arthritis1
Arrythmia03
Peptic ulcer1
Gastrointestinal bleeding03
Residual joint or skeletal problem03
Glaucoma03
Angina03
Cerebrovascular disease1
Osteoporosis1
Congestive heart disease1
Pulmonary disease (mild)0
Pulmonary disease (moderate)1
Pulmonary disease (severe)1
Myocardial infarction1
Valvular disease03
Dementia1
Other dementia1
Rheumatological disease1
Coagulopathy2
Tinnitus03
Heart disease, not otherwise specified03
Macular degeneration03
Other nonmetastatic cancer2
Other metastatic cancer6
AIDS6
Blind NOS03
Deaf NOS03

Ancillary