Abstract
Summary
Risk for falls and fractures increases after breast cancer or other cancer diagnosis in postmenopausal women. Factors other than falls may be the major causes for the increased fracture risk.
Introduction
Cancer treatment and prognosis may have detrimental effects on bone health. However, there is a lack of prospective investigations on fracture risk among incident cancer cases.
Materials and Methods
In this study, postmenopausal women (N = 146,959) from the Women's Health Initiative prospective cohort, who had no cancer history at baseline were followed for up to 9 years and classified into no cancer, incident breast cancer (BC) and incident other cancer (OC) groups. The main outcomes measured were incident fractures and falls before and after cancer diagnosis. Hazards ratios (HR) and 95% confidence intervals (CI) were computed from Cox proportional hazards models.
Results
While hip fracture risk before a cancer diagnosis was similar between the no cancer and cancer groups, hip fracture risk was significantly higher after BC diagnosis (HR = 1.55, CI = 1.13-2.11) and the elevated risk was even more notable after OC diagnosis (HR = 2.09, CI = 1.65-2.65). Risk of falls also increased after BC (HR = 1.15, CI = 1.06-1.25) or OC diagnosis (HR = 1.27, CI = 1.18-1.36), but could not fully explain the elevated hip fracture risk. Incident clinical vertebral and total fractures were also significantly increased after OC diagnosis (p < 0.05).
Conclusions
Postmenopausal women have significantly elevated risks for falls and fractures after a cancer diagnosis. The causes for this increased risk remained to be investigated.
Keywords: Breast cancer diagnosis, Cancer diagnosis, Falls, Fractures, Postmenopausal women, Prospective cohort
Introduction
Approximately four in ten white women age 50 or older in the United States will experience a hip, spine or wrist fracture sometime during the remainder of their lives [1]. Even when women have a higher bone density and lower fracture risk before their breast cancer diagnosis [2, 3], bone loss may be accelerated with cancer treatment and the presence of cancer itself [4-7]. Recent findings from the Women's Health Initiative (WHI) study [8] have shown a higher fracture risk among women who reported a breast cancer diagnosis history at the WHI enrollment in comparison to postmenopausal women without a history of cancer. However, fracture risk among incident postmenopausal breast cancer cases, as well as the differences in the fracture risk before and after a breast cancer diagnosis in breast cancer survivors are still unknown. As a diagnosis of other cancers may involve some of the same risk factors, such as chemotherapy use, a higher fracture risk after cancer diagnosis may not be unique to breast cancer [9]. Falls are a major risk factor for fractures since 90% of fractures result from a fall [10]. The susceptibility of women to falls related to a diagnosis of breast cancer or other cancer is also understudied.
This objective of this study was to investigate the risk of falls and fractures following a cancer diagnosis in postmenopausal women. We hypothesized that (1) in comparison to their postmenopausal peers, women who are diagnosed with invasive breast cancer have significantly higher risk for fracture after their breast cancer diagnosis but not before the diagnosis; (2) fracture risk increases in postmenopausal women diagnosed with other cancer; (3) the risk of falls elevates after, but not before, women are diagnosed with either breast cancer or other cancer; and (4) the cancer-related increase in fall does not fully explain the higher fracture risk associated with a cancer diagnosis.
Materials and Methods
Participants
This report was based on the data collected in the WHI. Details regarding the WHI study design and recruitment method have been published elsewhere [11, 12]. The primary aim of the WHI was to investigate risk factors of and intervention effects on cancers, cardiovascular diseases, and osteoporotic fractures, in a large multiethnic cohort of postmenopausal women in the United States. WHI included four clinical trials (CT) and one observational study (OS). Participants were recruited at 40 WHI clinic centers across the nation. The major inclusion criteria were being postmenopausal, age between 50 and 79, not participating in other clinical trials at the time of enrollment, and not likely to die or move within 3 years. Additional exclusion criteria were applied to the participants in the CT. Women who had no history of cancers before the WHI enrollment were included in our analyses. Among them, 81,601 women were in the OS and 65,358 in the CT. The average (SD) follow-up time, from randomization in the CT or enrollment into the OS, was 7.8 (1.7) years. In our analysis, women were classified into three groups based on their follow-up result: no cancer, incident invasive breast cancer (BC), and other incident cancer (OC).
Data collection
At baseline, participants completed a set of questionnaires regarding demographics, medical history, health status, reproductive history, medication, physical activity, dietary intake, fracture history and other lifestyle factors. Height and weight measurements were also conducted during the initial clinical visit. Follow-up information on selected health outcomes, medication, and lifestyle factors was collected from self-administered questionnaires annually in the OS participants and biannually in the CT participants. Self-reported incident cancer diagnoses were adjudicated by physician reviews of medical records and pathology reports at each clinical center. Information on estrogen receptors from pathology reports and selected breast cancer treatment was used in some analyses.
Among CT participants, all self-reported fractures were confirmed by trained physician adjudicators reviewing medical records and reports. For OS participants, only hip fractures were adjudicated and the remaining fractures were based on self-reported information. No study X-rays were examined for the presence of sub-clinical vertebral fractures. In a WHI validation study, agreement between self-reported fracture and clinical chart review was over 70% on average, although it varied by skeletal sites [13]. In this report, the total fracture category included all clinical fractures excluding factures of face, skull, toe, finger, rib and chest. Data on the number of falls during the follow-up were ascertained from annual or semiannual self-reports.
Statistical analysis
SAS/STAT (version 9.1.3 for Windows, SAS Institute, Cary, NC, USA) program was used. All the tests were two tailed and at a significance level of 0.05. Age-adjusted annual incidence rates were calculated for hip, clinical vertebral, forearm, and total fractures by cancer group and by pre- and post-diagnosis period. The same approach was used to compute age-adjusted annual incidence rates for falls.
Primary analyses used time-to-event methods based on the Cox proportional hazards model, with time from randomization in the CT, and time from enrollment in the OS to time of fracture or fall as the basic time variable. The main exposure variables, incident breast cancer and incident other cancer, were treated as time-varying dichotomous covariates. For example, the covariate breast cancer(t) is equal to 0 for t ≤ time_to_incident_breast_cancer and equal to 1 for t > time_to_incident_breast_cancer, where t is survival time [14]. Women who were diagnosed with breast cancer in situ were included in the Cox proportional hazards model and censored after their breast cancer in situ diagnosis.
The possibility that baseline characteristics may confound the relationship of exposure to the outcome variables was accommodated by including baseline risk factors of fracture previously described [15], we used a slightly finer adjustment by differentiating between estrogen +progesterone and estrogen-alone use and instead of adjusting for body mass index, we adjusted for height (linear and quartiles) and weight (linear and quartiles). We also included family history of cancer (yes/no), Gail risk (linear and quartiles), total calcium intake (linear and quartiles), total vitamin D intake (linear and quartiles), total energy (linear and quartiles), physical activity (linear and quartiles), alcohol intake (linear and quartiles), diet modification (DM) treatment assignment, hormone therapy (HT) assignment, calcium/vitamin D (CaD) supplementation assignment, and history of fracture after the age of 55 and allowed baseline hazard functions to vary by age (5 years).
For the secondary analyses, we divided incident breast cancer by receptor status and differentiated ER+ breast cancer by users and non-users of tamoxifen; these exposure variables were also treated as time varying dichotomous covariates. We investigated whether the number of falls (≥2) and change in height from baseline were intermediate outcomes or independent risk factors by including them as time-varying covariates [16]. To explore possible explanations for an increased fracture or fall risk after a cancer diagnosis, a descriptive analysis was conducted among participants in the WHI CT to show differences in using osteoporotic-related medications and selected cancer treatments.
Results
This study included a total of 146,959 women. Among them, 132,840 had no cancer diagnosis at baseline or during the follow-up; 4,804 were diagnosed with invasive breast cancer; 1,073 with non-invasive breast cancer (in situ); and 8,242 with other types of cancers, including colorectal (16.3% of other cancers), lung (13.1%), endometrial (9.4%), melanoma of the skin (9.2%) and non-Hodgkins lymphoma (6.6%). Given the large sample size, most of the baseline characteristics were statistically different between groups (Tables 1 and 2). However, these groups were comparable in the percentage of parents who ever had a broken bone after age 40 (p < 0.95). The percentage of women with a fracture after age 55 in the incident invasive BC group (11.7%) was similar to the non-cancer group (12.0%), but it was higher in the OC group (14.6%, p < 0.001).
Table 1.
Baseline characteristics in the WHI CT and OS (n=146,959a) – Categorical Variables
| No-cancer (N=132,840) | Incident Other Cancer (N=8242) | Incident Invasive BC (N=4804) | ||
|---|---|---|---|---|
| N (%) | N (%) | N (%) | P-Valueb | |
| Ethnicity | ||||
| White | 10,854 (81.7) | 7152 (86.8) | 4186 (87.1) | |
| Black | 12,431 (9.4) | 608 (7.4) | 328 (6.8) | |
| Hispanic | 5746 (4.3) | 196 (2.4) | 124 (2.6) | <0.001 |
| American Indian | 594 (0.4) | 30 (0.4) | 16 (0.3) | |
| Asian/Pacific Islander | 3649 (2.7) | 148 (1.8) | 105 (2.2) | |
| Unknown | 1877 (1.4) | 108 (1.3) | 45 (0.9) | |
| Education | ||||
| 0-8 years | 2272 (1.7) | 119 (1.5) | 38 (0.8) | |
| Some high school | 4887 (3.7) | 328 (4.0) | 139 (2.9) | <0.001 |
| High school diploma/GED | 22,859 (17.3) | 1477 (18.1) | 760 (15.9) | |
| School after high school | 49,952 (37.9) | 3117 (38.1) | 1740 (36.5) | |
| College degree or higher | 51,870 (39.3) | 3139 (38.4) | 2088 (43.8) | |
| Number of pregnancies | ||||
| Never pregnant | 11,794 (8.9) | 815 (9.9) | 516 (10.8) | |
| 1 | 9100 (6.9) | 541 (6.6) | 341 (7.1) | <0.001 |
| 2-4 | 78,227 (59.2) | 4707 (57.3) | 2863 (60.0) | |
| 5+ | 33,063 (25.0) | 2145 (26.1) | 1053 (22.1) | |
| Parent broke bone after age 40 | 48,359 (36.4) | 3014 (36.6) | 1746 (36.3) | 0.95 |
| Corticosteroid Use >= 2 years | 687 (0.5) | 47 (0.6) | 22 (0.5) | 0.68 |
| Current smoker at baseline | 8851 (6.8) | 904 (11.1) | 290 (6.1) | <0.001 |
| Unopposed Estrogen Duration by Category | ||||
| None | 85,925 (64.7) | 5459 (66.2) | 3158 (65.7) | |
| < 5 Years | 17,579 (13.2) | 1051 (12.8) | 595 (12.4) | <0.001 |
| 5 - < 10 Years | 9293 (7.0) | 505 (6.1) | 308 (6.4) | |
| 10 - < 15 Years | 7308 (5.5) | 386 (4.7) | 252 (5.2) | |
| 15+ | 12,733 (9.6) | 841 (10.2) | 491 (10.2) | |
| Estrogen + Progest Duration by Category | ||||
| None | 97,356 (73.3) | 6219 (75.5) | 3147 (65.5) | |
| < 5 Years | 18,210 (13.7) | 919 (11.2) | 703 (14.6) | <0.001 |
| 5 - < 10 Years | 9587 (7.2) | 546 (6.6) | 493 (10.3) | |
| 10 - < 15 Years | 5246 (3.9) | 368 (4.5) | 296 (6.2) | |
| 15+ | 2439 (1.8) | 190 (2.3) | 165 (3.4) | |
| Baseline current HT user by duration | ||||
| Not current HT user | 78,129 (58.9) | 5275 (64.1) | 2495 (52.0) | <0.001 |
| Current HT <5yr | 15,926 (12.0) | 725 (8.8) | 598 (12.5) | |
| Current HT >= 5yr | 38,674 (29.1) | 2230 (27.1) | 1706 (35.5) | |
| Broke bone at/after 55 | 15,979 (12.0) | 1204 (14.6) | 564 (11.7) | <0.001 |
| Living w/partner | 83,031 (62.8) | 4857 (59.2) | 3015 (63.0) | <0.001 |
| Excellent/VG/Good v. Fair/Poor health | 120,477 (91.3) | 7393 (90.2) | 4407 (92.3) | <0.001 |
| History of arthritis | 61,300 (46.6) | 4087 (50.0) | 2248 (47.4) | <0.001 |
| Sedatives, psychotics or hypnotics drug use | 7899 (5.9) | 525 (6.4) | 278 (5.8) | 0.25 |
| CESD>=0.009 or anti-depressant use | 36,729 (28.4) | 2290 (28.5) | 1291 (27.5) | 0.36 |
| Years since menopause | ||||
| <10 | 41,053 (32.8) | 1869 (24.3) | 1515 (33.3) | <0.001 |
| 10 – 20 | 46,128 (36.9) | 2846 (36.9) | 1646 (36.2) | |
| ≥ 20 | 37,822 (30.3) | 2990 (38.8) | 1388 (30.5) | |
| 4th quartile of height (>161.9 cm) vs. others | 32,199 (24.4) | 2106 (25.7) | 1269 (26.5) | <0.001 |
| 4th quartile of weight (>70.5 kg) vs. others | 32,926 (24.9) | 2197 (26.8) | 1300 (27.1) | <0.001 |
| 4th quartile of FFQ caffeine intake (>188 mg) vs. others | 32,196 (25.0) | 2151 (26.9) | 1156 (24.7) | <0.001 |
| Family history of cancer | 83,196 (62.6) | 5474 (66.4) | 3207 (66.8) | <0.001 |
| Study | ||||
| E active | 4642 (3.5) | 293 (3.6) | 108 (2.2) | |
| E placebo | 4663 (3.5) | 293 (3.6) | 138 (2.9) | |
| E+P active | 7467 (5.6) | 525 (6.4) | 278 (5.8) | <0.001 |
| E+P placebo | 7193 (5.4) | 494 (6.0) | 204 (4.2) | |
| CT not randomized to HT | 34,929 (26.3) | 2220 (26.9) | 1391 (29.0) | |
| OS | 73,946 (55.7) | 4417 (53.6) | 2685 (55.9) | |
| Hysterectomy at randomization | ||||
| Yes | 53,894 (40.6) | 3052 (37.1) | 1705 (35.5) | <0.001 |
| No | 78,878 (59.4) | 5184 (62.9) | 3095 (64.5) | |
| Broke bone at/after 55 | 15,979 (12.0) | 1204 (14.6) | 564 (11.7) | <0.001 |
Baseline characteristics of the 1073 participants with non-invasive breast cancer are not displayed
From a chi-square test of association
Table 2.
Baseline characteristics in the WHI CT and OS – Continuous Variables
| No-cancer (N=132,840) |
Incident Other Cancer (N=8242) |
Incident Invasive BC (N=4804) |
|||||
|---|---|---|---|---|---|---|---|
| N | Mean (SD) | N | Mean (SD) | N | Mean (SD) | P-Valuea | |
| Age at screening, years | 132,840 | 62.9 (7.2) | 8242 | 65.2 (7.0) | 4804 | 63.6 (7.0) | <0.001 |
| Weight, kg | 132,408 | 73.5 (16.9) | 8211 | 74.3 (17.2) | 4791 | 74.5 (16.7) | <0.001 |
| Height, cm | 132,038 | 161.7 (6.7) | 8196 | 162.0 (6.6) | 4784 | 162.2 (6.6) | <0.001 |
| Gail 5 year risk | 132,840 | 1.7 (1.0) | 8242 | 1.9 (1.0) | 4804 | 2.0 (1.1) | <0.001 |
| Dietary Energy (kcal) | 128,808 | 1645.1 (646.2) | 8013 | 1647.5 (641.9) | 4674 | 1654.0 (622.1) | 0.63 |
| BL Total expend from phys act (kcal/week per kg) | 126,444 | 12.4 (13.8) | 7743 | 12.2 (13.7) | 4528 | 12.1 (12.9) | 0.05 |
| Dietary Alcohol (g) | 128,808 | 5.1 (10.4) | 8013 | 5.7 (11.4) | 4674 | 6.0 (11.3) | <0.001 |
From a one-way ANOVA
Age-adjusted annual incidence rates of fractures and falls were presented in Table 3. Fracture incidence rates increased after breast cancer or other cancer diagnosis. Before cancer diagnosis, annualized incident hip fracture rates were 0.13% and 0.15% for the BC and the OC group respectively and these rates were similar to the rate for the non-cancer group (0.15%). After a cancer diagnosis, the annual incidence rates of hip fractures increased to 0.25% in the BC group and 0.40% in the OC group. The age-adjusted incidence rate of falls (two or more falls) was also slightly higher after cancer diagnosis, but not before. We also limited the analysis on the annual percent of total fracture to 2 years as well as 1 year before the cancer diagnosis. The results showed similar patterns as in Table 3, suggesting fracture risk was relatively lower before breast cancer diagnosis, but was higher before other cancer diagnosis in comparison to women without a cancer diagnosis (data not shown).
Table 3. Events and age-adjusted annualized percentages of hip fracture and falls before and after incident cancer in the WHI CT and OS (n=146,959).
| Hip Fracture | Forearm Fracturea | Vertebral Fracturea | Total Fracturea | >= 2 Fallsb | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Exposure | Phase | Mean (SD) Follow-up time-yrs | N of Events | Annualized %c | N of Events | Annualized % c | N of Events | Annualized % c | N of Events | Annualized % c | N of Events | Annualized % c |
| No-Cancer (N = 132,840) | 7.8 (1.6) | 1,449 | 0.15% | 4,676 | 0.47% | 1,868 | 0.19% | 11,996 | 1.24% | 40,085 | 4.72% | |
| Incident Other Cancer (N = 8,242) | Before | 4.3 (2.4) | 61 | 0.15% | 157 | 0.42% | 127 | 0.34% | 513 | 1.45% | 1584 | 5.06% |
| After | 2.7 (2.5) | 96 | 0.40% | 116 | 0.53% | 109 | 0.48% | 396 | 1.89% | 982 | 5.76% | |
| Incident Breast Cancer (N = 4,804) | Before | 4.2 (2.4) | 25 | 0.13% | 84 | 0.43% | 31 | 0.16% | 216 | 1.12% | 854 | 4.89% |
| After | 3.9 (2.3) | 48 | 0.25% | 92 | 0.50% | 56 | 0.30% | 240 | 1.36% | 753 | 5.30% | |
OS data based on self-report
Based on self-report
Adjusted for age at baseline by 1-year groups
Multiple potential confounding factors, as suggested by the analyses in Tables 1 and 2, were examined with the fracture and fall outcomes. The following variables were adjusted in all the Cox proportional hazards models: age (linear), years since menopause, ethnicity, education, living with a partner, smoking, general health, parity, duration of E-alone at baseline, duration of E+P at baseline, sedatives or psychotics or hypnotic drug use, depressive symptoms or anti-depressant use, arthritis, corticosteroid use ≥ 2 years, weight (quartiles, linear), height (quartiles, linear), total energy expended from physical activity (linear, quartile), total calcium intake (logged linear, quartiles), total vitamin D intake (logged linear, quartiles), total energy intake (logged linear, quartiles), total alcohol intake (linear, quartiles), high caffeine, parental fracture after age 40, family history of cancer, Gail score (quartiles, linear) and stratified by age (5-year groups), HT trial assignment, DM trial assignment, CaD trial assignment, study participation, hysterectomy status and history of fracture after 55 years of age.
Results from Cox proportional hazards models (Table 4) showed 15% and 27% increased risk for falls after BC or OC diagnosis respectively. The difference in fall risk between the BC and OC was borderline statistically significant (P = 0.07). In a secondary data analysis, the risk of falls did not increase in BC who had estrogen receptor negative tumors.
Table 4. Risk of ≥ 2 falls after incident cancer in the WHI CT and OS (n=146,959).
| Number of Events | (Annualized Percent)a | Hazard Ratio (95% CI)b | P-valuec | P-valued | |
|---|---|---|---|---|---|
| No-Cancer | 42,708e | (4.73%) | 1.00 | ||
| Incident Other Cancer | 982 | (5.76%) | 1.27 (1.18, 1.36) | <0.001 | |
| Incident Breast Cancer | 753 | (5.30%) | 1.15 (1.06, 1.25) | <0.001 | 0.07 |
| ER+ | 568 | (5.33%) | 1.17 (1.06, 1.28) | 0.001 | 0.16 |
| ER- | 100 | (4.72%) | 1.02 (0.82, 1.27) | 0.86 | 0.07 |
| Other | 85 | (5.66%) | 1.21 (0.96, 1.53) | 0.11 | 0.73 |
Adjusted for age at baseline by 10-year groups.
From a proportional hazards model, where incident cancer status is fit as time-dependent variable, adjusted for age (linear),years since menopause, ethnicity, education, living with partner, smoking, general health, parity, duration of E-alone at baseline, duration of E+P at baseline, sedatives or psychotics or hypnotic drug use, depressive symptoms or anti-depressant use, arthritis, corticosteroid use >= 2yrs, weight (quartiles, linear), height (quartiles, linear), total energy expended from physical activity(linear, quartile), total calcium intake(logged linear, quartiles), total vitamin D intake (logged linear, quartiles), total energy intake (logged linear, quartiles), total alcohol intake (linear, quartiles), high caffeine, parent broke bone after age 40, family history of cancer, Gail score (quartiles, linear) and stratified by age (5 year groups), HT trial assignment, DM trial assignment, CaD trial assignment, study participation, hysterectomy status and history of fracture after 55 years of age.
From the described proportional hazards model testing whether the hazard ratio is different from unity.
From the described proportional hazards model testing whether the hazard ratio for breast cancer, or breast cancer type, is different from the hazard ratio for other cancer.
42768 falls(≥ 2) = 40085 incidents of falling (>=2) from the no- cancer group + 1584 incidents of falling (≥ 2) prior to incident other cancer + 854 incidents of falling (>=2) prior to incident invasive breast cancer + 245 incidents of falling (≥ 2) prior to incident non-invasive breast cancer.
In comparison to the no-cancer group, hip fracture risk significantly increased after BC or OC diagnosis (Table 5). After adjusting for covariates the hazards ratio (HR) of hip fracture was 1.55 (95% confidence interval [95%CI] = 1.13 to 2.11) in women with incident breast cancer and 2.09 (95%CI = 1.65 to 2.65) in women with incident other cancer. The increased hip fracture risk seemed more obvious in the ER- group. Although fall is a significant risk factor for hip fracture, HR=2.05 (95%CI = 1.82, 2.31 and p < 0.001), the increased number of falls did not explain the increased hip fracture risk in incident cancer cases. After adjusting for falls, as a time dependent covariate, the HR (95% CI) were 1.54 (1.13, 2.11) and 2.09 (1.65, 2.64) for BC and OC, respectively, which were very similar to the results in Table 5.
Table 5. Risk of hip fracture after incident cancer in the WHI CT and OS (n=146,959).
| Number of Events | (Annualized Percent)a | Hazard Ratio (95% CI)b | P-valuec | P-valued | |
|---|---|---|---|---|---|
| No-Cancer | 1,539e | (0.15%) | 1.00 | ||
| Incident Other Cancer | 96 | (0.40%) | 2.09 (1.65, 2.65) | <0.001 | |
| Incident Breast Cancer | 48 | (0.25%) | 1.55 (1.13, 2.11) | 0.006 | 0.12 |
| ER+ | 35 | (0.24%) | 1.44 (1.00, 2.06) | 0.05 | 0.08 |
| ER- | 8 | (0.34%) | 2.02 (0.90, 4.55) | 0.09 | 0.94 |
| Other | 5 | (0.25%) | 1.89 (0.78, 4.62) | 0.16 | 0.83 |
Adjusted for age at baseline by 10-year groups.
From a proportional hazards model using covariates and stratification variables described in Table 4.
From the described proportional hazards model testing whether the hazard ratio is different from unity.
From the described proportional hazards model testing whether the hazard ratio for breast cancer, or breast cancer type, is different from the hazard ratio for other cancer.
1,539 hip fractures = 1,449 fractures from the no- cancer group + 61 fractures prior to incident other cancer + 25 fractures prior to incident invasive breast cancer + 4 fractures prior to incident non-invasive breast cancer.
There were apparent increases in clinical vertebral fracture risk in breast cancer cases and in other cancer cases (Table 6), but the increased risk of fracture was only statistically significant (p < 0.001) in the OC group (HR = 1.86; 95%CI = 1.49 to 2.32). Falls increased risk of spinal fracture: HR =1.76 (95%CI = 1.58, 1.95). However, the results for clinical vertebral fractures were essentially unchanged: HR (95%) was 1.25 (0.93, 1.68) and 1.86 (1.49, 2.32) for BC and OC, respectively after controlling for falls.
Table 6. Risk of vertebral fracture after incident cancer in the WHI CT and OS (n=146,959).
| Number of Events | (Annualized Percent) a | Hazard Ratio (95% CI)b | P-valuec | P-valued | |
|---|---|---|---|---|---|
| No-Cancer | 2,033e | (0.19%) | 1.00 | ||
| Incident Other Cancer | 109 | (0.48%) | 1.86 (1.49, 2.32) | <0.001 | |
| Incident Breast Cancer | 56 | (0.30%) | 1.26 (0.94, 1.69) | 0.12 | 0.03 |
| ER+ | 44 | (0.31%) | 1.23 (0.88, 1.72) | 0.22 | 0.04 |
| ER- | 7 | (0.31%) | 1.62 (0.77, 3.42) | 0.21 | 0.73 |
| Other | 5 | (0.25%) | 1.04 (0.39, 2.79) | 0.94 | 0.26 |
Adjusted for age at baseline by 10-year groups.
From a proportional hazards model using covariates and stratification variables described in Table 4.
From the described proportional hazards model testing whether the hazard ratio is different from unity.
From the described proportional hazards model testing whether the hazard ratio for breast cancer, or breast cancer type, is different from the hazard ratio for other cancer.
2,033 vertebral fractures = 1,868 fractures from the no-cancer group + 127 fractures prior to incident other cancer + 31 fractures prior to incident invasive breast cancer + 7 fractures prior to incident non-invasive breast cancer.
Forearm fracture risk did not increase after breast cancer diagnosis (HR = 1.00) and increased slightly for other cancer diagnosis (HR = 1.07), but the amount of increase was small and did not reach statistical significance (Table 7). A significant increase in total fracture risk (HR = 1.33; 95%CI = 1.18 to 1.49) was found in women diagnosed with cancer other than invasive breast cancer (Table 8).
Table 7.
Risk of lower arm fracture after incident cancer in the WHI CT and OS (n=146,959)
| Number of Events | (Annualized Percent)a | Hazard Ratio (95% CI)b | P-valuec | P-valued | |
|---|---|---|---|---|---|
| No-Cancer | 4,938e | (0.46%) | |||
| Incident Other Cancer | 116 | (0.53%) | 1.07 (0.87, 1.31) | 0.54 | |
| Incident Breast Cancer | 92 | (0.50%) | 1.00 (0.79, 1.26) | 0.97 | 0.66 |
| ER+ | 72 | (0.52%) | 0.99 (0.76, 1.29) | 0.94 | 0.66 |
| ER- | 9 | (0.37%) | 0.89 (0.44, 1.78) | 0.74 | 0.62 |
| Other | 11 | (0.56%) | 1.16 (0.60, 2.24) | 0.66 | 0.81 |
Adjusted for age at baseline by 10-year groups.
From a proportional hazards model using covariates and stratification variables described in Table 4.
From the described proportional hazards model testing whether the hazard ratio is different from unity.
From the described proportional hazards model testing whether the hazard ratio for breast cancer, or breast cancer type, is different from the hazard ratio for other cancer.
4,938 lower arm fractures = 4,676 fractures from the no-cancer group + 157 fractures prior to incident other cancer + 84 fractures prior to incident invasive breast cancer + 21 fractures prior to incident non-invasive breast cancer.
Table 8. Risk of total fracture after incident cancer in the WHI CT and OS (n=146,959).
| Number of Events | (Annualized Percent)a | Hazard Ratio (95% CI)b | P-valuec | P-valued | |
|---|---|---|---|---|---|
| No-Cancer | 12,775e | (1.24%) | |||
| Incident Other Cancer | 396 | (1.89%) | 1.33 (1.18, 1.49) | <0.001 | |
| Incident Breast Cancer | 240 | (1.36%) | 1.02 (0.89, 1.18) | 0.74 | 0.005 |
| ER+ | 180 | (1.35%) | 0.99 (0.84, 1.17) | 0.87 | 0.004 |
| ER- | 33 | (1.44%) | 1.18 (0.81, 1.70) | 0.39 | 0.54 |
| Other | 27 | (1.44%) | 1.12 (0.73, 1.70) | 0.61 | 0.44 |
Adjusted for age at baseline by 10-year groups.
From a proportional hazards model using covariates and stratification variables described in Table 4.
From the described proportional hazards model testing whether the hazard ratio is different from unity.
From the described proportional hazards model testing whether the hazard ratio for breast cancer, or breast cancer type, is different from the hazard ratio for other cancer.
12,775 total fractures = 11,996 fractures from the no- cancer group + 513 fractures prior to incident other cancer + 216 fractures prior to incident invasive breast cancer + 50 fractures prior to incident non-invasive breast cancer.
In an exploratory analysis about the potential explanations for the increased risk of fractures, age adjusted frequencies of selected medication use for the WHI CT was calculated and presented in Table 9. The percentage of tamoxifen use increased to 40.3% and the percentage of aromatase inhibitors use increased to 7.3% after breast cancer diagnosis. The percent of women on tamoxifen and aromatase inhibitors for other cancer was small. The use of bisphosphonates was more common (13.2%) in the no-cancer group than the BC or OC group either before (BC = 2.8%, OC = 4.5%) or after the cancer diagnosis (BC = 10.7%, OC = 6.5%). Less than 0.5% women who were on antiestrogen used Fareston (toremifene citrate) instead of tamoxifen (data not shown).
Table 9. Medication use during baseline or follow-up WHI CT (n=64,838).
| Number and Percenta of Participant Using Each Agent | |||||
|---|---|---|---|---|---|
| Non cancer n=58,894 | Breast Cancer Pt n=2,119 | Other Cancers n=3,825 | |||
| Before | After diagnosis | Before | After Diagnosis | ||
| Tamoxifen | 50 (0.1%) | 1 (<0.1%) | 855 (40.3%) | 4 (0.1%) | 10 (0.3%) |
| Aromatase Inhibitors | 1 (<0.1%) | 0 | 155 (7.3%) | 0 | 4 (0.1%) |
| Raloxifene | 1834 (3.1%) | 30 (1.4%) | 34 (1.6%) | 40 (1.0%) | 58 (1.5%) |
| Bisphosphonates | 7560 (13.2%) | 58 (2.8%) | 227 (10.7%) | 175 (4.5%) | 259 (6.5%) |
Adjusted by age at baseline.
Though not statistically significant, tamoxifen use was associated with a lower fracture risk. The HR of total fracture was 0.95 (95%CI = 0.72, 1.26) for women without tamoxifen therapy and 0.78 (95%CI = 0.51, 1.18) for women with tamoxifen therapy and ER+ BC. Aromatase inhibitors significantly increased risk of fractures in women with ER+ breast tumors: the HR for the total fractures was 2.76 (95% CI = 1.28, 5.92) for aromatase inhibitors users and 0.83 (95%CI = 0.65, 1.06) for the non-users. It should be noted that the number of fractures was small in these groups; of the 102 total fractures cases in the CT who had an ER+ invasive breast cancer, 51 were tamoxifen users after incident breast cancer while 51 were non-users and 12 were aromatase inhibitor users after incident invasive breast cancer while 90 were not. Fracture risk could not be assessed among the few participants taking raloxifene.
Discussion
This study provides new evidence indicating that while incident cancer cases did have a comparable fracture risk as their non-cancer peers of the same age before they were diagnosed with cancer, both fracture and fall risks are significantly elevated after a diagnosis of either breast cancer or any other cancer in postmenopausal women. The increased fracture risk is even more notable after a diagnosis of cancer other than invasive breast cancer.
Several factors have been considered as the potential causes for a higher fracture risk in postmenopausal breast cancer survivors. Chemotherapy, aromatase inhibitor use and the tumor itself may all direct and indirectly interfere with normal bone metabolism [17]. It is well-known that bone metastases increase the risk for pathological fractures. Contraindication of postmenopausal hormone therapy after breast cancer therapy is common, especially in hormonally sensitive breast cancer survivors, and it may increase bone turnover rate and eventually lead to a higher fracture rate. However, an increased risk of fracture in breast cancer survivors has been identified in some [8, 18, 19], but not all reports [20]. Previously, we reported that WHI OS participants with a breast cancer history at baseline had a significantly increased risk of clinical fractures (except for hip fractures) in comparison to women who did not have any cancer history [8]. Unfortunately, more than 77% of cancer survivors who had osteoporosis were undiagnosed at the time of WHI enrollment. The results reported here have not only confirmed our findings on a higher fracture risk among breast cancer cases, but also provided the first piece of prospective evidence showing that the higher fracture risk was only evidenced post-cancer diagnosis, not before.
Interestingly, there are some discrepancies on skeletal sites for the increased fracture risk between this current study using incident breast cancer cases and the previous report with the prevalent cases [8]. In the current study, the most remarkably increased fracture risk was found for the hip site, while total fracture risk was not increased after BC diagnosis. In contrast, fracture risk significantly increased for all fractures but not for the hip fracture risk specifically in the previous report with prevalent breast cancer cases. These may be due to the fact that there were fewer hip fractures, but more other fractures among the prevalent cases in comparison to the incident cases. Whether cancer treatment, time since BC diagnosis, age at BC diagnosis and the time period for fracture ascertainment, contribute to these discrepancies are interesting research questions for future investigation.
The association between hormonal receptor status of breast cancer and fracture or fall risk may well be related in part to the commonly used treatment for such cancer types with tamoxifen. As a partial estrogen agonist, tamoxifen decreases bone mineral density in premenopausal women, but has the opposite effect in postmenopausal women [21]. In the National Surgical Adjuvant Breast and Bowel Project P-1 Study, tamoxifen use resulted in a reduction in hip, radius and spine fractures compared to placebo in postmenopausal women. In contrast, the use of aromatase inhibitors in this study seemed to increase fracture risk among women with ER+ BC [22]. The elevated fracture risk with aromatase inhibitors use in our study sample is in agreement with the previous findings.
To our knowledge, this is the first paper demonstrating that fracture risk is increased in postmenopausal women after a diagnosis of cancers other than breast cancer. Intuitively, some of the risk factors for increased fracture associated with invasive breast cancer, such as use of chemotherapy and bone metastasis may also contribute to the higher fracture risk in other cancers. Further study is needed to identify high-risk groups of women and to develop strategies to prevent and manage the additional adverse health outcomes associated with cancer.
Falls are a major risk factor for fracture and the etiology of falls is multifactorial, including vitamin D insufficiency; an age-related decrease in muscle size and strength; neurological disorders of vision, balance and coordination; the use of sedatives and other medications which impair cognition [23]; and hospitalization. The cause of increased falls seen in cancer patients in this study is not known, but might include all of the above. Falls are very common in older people. Approximately 10–15% of falls in the elderly result in fracture [24] and more than 90% of fractures are associated with a fall [10]. We found a strong relationship between fall and fractures in this study. However, in spite of the increased number of falls after a cancer diagnosis, this higher rate of falling could not explain the higher fracture risk in cancer cases.
Recent findings have suggested that antiresorptive agents may effectively prevent bone loss in breast cancer survivors [25]. It is striking that in our study the percentage of women on bisphosphonates was even lower in breast cancer cases than in women without a cancer diagnosis in spite of the increased fracture risk. Based on our findings, breast cancer patients who have ER+ breast cancer tumors and are on aromatase inhibitors were at higher fracture risk. It is recognized that these women should be evaluated and treated as necessary to prevent fractures. However, our study indicates that fracture prevention needs to be improved in this high-risk cancer survivor population.
To date, this is the largest and most comprehensive prospective study addressing fracture risk, a significant health problem in millions of cancer survivors. The large multiethnic cohort, confirmed cancer diagnosis, adjudicated hip fractures in all women, fracture information before and after cancer diagnosis, and detailed information on tumor receptors and medication use in BC are some major strengths of this study. Nevertheless, the current study has several limitations. Not all fractures used in the analyses were adjudicated. However, we have repeated all the analyses using only the adjudicated fractures in the CT participants and results were similar, suggesting errors in self-reported fractures did not significantly alter the findings of this study. We could not distinguish pathological fractures from other fractures in this analysis. We have used self-reported falls in this study and the validity of this fall assessment tool is uncertain. Assuming the measurement errors in self-reported falls are random, then the relationship between fractures and fall is likely to be underestimated in this study. These limitations have reflected the challenges in prospective investigations on fracture and fall risk in the cancer survivor populations.
In conclusion, a significant increase in fracture risk at selected skeletal sites is seen in postmenopausal women following a diagnosis of invasive breast cancer or other cancer. Despite the higher rate of falls in women with diagnosed cancer, falls are less likely to be the major cause of the increased risk of fracture associated with a cancer diagnosis. Further study is needed to inform development of strategies to reduce the burden of fracture to the large number of cancer survivors in the United States and worldwide.
Appendix
Program Office
(National Heart, Lung, and Blood Institute, Bethesda, Maryland) Elizabeth Nabel, Jacques Rossouw, Shari Ludlam, Linda Pottern, Joan McGowan, Leslie Ford, and Nancy Geller.
Clinical Coordinating Center
(Fred Hutchinson Cancer Research Center, Seattle, WA)
Ross Prentice, Garnet Anderson, Andrea LaCroix, Charles L. Kooperberg, Ruth E. Patterson, Anne McTiernan; (Wake Forest University School of Medicine, Winston-Salem, NC) Sally Shumaker; (Medical Research Labs, Highland Heights, KY) Evan Stein; (University of California at San Francisco, San Francisco, CA) Steven Cummings.
Clinical Centers
(Albert Einstein College of Medicine, Bronx, NY) Sylvia Wassertheil-Smoller; (Baylor College of Medicine, Houston, TX) Jennifer Hays; (Brigham and Women's Hospital, Harvard Medical School, Boston, MA) JoAnn Manson; (Brown University, Providence, RI) Annlouise R. Assaf; (Emory University, Atlanta, GA) Lawrence Phillips; (Fred Hutchinson Cancer Research Center, Seattle, WA) Shirley Beresford; (George Washington University Medical Center, Washington, DC) Judith Hsia; (Los Angeles Biomedical Research Institute at Harbor- UCLA Medical Center, Torrance, CA) Rowan Chlebowski; (Kaiser Permanente Center for Health Research, Portland, OR) Evelyn Whitlock; (Kaiser Permanente Division of Research, Oakland, CA) Bette Caan; (Medical College of Wisconsin, Milwaukee, WI) Jane Morley Kotchen; (MedStar Research Institute/Howard University, Washington, DC) Barbara V. Howard; (Northwestern University, Chicago/Evanston, IL) Linda Van Horn; (Rush Medical Center, Chicago, IL) Henry Black; (Stanford Prevention Research Center, Stanford, CA) Marcia L. Stefanick; (State University of New York at Stony Brook, Stony Brook, NY) Dorothy Lane; (The Ohio State University, Columbus, OH) Rebecca Jackson; (University of Alabama at Birmingham, Birmingham, AL) Cora E. Lewis; (University of Arizona, Tucson/Phoenix, AZ) Tamsen Bassford; (University at Buffalo, Buffalo, NY) Jean Wactawski-Wende; (University of California at Davis, Sacramento, CA) John Robbins; (University of California at Irvine, CA) F. Allan Hubbell; (University of California at Los Angeles, Los Angeles, CA) Howard Judd; (University of California at San Diego, LaJolla/Chula Vista, CA) Robert D. Langer; (University of Cincinnati, Cincinnati, OH) Margery Gass; (University of Florida, Gainesville/Jacksonville, FL) Marian Limacher; (University of Hawaii, Honolulu, HI) David Curb; (University of Iowa, Iowa City/Davenport, IA) Robert Wallace; (University of Massachusetts/Fallon Clinic, Worcester, MA) Judith Ockene; (University of Medicine and Dentistry of New Jersey, Newark, NJ) Norman Lasser; (University of Miami, Miami, FL) Mary Jo O'Sullivan; (University of Minnesota, Minneapolis, MN) Karen Margolis; (University of Nevada, Reno, NV) Robert Brunner; (University of North Carolina, Chapel Hill, NC) Gerardo Heiss; (University of Pittsburgh, Pittsburgh, PA) Lewis Kuller; (University of Tennessee, Memphis, TN) Karen C. Johnson; (University of Texas Health Science Center, San Antonio, TX) Robert Brzyski; (University of Wisconsin, Madison, WI) Gloria E. Sarto; (Wake Forest University School of Medicine, Winston-Salem, NC) Denise Bonds; (Wayne State University School of Medicine/Hutzel Hospital, Detroit, MI) Susan Hendrix.
Acknowledgments
The WHI program is funded by the National Heart, Lung and Blood Institute, U.S. Department of Health and Human Services. No additional funding was received for this project. The authors had full access to all of the data in the study, take responsibility for the integrity of the data, and the accuracy of the data analysis.
References
- 1.Cummings SR, Melton LJ. Epidemiology and outcomes of osteoporotic fractures. Lancet. 2002;359:1761–1767. doi: 10.1016/S0140-6736(02)08657-9. [DOI] [PubMed] [Google Scholar]
- 2.Cauley JA, Lucas FL, Kuller LH, et al. Bone mineral density and risk of breast cancer in older women: the study of osteoporotic fractures. Study of Osteoporotic Fractures Research Group. Jama. 1996;276:1404–1408. see comment. [PubMed] [Google Scholar]
- 3.Newcomb PA, Trentham-Dietz A, Egan KM, et al. Fracture history and risk of breast and endometrial cancer. Am J Epidemiol. 2001;153:1071–1078. doi: 10.1093/aje/153.11.1071. [DOI] [PubMed] [Google Scholar]
- 4.Chen Z, Maricic M, Pettinger M, et al. Osteoporosis and rate of bone loss among postmenopausal survivors of breast cancer. Cancer. 2005;104:1520–1530. doi: 10.1002/cncr.21335. [DOI] [PubMed] [Google Scholar]
- 5.Friedlaender GE, Tross RB, Doganis AC, et al. Effects of chemotherapeutic agents on bone. I. Short-term methotrexate and doxorubicin (adriamycin) treatment in a rat model. J Bone Joint Surg Am. 1984;66:602–607. [PubMed] [Google Scholar]
- 6.May KP, West SG, McDermott MT, et al. The effect of low-dose methotrexate on bone metabolism and histomorphometry in rats. Arthritis Rheum. 1994;37:201–206. doi: 10.1002/art.1780370208. [DOI] [PubMed] [Google Scholar]
- 7.Twiss JJ, Waltman N, Ott CD, et al. Bone mineral density in postmenopausal breast cancer survivors. J Am Acad Nurse Pract. 2001;13:276–284. doi: 10.1111/j.1745-7599.2001.tb00035.x. [DOI] [PubMed] [Google Scholar]
- 8.Chen Z, Maricic M, Bassford TL, et al. Fracture risk among breast cancer survivors: results from the Women's Health Initiative Observational Study. Arch Intern Med. 2005;165:552–558. doi: 10.1001/archinte.165.5.552. [DOI] [PubMed] [Google Scholar]
- 9.Michaud LB, Goodin S. Cancer-treatment-induced bone loss, part 1. Am J Health Syst Pharm. 2006;63:419–430. doi: 10.2146/ajhp050045.p1. [DOI] [PubMed] [Google Scholar]
- 10.Rivara FP, Grossman DC, Cummings P. Injury prevention. Second of two parts. N Engl J Med. 1997;337:613–618. doi: 10.1056/NEJM199708283370907. see comment. [DOI] [PubMed] [Google Scholar]
- 11.Design of the Women's Health Initiative clinical trial and observational study. The Women's Health Initiative Study Group. Control Clin Trials. 1998;19:61–109. doi: 10.1016/s0197-2456(97)00078-0. [DOI] [PubMed] [Google Scholar]
- 12.Langer RD, White E, Lewis CE, et al. The Women's Health Initiative Observational Study: baseline characteristics of participants and reliability of baseline measures. Ann Epidemiol. 2003;13:S107–121. doi: 10.1016/s1047-2797(03)00047-4. [DOI] [PubMed] [Google Scholar]
- 13.Chen Z, Kooperberg C, Pettinger MB, et al. Validity of self-report for fractures among a multiethnic cohort of postmenopausal women: results from the Women's Health Initiative observational study and clinical trials. Menopause. 2004;11:264–274. doi: 10.1097/01.gme.0000094210.15096.fd. [DOI] [PubMed] [Google Scholar]
- 14.Hosmer DW, Lemeshow S. Applied survival analysis : regression modeling of time to event data. Wiley; New York: 1999. [Google Scholar]
- 15.Cauley JA, Wu L, Wampler NS, et al. Clinical Risk Factors for Fractures in Multiethnic Women: The Women's Health Initiative. Journal of Bone and Mineral Research. doi: 10.1359/jbmr.070713. [DOI] [PubMed] [Google Scholar]
- 16.Prentice RL. Surrogate endpoints in clinical trials: definition and operational criteria. Stat Med. 1989;8:431–440. doi: 10.1002/sim.4780080407. [DOI] [PubMed] [Google Scholar]
- 17.Delmas PD, Fontana A. Bone loss induced by cancer treatment and its management. Eur J Cancer. 1998;34:260–262. doi: 10.1016/s0959-8049(97)10135-6. [DOI] [PubMed] [Google Scholar]
- 18.Adami HO, Zack M, Kressner U, et al. Hip fractures in women with breast cancer. Am J Epidemiol. 1990;132:877–883. doi: 10.1093/oxfordjournals.aje.a115730. [DOI] [PubMed] [Google Scholar]
- 19.Kanis JA, McCloskey EV, Powles T, et al. A high incidence of vertebral fracture in women with breast cancer. Br J Cancer. 1999;79:1179–1181. doi: 10.1038/sj.bjc.6690188. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Lamont EB, Lauderdale DS. Low risk of hip fracture among elderly breast cancer survivors. Ann Epidemiol. 2003;13:698–703. doi: 10.1016/s1047-2797(03)00056-5. [DOI] [PubMed] [Google Scholar]
- 21.Chang J, Powles TJ, Ashley SE, et al. The effect of tamoxifen and hormone replacement therapy on serum cholesterol, bone mineral density and coagulation factors in healthy postmenopausal women participating in a randomised, controlled tamoxifen prevention study. Ann Oncol. 1996;7:671–675. doi: 10.1093/oxfordjournals.annonc.a010715. see comment. [DOI] [PubMed] [Google Scholar]
- 22.Fisher B, Costantino JP, Wickerham DL, et al. Tamoxifen for prevention of breast cancer: report of the National Surgical Adjuvant Breast and Bowel Project P-1 Study. J Natl Cancer Inst. 1998;90:1371–1388. doi: 10.1093/jnci/90.18.1371. see comment. [DOI] [PubMed] [Google Scholar]
- 23.Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med. 1988;319:1701–1707. doi: 10.1056/NEJM198812293192604. [DOI] [PubMed] [Google Scholar]
- 24.Nevitt MC, Cummings SR, Hudes ES. Risk factors for injurious falls: a prospective study. J Gerontol. 1991;46:M164–170. doi: 10.1093/geronj/46.5.m164. [DOI] [PubMed] [Google Scholar]
- 25.Greenspan SL, Bhattacharya RK, Sereika SM, et al. Prevention of bone loss in survivors of breast cancer: A randomized, double-blind, placebo-controlled clinical trial. J Clin Endocrinol Metab. 2007;92:131–136. doi: 10.1210/jc.2006-1272. [DOI] [PubMed] [Google Scholar]