Journal of Cancer Survivorship

, Volume 8, Issue 3, pp 384–393

Determinants of newly diagnosed comorbidities among breast cancer survivors

Authors

    • University Cancer Center HamburgUniversity Medical Center Hamburg-Eppendorf
    • Clinical Cancer Registry/Cancer Epidemiology, University Cancer Center Hamburg (UCCH)University Medical Center Hamburg-Eppendorf
  • Daniela Gornyk
    • University Cancer Center HamburgUniversity Medical Center Hamburg-Eppendorf
  • Judith Heinz
    • University Cancer Center HamburgUniversity Medical Center Hamburg-Eppendorf
  • Alina Vrieling
    • Department of Cancer EpidemiologyGerman Cancer Research Center (DKFZ)
    • Department for Health EvidenceRadboud University Medical Centre
  • Petra Seibold
    • Department of Cancer EpidemiologyGerman Cancer Research Center (DKFZ)
  • Jenny Chang-Claude
    • Department of Cancer EpidemiologyGerman Cancer Research Center (DKFZ)
  • Dieter Flesch-Janys
    • University Cancer Center HamburgUniversity Medical Center Hamburg-Eppendorf
Article

DOI: 10.1007/s11764-013-0338-y

Cite this article as:
Obi, N., Gornyk, D., Heinz, J. et al. J Cancer Surviv (2014) 8: 384. doi:10.1007/s11764-013-0338-y

Abstract

Purpose

Comorbid conditions have become increasingly relevant for breast cancer care given the large numbers of long-term survivors. Our aim was to identify potential determinants associated with the development of comorbidities after breast cancer.

Methods

Self-reported comorbidities and lifestyle were assessed at recruitment and after a median follow up of 69.4 months from diagnosis in a population-based cohort of breast cancer cases aged 50 to 74 years at diagnosis (MARIEplus study). Tumor and therapy data were extracted from medical records. Determinants potentially associated with incident diagnoses of hypertension, cardiovascular diseases (CVD), and osteoporosis were assessed using multivariable Cox proportional hazard regression models.

Results

Follow-up interview was completed by 2,542 women (76.4 % of eligible patients). A diagnosis of hypertension was significantly associated with age, higher education (hazard ratio (HR) 0.54, CI 0.37–0.79), baseline body mass index (BMI; ≥30 kg/m2; HR, 1.90; CI, 1.24–2.90), and trastuzumab medication (HR, 2.16; CI, 1.09–4.33). An increased risk for CVD was associated with age, BMI, and intake of aromatase inhibitors (AI; HR, 1.42; CI, 1.09–1.84). Risk of osteoporosis was also positively associated with AI treatment (HR, 2.15; CI, 1.64–2.82) but inversely associated with a higher BMI (≥30 kg/m2; HR, 0.50; CI, 0.31–0.79).

Conclusion

In breast cancer survivors, treatment with AI constituted a risk factor for incident CVD and osteoporosis. Besides known risk factors, patients who were treated with trastuzumab may have an increased risk for hypertension.

Implications for cancer survivors

Reducing overweight and regular sport/cycling activities may help to prevent CVD after breast cancer. Patients should be monitored for risk factors and advised on possible cardiac side effects of AI and trastuzumab.

Keywords

Breast cancer survivorsComorbidityHypertensionOsteoporosisAromatase inhibitorTrastuzumab

Abbreviation

CI

95 % CI

Introduction

Breast cancer survivors are a steadily growing population because mortality has declined since the 1990s. The 5-year relative survival rate has improved to approximately 90 % in the USA [1] and 85 % in the UK [2] due to improved therapy options and early cancer detection programs. At the same time, aging of cancer survivors is frequently accompanied by comorbid conditions [3, 4] that might lead to impairment of function, quality of life, and prognosis [5]. In fact, differences in survival rates after breast cancer have been attributed to prevalent comorbid conditions independently from tumor stage at diagnosis [6, 7]. Comorbidities are generally defined as concurrent, etiologically independent chronic health conditions, unrelated to the disease under study [8]. To study incidence and determinants of comorbid conditions in breast cancer survivors, it is important to distinguish comorbidities present at diagnosis from those occurring incidentally after diagnosis. The latter also includes adverse effects potentially related to long-term cancer treatment with tamoxifen [9], aromatase inhibitors (AI) [10], and trastuzumab [11]. Observational studies that have evaluated determinants for newly diagnosed comorbidities after breast cancer are scarce. One study focused on adjuvant therapy and its impact on overall comorbid burden but did not consider lifestyle factors in the analysis [12]. Another study was based on an administrative database of primary care records with limited information on lifestyle [13]. Understanding the role of lifestyle factors is important for the management of adverse treatment effects and long-term health of cancer survivors. Therefore, this prospective study of a breast cancer cohort aims to identify risk factors for developing three of the most common comorbidities during a median follow-up of 5.8 years: hypertension, cardiovascular diseases (CVD), and osteoporosis.

Methods

A prospective case–cohort study of breast cancer patients recruited in the population-based case–control study MARIE (acronym for Mamma carcinoma Risk factor Investigation; MARIEplus) was carried out in the city and state of Hamburg and the Rhine-Neckar-Karlsruhe region in Germany, focusing on risk factors for newly diagnosed comorbidities, lifestyle changes, course of therapy, and prognosis. Methods of the MARIE study have been described elsewhere [14]. In brief, breast cancer patients aged between 50 and 74 years with primary invasive breast cancer or carcinoma in situ diagnosed between 2001 and 2005 in the study regions Hamburg and Rhine-Neckar-Karlsruhe were eligible for inclusion in the MARIE study. They were invited to participate in the study while still in the hospital, after discharge or notification from the Hamburg cancer registry. Median time from diagnosis to recruitment was 4 months. Baseline data, e.g., lifestyle, body mass index (BMI), and comorbidities, were assessed in standardized personal interviews. In the follow-up study, MARIEplus, vital status and causes of death were verified through 12/31/2009 for the whole cohort. A standardized computer-assisted telephone interview was carried out in survivors from May to September 2009, to obtain information on lifestyle factors, therapy (e.g., hormonal therapy), recurrences, second tumors, and newly diagnosed comorbidities. Additionally, tumor characteristics and treatment of the primary tumor, information on recurrence, and second tumors were extracted from medical records.

Considering comorbidities, patients were asked at baseline and at follow-up whether they had had a physician-confirmed diagnosis specified in a list of diseases, including hypertension, myocardial infarction, angina pectoris, stroke, arterial obstructive disease, diabetes, osteoporosis, arthritis/rheumatism, inflammatory bowel disease, chronic bladder disease, chronic liver disease, thyroid disease, and chronic pulmonary disease. If answered with “yes”, the patients were asked in which year the specific condition occurred first. Patients were assumed to have had the respective disease if they answered the question at baseline, follow-up or both assessments with “yes” and reported a year of diagnosis. In case of differences in date of first diagnosis between baseline interview and follow-up survey, the date given at baseline was used, assuming a lower probability of a recall bias at this time point. Using the reported year of diagnosis of the comorbidity and the date of diagnosis of breast cancer, it was determined whether the reported comorbidity occurred before or after the breast cancer diagnosis. If the comorbid condition was reported at baseline to have occurred in the same year as the breast cancer, we considered it “before or at breast cancer diagnosis”, and if first mentioned in the follow-up interview, we considered its first occurrence “after the breast cancer”. However, if a patient did not report to have had the disease at baseline but reported in the follow-up interview that the year of first diagnosis was prior to the diagnosis of breast cancer, data were set to missing and excluded from the analysis. Patients who already had the specified disease before the breast cancer diagnosis or did not remember the date of diagnosis were excluded from the respective analyses (Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs11764-013-0338-y/MediaObjects/11764_2013_338_Fig1_HTML.gif
Fig. 1

Flow diagram of the MARIEplus breast cancer cohort and subgroups of analyses

Out of the 13 reported comorbid conditions, the three most common diseases, hypertension, CVD (i.e., myocardial infarction, stroke, angina pectoris, and arterial obstructive disease), and osteoporosis, newly occurring after breast cancer diagnosis were selected as outcome variables, since they yielded sufficient numbers for analysis (Fig. 1). Although arthritis, inflammatory diseases of the spine, and rheumatism were very frequently reported conditions, they were not considered as outcome because patients were likely to have included also non-inflammatory diseases (e.g., lower back pain) in their response.

The study was approved by the ethics committees of the University of Heidelberg and the Ethics Committee of the Medical Council of Hamburg and conducted in accordance with the Declaration of Helsinki. All study participants gave written informed consent.

Statistical analyses

To assess the associations between determinants and disease outcomes of interest, hazard ratios (HR) with 95 % confidence intervals (95 % CI) were estimated using Cox proportional hazards models. Each model included known and suspected risk factors and potential confounding variables mainly derived from the baseline interview: age (continuous), study region, postmenopausal at breast cancer diagnosis (yes, no), marital status (married, single, separated or divorced, widowed), education (low, middle, high, calculated by the two variables school education and apprenticeship or academic study), BMI at recruitment (categorized according to the World’s Health Organization’s recommendations, <22.5, <25, <30, 30+ kg/m2), age at menarche (<12, 12–<15, 15+ years), menopausal hormone therapy (MHT) at breast cancer diagnosis (never, former, current), parity (none, 1, 2, 3+), summarized measure of recreational physical activity since age of 50 years, i.e., sports and cycling (metabolic equivalent × hours/week [MET], 0, <12, <24, <42, ≥42) [15], smoking habits at breast cancer diagnosis (never, former, current), and alcohol consumption (0, >0-<19, 19+ g/day). Further included were tumor stage (Ca in situ, 1, 2, 3/4 according to UICC), a combined variable of her2/neu receptor status and trastuzumab medication (negative her2/neu /no trastuzumab, positive/with trastuzumab, positive/no trastuzumab, unknown) and other types of therapy, including chemotherapy, radiotherapy, intake of tamoxifen, AI, trastuzumab, and/or bisphosphonates (yes vs. no/unknown, respectively; Table 1). Additionally, baseline information on diabetes mellitus (yes, no) was included in the models as a potential risk factor for hypertension and CVD, respectively, and prevalent hypertension as a risk factor for CVD. Follow-up time was calculated as the time between date of diagnosis and date of comorbidity event or censoring at date of telephone interview. To assess potential participation bias baseline characteristics of follow-up participants and nonparticipants as well as deceased women were compared (Table 1). P values of less than 0.05 were considered significant. All analyses were conducted with PASW 18 (SPSS).
Table 1

Baseline characteristics of the breast cancer cohort by participation status in the MARIEplus study (follow-up interview participants [present analysis group], nonresponder to follow-up and deceased)

 

Follow-up interview

Nonresponder

Deceased

N

%

N

%

N

%

 

2,542

100

785

100

485

100

Age (mean, SD)

62.1

5.9

62.6

6.2

63.4

6.3

Region

 Hamburg

1,345

52.9

489

62.3

240

49.5

 RNK

1,197

47.1

296

37.7

245

50.5

Postmenopausal

 No

240

9.4

73

9.3

36

7.4

 Yes

2,302

90.6

712

90.7

449

92.6

Marital status

 Married

1,709

67.3

479

61.0

269

55.6

 Widowed

345

13.6

118

15.0

107

22.1

 Separated/divorced

328

12.9

134

17.1

77

15.9

 Single

159

6.3

54

6.9

31

6.4

 Unknown

1

 

 

 

Education

 Low

1,405

55.3

459

58.5

315

64.9

 Medium

733

28.8

210

26.8

117

24.1

 High

404

15.9

115

14.6

52

10.7

 Unknown

 

1

0.1

1

0.2

BMI (kg/m2)

 

 <22.5

548

21.6

176

22.4

104

21.4

 <25

676

26.6

186

23.7

93

19.2

 <30

922

36.3

280

35.7

182

37.5

 30+

394

15.5

142

18.1

105

21.6

 Unknown

2

0.1

1

0.1

1

0.2

Age at menarche (years)

 <12

213

8.4

62

7.9

49

10.1

 12–<15

1,671

65.7

488

62.2

306

63.2

 15+

654

25.7

234

29.8

129

26.7

 Unknown

4

0.2

1

0.1

1

0.2

Parity (≥28 weeks)

 None

456

17.9

134

17.1

91

18.8

 1

731

28.8

226

28.8

139

28.7

 2

950

37.4

291

37.1

163

33.6

 3+

405

15.9

134

17.1

92

19.0

Menopausal HTa

 Never

753

29.6

278

35.4

253

52.2

 Past

514

20.2

137

17.5

91

18.8

 Current

1,261

49.6

363

46.2

135

27.8

 Unknown

14

0.6

7

0.9

6

1.2

Physical activity b (MET × h/week)

 0 (no sports/cycling)

442

17.4

192

24.5

143

29.5

 >0–<12

642

25.3

183

23.3

100

20.6

 <24

526

20.7

142

18.1

85

17.5

 <42

478

18.8

127

16.2

81

16.7

 ≥42

434

17.1

131

16.7

73

15.1

 Unknown

20

0.8

10

1.3

3

0.6

Smoking

 Never smoker

1,344

52.9

409

52.1

252

52.0

 Former smoker

747

29.4

192

24.5

114

23.5

 Current smoker

451

17.7

184

23.4

119

24.5

Alcohol consumption (g/day)

 0

330

13.0

140

17.8

72

14.8

 >0–<19

1,883

74.1

535

68.2

340

70.1

 19+

325

12.8

109

13.9

72

14.8

 Unknown

4

0.2

1

0.1

1

0.2

Tumor stage

 In situ

162

6.4

63

8.0

7

1.4

 1

1,161

45.7

315

40.1

90

18.6

 2

916

36.0

305

38.9

169

34.8

 3/4

223

8.8

77

9.8

171

35.3

 Unknown

80

3.1

25

3.2

48

9.9

Radiotherapy

 No

531

20.9

202

25.7

155

32.0

 Yes

2,000

78.7

563

71.7

315

64.9

 Unknown

11

0.4

20

2.5

15

3.1

Chemotherapy

 No

1,367

53.8

390

49.7

168

34.6

 Yes

1,163

45.8

361

46.0

304

62.7

 Unknown

12

0.5

34

4.3

13

2.7

Tamoxifen

 No

771

30.3

184

13.4

164

33.6

 Yes

1,751

68.9

502

63.9

237

48.9

 Unknown

20

0.8

99

12.6

84

17.3

Aromatase inhibitors

 No

1,332

52.4

521

56.3

293

49.3

 Yes

1,174

46.2

165

21.0

162

33.4

 Unknown

36

1.4

99

12.6

84

17.3

Her2neu/trastuzumab

 Negative/no trastuzumab (incl. Ca in situ)

1,905

74.9

574

73.1

336

69.3

 Positive/trastuzumab

46

1.8

15

1.9

45

9.3

 Positive/no trastuzumab

366

14.4

133

16.9

72

14.8

 Unknown

225

8.9

63

8.0

32

6.6

Bisphosphonates

 No/unknown

2,476

97.4

749

95.4

354

73.0

 Yes

66

2.6

36

4.6

131

27.0

Diabetes

 No

2,365

93.0

716

91.2

422

87.0

 Yes

167

6.6

67

8.5

61

12.6

 Unknown

10

0.4

2

0.3

2

0.4

Hypertension

 No

1,561

61.4

466

59.4

255

52.6

 Yes

951

37.4

315

40.1

229

47.2

 Unknown

30

1.2

4

0.5

1

0.2

aMenopausal HT, menopausal hormone therapy (estrogen and/or progesterone)

bPhysical activity refers to recreational PA since age 50 years, including sports and cycling in MET, metabolic equivalents × hours/week

Results

Of all 3,813 breast cancer patients enrolled in the MARIE study, 3,327 were eligible at follow up. Of the latter, 2,542 (76.4 %) completed the follow-up interview and were included in the analysis (Fig. 1). Median time from diagnosis to follow-up interview was 69.3 months. Of the nonparticipating women, 785 did not respond or refused to participate, one was lost to follow-up, and 485 were deceased during follow-up.

Table 1 displays baseline characteristics and therapy data of follow-up participants, nonparticipants, and deceased participants. At baseline, follow-up participants had a mean age of 62.1 years (standard deviation (SD), 5.9 years), 67.3 % were married, and 55.3 % had a low education. The mean BMI was 25.8 kg/m2 (SD, 4.3) and 15.5 % had a BMI ≥30 kg/m2 at baseline interview. About half of the women were current MHT users at diagnosis (49.6 %). Regarding cancer treatment, 78.7 % had received radiotherapy and 45.8 % chemotherapy. Furthermore, 68.9 % women had been treated with tamoxifen, 46.2 % with AI, 1.8 % with trastuzumab, and 2.6 % with bisphosphonates.

Compared with follow-up participants, nonparticipants/deceased women were older, less often married, less frequently MHT users, less physically active, and more likely to be a current smoker (Table 1). A larger proportion of the nonparticipants reported at baseline to never consume alcohol. Nonparticipants more frequently suffered from hypertension and diabetes at baseline. Fewer nonparticipants had radiation therapy, and more nonparticipants were treated with tamoxifen. Additionally, deceased women had a higher BMI, were less educated, and more frequently widowed than participants. Deceased women had higher tumor stages and were more likely to have received chemotherapy and/or radiation. At baseline, 12.6 % of the deceased suffered from diabetes compared to 6.6 % of follow-up participants.

Prior to breast cancer, 951 (37.4 %) follow-up participants had had a diagnosis of hypertension and 312 (12.3 %) reported to be newly diagnosed with hypertension (Table 2). CVD had been diagnosed before cancer diagnosis in 286 (11.3 %) and newly occurred in 267 (10.5 %) women at follow up. Osteoporosis was present in 236 (9.3 %) women prior to breast cancer, and newly diagnosed in 267 (10.5 %) women after the breast cancer diagnosis.
Table 2

Reported comorbidities stratified by time of onset in relation to breast cancer diagnosis (2,542 follow-up participants)

 

Never

Before or at diagnosis

After diagnosis

Unknown

Total

n

%

n

%

n

%

n

%

n

%

Hypertension

1,249

49.1

951

37.4

312

12.3

30

1.2

2,542

100

CVD

1,902

74.8

286

11.3

267

10.5

87

3.4

2,542

100

 Angina pectoris

2,350

92.4

80

3.1

90

3.5

22

0.9

2,542

100

 Myocardial infarction

2,481

97.6

39

1.5

19

0.7

3

0.1

2,542

100

 Stroke

2,440

96.0

53

2.1

44

1.7

5

0.2

2,542

100

 Arterial occlusion disease

2,103

82.7

156

6.1

197

7.7

86

3.4

2,542

100

Osteoporosis

1,983

78.0

236

9.3

267

10.5

56

2.2

2,542

100

Results of the Cox proportional hazards regression models for hypertension, CVD, and osteoporosis are shown in Table 3. Higher education (HR, 0.54; 95 % CI, 0.37–0.79) was associated with decreased risk for hypertension, whereas age (HR per year of age, 1.03; CI, 1.00–1.05) and BMI at baseline interview (BMI >25 kg/m2 (HR, 1.55; CI, 1.13–2.15); BMI ≥30 kg/m2 (HR, 1.90; CI, 1.24–2.90), respectively) were associated with increased risk. Recreational physical activity (sports and cycling) was not related to hypertension. Compared to those with stage I tumors, patients with in situ carcinoma were at higher risk (HR, 1.78; CI, 1.11–2.86), while higher tumor stages were not associated with risk of hypertension. Higher tumor stages were not associated with risk of hypertension. Compared to women with her2/neu negative tumors, those with her2/neu positive tumors who received trastuzumab had a 2.16-fold risk of developing hypertension (CI, 1.07–4.33). Women with her2/neu positive tumors not treated with trastuzumab showed no significant association with hypertension.
Table 3

Cox proportional hazards models of reported newly diagnosed comorbidities in the MARIEplus study

  

Hypertension (299 events/1,503)

CVD (254 events/2,058)

Osteoporosis (259 events/2,163)

n

HR

95 % CI

n

HR

95 % CI

n

HR

95 % CI

Age (years)

(Continuous)

1,503

1.03

(1.00, 1.05)*

2,058

1.04

(1.01, 1.07)**

2,163

1.01

(0.98, 1.04)

Region (Hamburg)

RNK (ref.a)

825

0.79

(0.61, 1.02)

1,101

0.74

(0.56, 0.97)*

1,146

0.78

(0.59, 1.02)

Postmenopausal b

 

1,328

0.93

(0.60, 1.44)

1,847

0.86

(0.51, 1.44)

1,943

0.88

(0.55, 1.41)

Marital status

Married (ref.)

1,020

1.0

 

1,407

1.0

 

1,475

1.0

 

Single

97

0.90

(0.53, 1.53)

131

0.77

(0.39, 150)

132

0.69

(0.34, 1.41)

Seperated/divorced

211

0.95

(0.67, 1.36)

276

0.86

(0.58, 1.29)

276

1.14

(0.79, 1.64)

Widowed

175

1.13

(0.79, 1.62)

244

1.21

(0.84, 1.73)

274

1.42

(0.99, 2.02)

Education

Low (ref.)

763

1.0

 

1,095

1.0

 

1,171

1.0

 

Medium

457

0.83

(0.64, 1.09)

622

0.84

(0.63, 1.14)

636

1.00

(0.76, 1.33)

High

283

0.54

(0.37, 0.79)**

341

0.87

(0.59, 1.28)

356

0.80

(0.55, 1.17)

BMI (kg/m2)

<22.5 (ref.)

419

1.0

 

472

1.0

 

456

1.0

 

<25

458

1.16

(0.82, 1.62)

557

0.92

(0.63, 1.35)

574

0.89

(0.63, 1.23)

<30

483

1.55

(1.13, 2.15)**

744

1.05

(0.73, 1.51)

798

0.66

(0.47, 0.92)*

≥30

143

1.90

(1.24, 2.90)**

285

1.77

(1.16, 2.71)**

335

0.50

(0.31, 0.79)**

Age at menarche (years)

<12 (ref.)

128

1.0

 

160

1.0

 

182

1.0

 

12–<15

992

0.84

(0.57, 1.24)

1,375

1.13

(0.70, 1.84)

1,423

1.02

(0.64, 1.62)

15+

383

0.79

(0.51, 1.22)

523

1.19

(0.70, 2.01)

558

1.10

(0.67, 1.82)

Parity (≥ 28. weeks)

0 (ref.)

290

1.0

 

378

1.0

 

384

1.0

 

1

436

1.13

(0.80, 1.60)

593

1.31

(0.86, 1.99)

620

1.24

(0.84, 1.83)

2

559

0.93

(0.65, 1.32)

768

1.13

(0.75, 1.72)

807

0.94

(0.63, 1.40)

3+

218

0.72

(0.46, 1.13)

319

1.18

(0.74, 1.91)

352

1.02

(0.64, 1.62)

MHTc

Never (ref.)

433

1.0

 

604

1.0

 

662

1.0

 

Past

288

1.36

(0.98, 1.90)

420

1.22

(0.85, 1.75)

444

0.95

(0.65, 1.39)

Current

782

1.01

(0.76, 1.34)

1,034

1.13

(0.83, 1.53)

1,057

1.09

(0.81, 1.46)

Phys. Activity, METd

0 (ref.)

227

1.0

 

336

1.0

 

371

1.0

 

>0–<2

361

0.98

(0.67, 1.42)

529

0.81

(0.55, 1.20)

560

0.76

(0.52, 1.12)

<24

327

0.82

(0.56, 1.21)

439

1.08

(0.73, 1.58)

451

0.79

(0.53, 1.18)

<42

305

0.95

(0.64, 1.40)

397

0.84

(0.55, 1.27)

415

0.84

(0.57, 1.26)

≥42

283

0.85

(0.57, 1.28)

357

0.57

(0.36, 0.91)*

366

0.75

(0.49, 1.15)

Smoking status

Never smoker (ref.)

722

1.0

 

1,074

1.0

 

1,131

1.0

 

Former smoker

470

1.10

(0.84, 1.44)

600

0.94

(0.69, 1.26)

643

0.93

(0.70, 1.25)

Current smoker

311

1.09

(0.79, 1.51)

384

0.92

(0.64, 1.32)

389

1.02

(0.72, 1.44)

Alcohol intake

0 g/day (ref.)

186

1.0

 

258

1.0

 

274

1.0

 

>0–<19 g/d

1,107

0.83

(0.59, 1.17)

1,540

0.80

(0.56, 1.14)

1,603

1.14

(0.77, 1.68)

19+ g/d

210

1.15

(0.74, 1.76)

260

1.12

(0.70, 1.81)

286

0.87

(0.51, 1.48)

Tumor stage

0 in situ

103

1.78

(1.11, 2.86)*

140

1.06

(0.59, 1.92)

149

1.07

(0.61, 1.89)

1 (ref.)

694

1.0

 

931

1.0

 

963

1.0

 

2

535

1.10

(0.81, 1.48)

743

0.92

(0.66, 1.27)

788

1.05

(0.76, 1.45)

3/4

119

1.33

(0.83, 2.14)

175

1.19

(0.74, 1.93)

194

1.18

(0.71, 1.94)

Unknown

52

1.22

(0.62, 2.42)

69

0.78

(0.36, 1.72)

69

1.31

(0.67, 2.56)

Radiotherapy

 

1,195

1.14

(0.83, 1.56)

1,616

1.00

(0.73, 1.38)

1,716

0.75

(0.56, 1.02)

Chemotherapy

 

717

0.88

(0.65, 1.19)

959

1.02

(0.73, 1.41)

1,022

0.88

(0.64, 1.21)

Tamoxifen

 

1,039

0.99

(0.76, 1.29)

1,443

1.02

(0.76, 1.37)

1,500

0.83

(0.63, 1.10)

Aromatase inhibitors

 

693

1.16

(0.90, 1.49)

949

1.42

(1.09, 1.84)**

1,017

2.15

(1.64, 2.82)**

Her2/neu, trastuzumab

Neg./no trastuzumab

1,118

1.0

 

1,543

1.0

 

1,612

1.0

 

Pos./trastuzumab

32

2.16

(1.07, 4.33)*

41

1.62

(0.70, 3.74)

40

0.57

(0.14, 2.34)

Pos./no trastuzumab

230

0.73

(0.50, 1.06)

304

1.47

(1.06, 2.05)*

317

1.10

(0.78, 1.56)

Unknown

130

1.13

(0.77, 1.64)

170

0.88

(0.55, 1.39)

194

1.19

(0.79, 1.79)

Bisphosphonates

 

53

1.29

(0.67, 2.40)

55

1.09

(0.50, 2.37)

 

 

Diabetes at diagnosis

 

49

0.73

(0.37, 1.44)

109

1.46

(0.92, 2.31)

 

Hypertension at diagnosis

 

 

721

0.83

(0.63, 1.10)

 

aref. reference

bDichotomous variables yes vs. no, no = reference category

cMHT menopausal hormone therapy (estrogen and/or progesterone)

dMET metabolic equivalents × hours/week, physical activity refers to recreational PA since age of 50 years including sports and cycling

*P < 0.05; **P < 0.01

For CVD, age (HR per year of age, 1.04; CI, 1.01–1.07), study region (HR for Hamburg, 0.74; CI, 0.56–0.97), BMI ≥30 kg/m2 (HR, 1.77; CI, 1.16–2.71), and the intake of AI (HR, 1.42; CI, 1.09–1.84) were positively associated with risk of disease. Patients with recreational physical activity of more than 42 MET/week had a reduced CVD risk (HR, 0.57; CI, 0.36–0.91), while lower levels were not associated with CVD. Furthermore, women with her2/neu positive tumors not treated with trastuzumab as well as those who received trastuzumab had a higher risk for CVD (HR, 1.41; CI, 1.01–1.97; HR, 1.61; CI, 0.70–3.73, respectively) compared with her2/neu negative tumors, although the latter association was statistically not significant.

The risk for osteoporosis was reduced in women with a higher BMI (>25 kg/m2 (HR, 0.66; CI, 0.47–0.92); ≥30 kg/m2 (HR, 0.50; CI, 0.31–0.79)) compared to women with a lower BMI. Patients treated with AI had a 2.15-fold risk (CI, 1.64–2.82) of a diagnosis of osteoporosis.

All other variables included in the models, e.g., smoking and alcohol consumption, showed no significant association with the incidence of self-reported hypertension, CVD, and osteoporosis after the breast cancer diagnosis.

Discussion

The present follow-up study of a large breast cancer cohort investigated potential demographic, lifestyle, and therapy-related determinants of self-reported new comorbidities during 5 years of follow-up. Among the examined factors, a few demographic and modifiable lifestyle factors were associated with hypertension, CVD, and/or osteoporosis, i.e., age, education, BMI at diagnosis, and a higher level of recreational physical activity. Of the therapy-related factors, treatment with AI was an independent risk factor for CVD and osteoporosis. In addition, patients with her2/neu positive tumors who had received trastuzumab had an elevated risk for hypertension. Results suggested a higher risk for CVD in her2/neu-positive tumors not related to trastuzumab.

Higher BMI at baseline was statistically significantly associated with all of the investigated comorbidities after breast cancer diagnosis. While increasing BMI was associated with increased risk for CVD and hypertension, it was associated with decreased risk of osteoporosis. The latter effect might be explained by a lower bone density and thus increased fracture risk in lean women [16]. The BMI used for this analysis was self-reported BMI at baseline interview, which is known to underestimate real BMI [17, 18]. For instance, in a subsample of the EPIC cohort, self-reported and measured height and weight agreed well, however, heavier and older women tended to overestimate height and underestimate weight [18]. Thus, the effects of BMI on hypertension, CVD, and osteoporosis in our study might be underestimated as well. Our findings regarding physical activity and CVD are in line with a recent meta-analysis of healthy cohorts [19] confirming that independently from BMI, recreational physical activity had a preventive effect on the development of CVD including stroke (combined HR for low versus high level 0.73; CI, 0.68–0.78).

In the present study, the use of AI was significantly associated with an increased risk for CVD. AI as part of the adjuvant systemic therapy are preferably used to prevent breast cancer recurrence in postmenopausal women with endocrine-responsive node-positive breast cancer [20]. In a meta-analysis of seven randomized clinical trials with 30,000 patients comparing AI with tamoxifen as initial adjuvant therapy in postmenopausal women with early-stage breast cancer, 5 years intake of AI compared with tamoxifen alone or in combination with 2–3 years of AI was associated with a higher risk of CVD (OR 1.26, CI 1.10–1.43) [10]. On the other hand, Ligibel et al. found no association of intake of AIs with myocardial infarction or ischemic stroke in a register-based study with a short follow-up of 30 months [21].

Intake of AIs doubled the risk for osteoporosis in our study. This result is in accordance with known adverse effects on bone mineral density and enhanced bone resorption in AI-treated women with early breast cancer [22, 23] and healthy postmenopausal women [24]. In most clinical trials of different AIs, impaired bone health and an increased fracture risk have been reported in patients treated with AI compared to patients treated with tamoxifen [2527]. In a meta-analysis of randomized clinical trials, 5 years of AI use compared to tamoxifen alone or sequential use with 2–3 years of an AI was associated with 47 % risk increase for bone fractures (OR, 1.47; CI, 1.34–1.61) [10].

We found a statistically significantly increased risk for hypertension associated with the intake of trastuzumab. Trastuzumab is known to reduce the risk of relapse in her2/neu receptor-positive patients up to 50 % and has become a standard component of adjuvant treatment in nonmetastatic her2/neu-positive tumors since 2006 [28]. Therefore, in the present cohort, most patients with her2/neu-positive tumors did not receive trastuzumab, and as a consequence the power to identify an increased risk was low. The herceptin adjuvant trial (HERA) clinical study mentioned hypertension as a side effect of trastuzumab in less than 1 % overall, however, its frequency was doubled in the treatment compared to the observational arm [29]. Trastuzumab has also been shown to be a risk factor for cardiovascular events in several studies [11, 30, 31], as well as in the latest update of the HERA study [32]. A retrospective cohort study, including 12,500 women aged 18 years or older diagnosed with breast cancer between 1999 and 2007, found a HR of 4.12 (CI, 2.30–7.42) for women who received trastuzumab without anthracycline and a HR of 7.19 (CI, 5.00–10.35) for women who received anthracycline plus trastuzumab to have an incident heart failure and/or cardiomyopathy [33]. Similarly, trastuzumab-related cardiac events after chemotherapy were observed in the NSABP B-31 randomized clinical trial [34]. In the present study, the risk for CVD was statistically significantly elevated in her2/neu-positive patients not treated with trastuzumab but not in her2/neu-positive patients treated with trastuzumab, though hazard ratios were of the same magnitude. Thus, it appears that there is no independent effect of trastuzumab on CVD after 5 years of follow-up, but rather her2/neu-positive tumors might exert a risk for CVD.

Risks for any of the comorbidities under study were not significantly related to radiotherapy, although there was a nonsignificant inverse association with osteoporosis that had not been reported so far. However, for cardiac adverse events to occur, the latency period might extend beyond our observation period. Cardiac events may also be restricted to subgroups as has been suggested for left side compared to right side breast cancer, the former being more adjacent to the heart, and thus might be exposed to relatively higher dose in the usual range [35, 36]. Chemotherapy was not associated with hypertension or CVD, although 70 % of participants received anthracyclines as primary adjuvant therapy, which is a well-known risk factor for long-term cardiac events, followed by 20 % cyclophosphamid, methotrexat, 5-fluoruracil (CMF)-like regimes. Hence, it is unlikely that a lack of effect might be due to the surrogate nature of this parameter not differentiating between anthracyclines and other sorts of chemotherapeutics. Similarly, there were no associations of tamoxifen use with CVD in one direction or the other. A reason could be that we did not separately investigate diseases that have been shown to be differentially associated with tamoxifen, e.g., deep venous thrombosis, stroke, and pulmonary embolism exerted adverse effects, while risk of myocardial infarction tended to decrease after tamoxifen use (reviewed in [31, 37]). In summary, the results regarding radiotherapy and tamoxifen but not chemotherapy are compatible to the single other study that assessed determinants of comorbidities after breast cancer [12], though that study used an unspecific summary measure as outcome.

The strength of this study is its large sample size and the high quality of the MARIE study data as was shown in the test–retest reliability of self-reported reproductive and lifestyle data [38]. The present study has also potential limitations. Though older breast cancer patients seem to be well able to provide information about their comorbidities [39], recall and information bias is possible. An earlier Dutch study reported a very good agreement with medical records in precisely defined chronic diseases, such as diabetes and cardiac diseases, while agreement for osteoarthritis and cerebrovascular conditions were low, explained by differential under- and over-reporting according to age, sex, physical mobility, and understanding [40]. Additionally, the severity of the comorbidities was not known. However, frequency of self-reported comorbidities in the present study compare well to other studies, e.g., the German telephone survey GEDA [41]. Prevalences of self-reported hypertension in German women at age of 50–64 years were 34 % and from 65 to 74 years 52.2 %. Myocardial infarction and stroke were slightly higher in GEDA, but in a general practices’ sample of around 22,000 women, it was reported in 3.5 and 2 %, respectively [42]. Estimates for the prevalence of osteoporosis strongly depend on data source and age, and were only slightly higher in a recent Canadian survey (age 50–70 years, 14.7 %; >71 years, 31.1 %) [43], but were comparable in GEDA (age 50–74 years, 13.4 %; >75 years, 26.7 %, see supplemental Table 1 in Fuchs et al. [41]. Thus, over-reporting seems not to be a relevant issue. Selection bias due to a healthy survivor effect and a healthy responder effect may have affected our results leading most likely to a loss in precision of estimates as follow-up participants had a better tumor stage and less comorbidities, were younger, and leaner than nonparticipants and deceased patients.

Conclusion

Results of the present study indicate that known risk factors for CVD, particularly BMI, also operate in breast cancer patients 5 years after diagnosis. Our study supports earlier findings of an adverse effect of AI on development of osteoporosis and CVD in a population-based cohort of breast cancer survivors. There is an elevated risk for hypertension after treatment with trastuzumab, whereas no excess risk for CVD was detected. As the number of patients who received trastuzumab was small, further research is needed to confirm our results.

Clinical implications and implications for cancer survivors

Follow-up care for breast cancer survivors needs to be attentive to potentially treatment-related comorbidities as well as in monitoring known risk factors for CVD and osteoporosis.

Advising particularly the elderly, less educated breast cancer survivors and those with higher BMI to reduce weight and perform regular sports activity could help to prevent new comorbidities. The higher risk for osteoporosis in low-weight patients may be balanced by early donation of bisphosphonates.

Acknowledgments

The MARIEPLUS study including this work was funded by the Deutsche Krebshilfe e.V., project number 108253/108419.

We are grateful to all participants of the MARIEplus study and thank our colleagues Ursula Eilber and Till Olchers for providing excellent support in collecting and preparing the data.

Conflict of interest

The authors declare that they have no conflict of interest.

Copyright information

© Springer Science+Business Media New York 2014