dove medical press web viewbmi, smoking status, alcohol consumption, ethnicity, rheumatoid...

$: Dove Medical Press Web viewBMI, smoking status, alcohol consumption, ethnicity, rheumatoid arthritis, osteoporosis and fractures, use of other osteoporosis drugs, vitamin D, NSAIDs,$
Bisphosphonates and primary breast cancer risk: an updated systematic review and meta-

analysis involving 963,995 women

Yupeng Liu, PhD; Xiaosan Zhang, PhD; Hongru Sun, MSc; Shu Zhao, PhD; Yuxue Zhang, MSc; Dapeng Li, MSc;

Qingyuan Zhang, PhD; Yashuang Zhao, PhD

SUPPLEMENTARY MATERIAL

Correspondence to: Dr. Yupeng Liu Department of Preventive Medicine, School of Public Health and Management

Wenzhou Medical UniversityWenzhou 325035, P. R. China

Tel/Fax: +86 (0) 577 8669 9182E-mail: [email protected] Professor Yashuang Zhao

Department of Epidemiology, School of Public HealthHarbin Medical UniversityHarbin 150086, P. R. China

Tel/Fax: +86 (0) 451 8750 2823E-mail: [email protected]

or Professor Qingyuan Zhang Department of Medical Oncology, The Third Affiliated Hospital

Harbin Medical UniversityHarbin150081, P. R. China

Tel/Fax: +86 (0) 451 8629 8116E-mail: [email protected]

Table of Contents Pages

Supplementary Discussion 1

Supplementary Tables 2-10

1 eTable 1. Characteristics of cohort studies included in this meta-analysis 2

2 eTable 2. Characteristics of case-control studies included in this meta-analysis 3

3 eTable 3. Characteristics of randomized controlled trials addressing the association between bisphosphonates and the risk of primary breast cancer

4

4 eTable 4. Sensitivity analyses by comparisons of the pooled results of the maximally and minimally adjusted effect estimates of the association between bisphosphonates and breast cancer risk

5

5 eTable 5. Sensitivity analyses excluding the study that has the most relative weight in subgroups 6

6 eTable 6. Characteristics of cohort studies addressing the association between bisphosphonates and the risk of contralateral breast cancer

7

7 eTable 7. Characteristics of case-control studies addressing the association between bisphosphonates and the risk of contralateral breast cancer

8

8 eTable 8. Publication bias 9

9 eTable 9. Levels of evidence for associations between bisphosphonates and primary breast cancer risk 10

Supplementary Figures 11-22

10 eFigure 1. Forest plot for the association between bisphosphonates and the risk of primary invasive breast cancer

11

11 eFigure 2. Forest plot for the association between bisphosphonates and primary breast cancer risk in postmenopausal women

12

12 eFigure 3. Subgroup analyses of the association between bisphosphonates and primary breast cancer risk according to duration of bisphosphonate use.

13

13 eFigure 4. Cumulative meta-analysis of the association between bisphosphonates and primary breast cancer risk according to duration of bisphosphonate use

14

14 eFigure 5. A sensitivity analysis by excluding retrospective case-control studies 15

15 eFigure 6. A sensitivity analysis by excluding randomized controlled trials 16

16 eFigure 7. Forest plot for the association between bisphosphonates and the risk of contralateral breast cancer in women with primary breast cancer

17

17 eFigure 8. A sensitivity analysis by combining the results from studies addressing the associations of bisphosphonates with the risk of both primary and contralateral breast cancer

18

18 eFigure 9. Meta-regression analyses of the association between bisphosphonates and primary breast cancer risk

19-20

19 eFigure 10. Funnel plots of potential publication 21

20 eFigure 11. Risk of bias of randomized controlled trials 22

Appendix 23-25

21 Appendix A. PRISMA Checklist 23-24

22 Appendix B. Methodological quality score of the cohort and case-control studies 25

Supplementary Discussion

Additionally, we assessed the statistical power of these two RCTs by using the software of PASS version 11.0.7 (NCSS

LLC., USA). In the Hue’s study, the power of statistical analyses for the association between bisphosphonates and BCa

was only 19.53% and 8.58% for FIT and HORIZON-PFT, respectively, based on the sample size (6194 participants in

FIT and 7580 in HORIZON-PFT), the incidence of breast cancer in the control groups (1.49% in FIT and 0.77% in

HORIZON-PFT), and the effect estimates (1.24 for FIT and 1.15 for HORIZON-PFT). Even though pooling the data

from the two trials, the power was only 21.71%, indicating an overall 78.29% risk of type II error. Given the limited

number of BCa patients during the follow-up period (165 patients from the combined dataset), this low power is not

surprising. In contrast, the included observational studies, especially prospective cohort studies, with a large sample size,

have an extremely high power of statistical analyses to detect the association. The pooling analysis of cohort studies had

a power of 99.81% (with a two-sided alpha of 0.05) to detect a 10% reduction of the risk of developing primary BCa in

the bisphosphonate users.

1 / 25

SUPPLEMENTARY TABLES

eTable 1. Characteristics of cohort studies included in this meta-analysisStudy, Country

Population characteristicsmean age (SD)/median age (range), years

Study period,Follow-up duration mean (SD) , years

No. of BPs-users(No. of cases)

No. of nonusers(No. of cases)

Types of BPs, No. of users (%)

Data sources of BPs prescription

BCa diagnosis Adjusted/controlled factors Quality Score* (%)

Chiang 2012 16

Taiwan, ChinaPostmenopausal women aged over 55BPs-users: 73.5(8.4)Non-users: 73.4(8.4)

1998-2010BPs-users: 4.3(2.5)Non-users: 4.9(2.6)

6906 (44) 20697 (165) All alendronate,Alendronate: 6909 (100.0)

NHIRD Hospital diagnosis,mostly confirmed by the pathological examination

Age, osteoporosis, hypertension, diabetes, benign breast disease, morbid obesity, statin use, chronic obstructive pulmonary disease, oestrogen use, dyslipidemia, chronic kidney disease, coronaryartery disease, colorectal polyp, etc.

80

Cardwell 2011 17

the United KingdomUKGPRD81% female with a mean age of 70 Both groups: 70.0(11.4)

1996-2006BPs-users: 4.5 (2.6) Non-users: 4.4 (2.6)

34049 (369) 34049 (501) Any BPs,Alendronate: 23834 (70.0)Nitrogen containingBPs: 29623 (87.0)

UKGPRD Hospital diagnosis,identified from the UKGPRD for diagnosis

Age, sex, BMI, alcohol, smoking, NSAID prescription, glucocorticoid steroid, vitamin D prescription and calcium, history of osteoporosis or osteopenia etc.

86

Vestergaard 2011 24

DenmarkWomen aged over 40Both groups: 71.1(10.7)

1996-2006NS

81329 (884) 261322 (3465) Alendronate: 45162(51.8)Etidronate: 34103 (39.2)Risedronate: 1156 (1.3)Ibandronate: 540 (0.6)Clodronate: 339 (0.4)Pamidronate: 21 (<0.1)Zoledronate: 8 (<0.1)

DNPD Hospital diagnosis,identified from the Danish cancer registry

Age, use of systemic hormone therapy, irradiation, chemotherapy, alcoholism, etc.

78

Chlebowski 2010 25

the United StatesPostmenopausal womenBoth groups: 63.8 (50-79)

1993-2005Both groups: 8.5

2816 (64) 151952 (5092) Alendronate: 2527 (89.7)Etidronate: 285 (10.1)Pamidronate: 1 (<0.1)Tiludronate: 1 (<0.1)More than one: 2 (0.1)

An interview-administered questionnaire

Hospital diagnosis, verified at each clinic by medical record and pathology report

Age, ethnicity, smoking, alcohol use, physical activity, BMI, mammogram in the last 2 years, chemotherapy, calcium, vitamin D, 5-year risk of hip fracture, Gail 5-year risk of BCa, etc.

81

Fournier 201718

FrenchThe E3N cohort

Postmenopausal Women 62.8 (6.4)BPs-users: 64.8 (6.6)Non-users: 62.3 (6.3)

2004-2011Both groups: 7.2 (1.7)

12935(308) 51503(2099) Any BPs†, Alendronate: 14071Risedronate: 11097Ibandronate: 4246Etidronate: 1461Zoledronate: 693Tiludronate: 2

A series of questionnaires

Self-report using a follow-up questionnaire and 95% cases with pathology reports

BMI, years of schooling, alcohol intake, personal history of fractures, osteoporosis and bone densitometries, parity and age at first birth, use of oral contraceptives, time since menopause, history of breast cancer in first-degree relatives,personal history of benign breast disease, hormone replacement therapy use, self-report of a mammogram performed during the previous follow-up cycle, raloxifene, and vitamin D

89

Abbreviations: BCa, breast cancer; BMI, body mass index; BPs, bisphosphonates; CIPD, the Catastrophic Illness Patient Database; DNPD, the Danish National Prescription Database; NHIRD, the National Health Insurance Research Database; NS, not stated; NSAID, nonsteroidal anti-inflammatory drug; SD, standard deviation; UKGPRD, the United Kingdom General Practice Research Database; The E3N cohort, the French prospective Etude Epidemiologique aupres de femmes de la Mutuelle Generale de l'Education Nationale cohort.*The quality score of each study is presented as a percentage of the maximum score.†Did not report the numbers of BPs users but reported the number of person-years of BPs users.

2 / 25

eTable 2. Characteristics of case-control studies included in this meta-analysis

Study, Country,Study design


Study period No. of cases(No. of BPs-users)

No.of controls(No. of BPs-users)




Rennert 2010 34

IsraelCase-control(Population-based individually matched by age, residence, and ethnic group)

Postmenopausal womenCase: 63.6Control: 65.6

2000- 1832 (193) 2207 (326) Alendronate: 519 (86.7)

CHS Hospital diagnosis,identified independently by the diagnosing hospitals

Age, residence, ethnicity, BMI, age at first pregnancy, family history of BCa, first degree, sports activity, fruits consumption, etc.

78

Newcomb 2010 33

the United StatesCase-control(Population-based, frequency matched by age)

Women aged 20–69Case: 54.2(8.9)Control: 54.8(8.9)

2003-2006 2936 (106) 2975 (161) Any BPs† An interview-administered questionnaire

Hospital diagnosis,identified from Wisconsin'smandatory cancer registry

Age at diagnosis, parity, age at first live birth, family history of breast cancer, BMI, menopausal status, type of hormone use, mammography, osteoporosis, smoking, height change etc.

73

Vinogradova 201319

the United Kingdoma nested case-control study from the QResearch and CPRD databases(Population-based individually matched by age, sex, practice and calendar year)

QResearch:Case: 69.4(9.8)Control: 69.4(9.7)CPRD:Case: 69.9(10.4)Control: 69.7(10.2)

1997-2011 QResearch:24489 (3827)CPRD:25444 (3769)

QResearch:113945 (17883)CPRD:118835 (17490)

Any BPsAlendronate Etidronate Risedronate Ibandronate

BNF From the two largest primary-care databases in the UK: QResearch and CPRD

BMI, smoking status, alcohol consumption, ethnicity, rheumatoid arthritis, osteoporosis and fractures, use of other osteoporosis drugs, vitamin D, NSAIDs, corticosteroids, acid-lowering drugs, years of data, family history of breast cancer, benign breast disease, use of oral contraceptives and hormone replacement therapy.

75

Abbreviations: BCa, breast cancer; BMI, body mass index; BNF, British National Formulary; BPs, bisphosphonates; CHS, the Clalit Health Services pharmacy records; NSAID, nonsteroidal anti-inflammatory drugs; SD, standard deviation; *The quality score of each study is presented as a percentage of the maximum score.†Did not report the types and numbers of BPs users.

3 / 25

eTable 3. Characteristics of randomized controlled trials addressing the association between bisphosphonates and the risk of primary breast cancer

Study, Country



No. of BPs-users(No. of Events)

No. of nonusers(No. of Events) BPs treatment BCa diagnosis Outcomes

/End-pointsJadad Score

Hue 2014 14

FIT a randomized double-blind placebo-controlled interventional trialthe United States

Postmenopausal women 55-81Alendronate: 68.1(6.2)Placebo: 68.2(6.1)

1992/1993-3.8 (0.8)

3103 (57) 3091 (46) Daily oral alendronate sodium (5mg/day for 2 years and 10 mg/day for 2 years)

Adverse events reports and supporting medical records†

Fracture 4

Hue 2014 14

HORIZON-PFT the United States

Postmenopausal women 65-89Zoledronic acid: 73.1 (5.3)Placebo: 73.0 (5.4)

2002-20112.8 (0.6)

3792 (33) 3788 (29) Annual intravenous infusions of zoledronic acid 5mg for 3 years

Adverse events reports and supporting medical records†

Fracture and BMD 4

†FIT and HORIZON-PFT were not initially designed to study breast cancer outcomes and identified new breast cancer cases using adverse events reports and supporting medical records.

4 / 25

eTable 4. Sensitivity analyses by comparisons of the pooled results of the maximally and minimally adjusted effect estimates of the association between bisphosphonates and breast cancer risk* Subgroups No. of

studies Effect estimate (95% CI) I2 (%) P-value for heterogeneity

Confounding RR† (P-value)

Total

Maximal 10 0.879 (0.826-0.936) 35.53 0.1241.060, (0.256)

Minimal 10 0.829 (0.766-0.898) 66.16 0.002

Case-control study

Maximal 3 0.834 (0.711-0.979) 61.79 0.073 1.090, (0.511)

Minimal 3 0.765 (0.625-0.936) 80.52 0.006

Cohort study

Maximal 5 0.874 (0.803-0.951) 33.18 0.2001.059, (0.373)

Minimal 5 0.825 (0.750-0.906) 53.89 0.070

Randomized controlled trials

Maximal 2 1.127 (0.808-1.573) 0.00 0.918 0.935, (0.772)

Minimal 2 1.205 (0.887-1.637) 0.00 0.815

Invasive BCa

Maximal 6 0.829 (0.692-0.994) 59.54 0.030 1.067, (0.602)

Minimal 6 0.770 (0.624-0.951) 75.53 0.001

Postmenopausal

Maximal 7 0.888 (0.798-0.989) 30.50 0.1951.043, (0.642)

Minimal 7 0.851 (0.737-0.982) 67.47 0.005

Abbreviations: BCa, breast cancer; BPs, bisphosphonates; CI, confidence interval.*We used a random-effects model to pool the effect estimates and therefore be more conservative.†Confounding RR was defined as the ratio of the pooled results of the maximally and the minimally adjusted effect estimates. If the P-value for confounding RR was less than 0.05, the impact of adjustment for confounders on the pooled results was statistically significant; otherwise, the confounding effect of the adjusted factors was small or null. For protective factors, if the confounding RR <1, it indicates that the adjusted confounders have obscured an inverse association; if the confounding RR >1, it indicates that the adjusted confounders have exaggerated an inverse association.

5 / 25

eTable 5. Sensitivity analyses excluding the study that has the most relative weight in subgroups*

Subgroups No. of studies Effect estimate (95% CI)Heterogeneity between-study

P-value for heterogeneity

between subgroupsI2 (%) P-value

Total 10 0.879 (0.826-0.936) 35.53 0.1240.738

Total† 9 0.864 (0.798-0.935) 30.08 0.178

Cohort study

Subtotal 5 0.874 (0.803-0.951) 33.18 0.2000.912

Subtotal† 4 0.882 (0.769-1.011) 47.30 0.128

Invasive BCa

Subtotal 6 0.829 (0.692-0.994) 59.54 0.030 0.647

Subtotal† 5 0.780 (0.646-0.941) 41.31 0.146

Postmenopausal

Subtotal 7 0.888 (0.798-0.989) 30.50 0.1950.539

Subtotal† 6 0.846 (0.757-0.946) 9.91 0.353

Duration of BPs use (year)

< 1

Subtotal 5 0.904 (0.842-0.970) 0.00 0.522 0.724

Subtotal† 4 0.873 (0.729-1.046) 1.33 0.385

≥1 (including 1-2, 2-3, 3-4, and >4)

Subtotal 19 0.745 (0.658-0.844) 74.61 <0.0010.779

Subtotal† 18 0.727 (0.647-0.818) 57.40 0.001

1-2

Subtotal 5 0.690 (0.534-0.893) 87.00 <0.001 0.561

Subtotal† 4 0.629 (0.527-0.750) 42.47 0.157

2-3

Subtotal 4 0.732 (0.554-0.967) 72.82 0.012 0.694

Subtotal† 3 0.797 (0.579-1.098) 67.65 0.045

3-4

Subtotal 6 0.862 (0.682-1.090) 59.40 0.0310.555

Subtotal† 5 0.941 (0.792-1.119) 0.00 0.665

>4

Subtotal 4 0.678 (0.525-0.875) 38.45 0.181 0.791

Subtotal† 3 0.721 (0.494-1.051) 54.51 0.111

Abbreviations: BCa, breast cancer; BPs, bisphosphonates; CI, confidence interval.*This sensitivity analysis was conducted only when the number of included studies in subgroup analyses was not less than 4. We used a random-effects model to pool the effect estimates and therefore be more conservative.†The pooled results of sensitivity analyses excluding the study that has the most relative weight in subgroups.

6 / 25

eTable 6. Characteristics of cohort studies addressing the association between bisphosphonates and the risk of contralateral breast cancerStudy, Country



No. of BPs-users(No. of cases)

No. of nonusers(No. of cases)




Kwan 201636

the United StatesThe KPNC and KPSC cohorts

KPNC 59.6 (12.5)BPs-users: 64.7 (11.0)Non-users: 58.8 (12.5)KPSC 59.7 (12.3)BPs-users: 64.5 (10.7)Non-users: 58.1 (12.3)

1996-2009KPNC BPs-users: 3.59Non-users: 6.67KPSC BPs-users: 3.41Non-users: 7.06

KPNC 1282 (35)KPSC1973 (26)

KPNC 7575 (199)KPSC5951 (172)

Any BPs,Alendronate: most commonly prescribed (>93 %)

From the Kaiser Permanente insurance records

Hospital diagnosis,from thecancer registries of the United States

Age at diagnosis, year of diagnosis, race/ethnicity, Charlson comorbidity index, AJCC stage, primary therapy, chemotherapy, ER/PR status, Her2 status, post-BC diagnosis osteopenia, osteoporosis, or fracture, and tamoxifen and aromatase inhibitor use

87

Korde 201837

the United StatesThe Quilt cohort joining the cases recruited into threeprevious population-based studies of breast cancer, the CARE, PACE and EMF studies

64.2 (45-79) 1993-201511.8 (2.5-17.6)

302 (23) 1511 (139) Any BPs,Alendronate 272(90.0)Risendronate 40(13.0)Zoledronate 12(4.0)

From medical record reviews

From the Cancer Surveillance System, the population-based SEER registry

Age at diagnosis and source study (the original three studies through which women were ascertained)

82

Abbreviations: BCa, breast cancer; BPs, bisphosphonates; KPNC, the Kaiser Permanente Northern California cohort; KPSC, the Kaiser Permanente Southern California cohort; SD, standard deviation. *The quality score of each study is presented as a percentage of the maximum score.

7 / 25

eTable 7. Characteristics of case-control studies addressing the association between bisphosphonates and the risk of contralateral breast cancer

Study, Country,Study design


Study period No. of cases(No. of BPs-users)

No.of controls(No. of BPs-users)




Monsees 201135

the United Statesa nested case-control study from the SEER database(Population-based individually matched by age atand year of first breast cancer diagnosis, ethnicity, historic stage of first breast cancer )

primary invasive, localized or regional SEER historic stage ER+ breast cancer women61.1 (40-79)

1990-2007 351 (32) 662 (85) Any nitrogenous BPsAlendronate:106(88.0)

The medical records of the oncology and/or primary care provider

SEER age and year of diagnosis with first primary breast cancer,SEER historic stage of the first primary breast cancer, county of residence, race/ethnicity and BMI at first breast cancer diagnosis

84

Abbreviations: BCa, breast cancer; BMI, body mass index; BPs, bisphosphonates; SD, standard deviation; SEER, the Surveillance Epidemiology and End Results.*The quality score of each study is presented as a percentage of the maximum score.

8 / 25

eTable 8. Publication bias

Subgroups No. of studies

P-value Summarized effect estimates (95% CI)*

For Begg's rank

correlation test

For Egger's linear

regression test

Adjusted

with trim and fillUnadjusted

Total 10 0.721 0.698 0.876 (0.822-0.932) 0.879 (0.826-0.936)

Case-Control study 3 0.296 0.117 0.920 (0.876-0.967) 0.834 (0.711-0.979)

Cohort study 5 0.807 0.881 0.839 (0.767-0.917) 0.877 (0.811-0.949)

Randomized controlled trial 2 NA NA NA 1.127 (0.808-1.573)

Invasive BCa 6 0.851 0.966 0.766 (0.637-0.922) 0.829 (0.692-0.994)

Postmenopausal 7 0.230 0.533 0.866 (0.779-0.962) 0.888 (0.798-0.989)

Duration of BPs treatment (year)

<1 5 0.624 0.835 0.904 (0.842-0.970) 0.904 (0.842-0.970)

Subtotal ≥1 19 0.132 0.053 0.734 (0.648-0.831) 0.745 (0.658-0.844)

1-2 5 0.462 0.061 0.818 (0.659-1.014) 0.690 (0.534-0.893)

2-3 4 0.734 0.915 0.796 (0.608-1.041) 0.732 (0.554-0.967)

3-4 6 0.060 0.014 0.826 (0.665-1.026) 0.862 (0.682-1.090)

>4 4 0.308 0.324 0.628 (0.484-0.816) 0.678 (0.525-0.875)

　　Abbreviations: BCa, breast cancer; BPs, bisphosphonates; CI, confidence interval; NA, not applicable. *The summarized effect estimate was calculated with a random-effects model.

9 / 25

eTable 9. Levels of evidence for associations between bisphosphonates and primary breast cancer risk

SubgroupsNo. of studies (%)

SubtotalLevel

of evidenceDecreased risk Increased risk No association

Total 5 (50) 0 5 (50) 10 Indecisive

Case-Control study 2 (67) 0 1(33) 3 Moderate

Cohort study 3 (60) 0 2 (40) 5 Moderate

RCTs 0 0 2(100) 2 Indecisive

Invasive BCa 3 (50) 0 3(50) 6 Indecisive

Postmenopausal 3 (43) 0 4 (57) 7 Indecisive

Duration of bisphosphonnate use (year)

< 1 2 (40) 0 3 (60) 5 Nil

≥ 1 (subtotal) 13 (68) 0 6 (32) 19 Strong

Abbreviations: BCa, breast cancer; BPs, bisphosphonates.

10 / 25

SUPPLEMENTARY FIGURES

eFigure 1. Forest plot for the association between bisphosphonates and the risk of primary

invasive breast cancer. Square markers indicate effect sizes of each study; horizontal lines, the

95% confidence intervals. The diamond data marker indicates the summarized effect size. The

vertical solid line indicates the overall pooled effect. Note that the studies are ranked in order of

their relative weights from random effects analysis. RR, relative risk.

Study

Fournier et al,18 2017

Overall (I2 = 59.54%, P = 0.03)

Chlebowski et al,25 2010

Rennert et al,34 2010

Newcomb et al,33 2010

RR (95% CI)

0.97 (0.83, 1.14)

0.83 (0.69, 0.99)

0.68 (0.52, 0.89)

0.70 (0.54, 0.90)

0.70 (0.54, 0.91)

Weight %

24.59

18.18

19.11

18.31

1.5 1 2Test for overall effect: z = -2.03; P = 0.04

Favor bisphosphonatesusers

Favor non-users

RR (95% CI)

100

Hue et al,14 2014_HORIZON-PFT

Hue et al,14 2014_FIT

1.15 (0.70, 1.90)

1.11 (0.71, 1.73)

11 / 25

eFigure 2. Forest plot for the association between bisphosphonates and primary breast cancer

risk in postmenopausal women. Square markers indicate effect sizes of each study; horizontal

lines, the 95% confidence intervals. The diamond data marker indicates the summarized effect size.

The vertical solid line indicates the overall pooled effect. Note that the studies are ranked in order of

their relative weights from random effects analysis. RR, relative risk.

Overall (I2 = 30.50%, P = 0.20)

Cardwell et al,17 2011




Study

Chiang et al,16 2012

0.89 (0.80, 0.99)

0.75 (0.63, 0.89)

0.90 (0.72, 1.12)

RR (95% CI)

0.80 (0.65, 0.99)

0.98 (0.85, 1.12)

0.92 (0.66, 1.28)

21.68

15.97

Weight %

17.05

27.25

8.65

1.5 1 2


Favor non-users

RR (95% CI)

Test for overall effect: z = -2.17; P = 0.03

100

Hue et al,14 2014 -FIT

Hue et al,14 2014 -HORIZON-PFT

1.11 (0.71, 1.73)

1.15 (0.70, 1.90)

12 / 25

eFigure 3. Subgroup analyses of the association between bisphosphonates and primary breast cancer

risk according to duration of bisphosphonate use. <1 indicates the subgroup of bisphosphonate users with

less than one year use of bisphosphonates; 1-2, one year or more but less than two years; 2-3, two years or

more but less than three years; 3-4, three years or more but less than four years; ≥4, 4 years or more. Square

markers indicate effect sizes of each study; horizontal lines, the 95% confidence intervals. The diamond data

marker indicates the summarized effect size. Note that the studies are ranked in order of their relative

weights from random effects analysis. RR, relative risk.

13 / 25



Study or Subgroup



Subtotal_<1 yr (I2 = 0.00%, P = 0.52)




Subtotal_>4 yr (I2 = 38.45%, P = 0.18)


Vinogradova et al,19 2013




Subtotal_3-4 yr (I2 = 59.40%, P = 0.03)


Subtotal_2-3 yr (I2 = 72.82%, P = 0.01)

Vinogradova et al,19 2013



Cardwell et al,17 2012Chlebowski et al,25 2010



Subtotal_1-2 yr (I2 = 87.88%, P < 0.01)



0.81 (0.57, 1.16)

0.57 (0.44, 0.74)

0.74 (0.61, 0.90)

1.04 (0.68, 1.59)

0.90 (0.84, 0.97)

0.61 (0.50, 0.75)

0.59 (0.47, 0.75)

0.77 (0.57, 1.04)

0.68 (0.52, 0.88)

RR (95% CI)

0.51 (0.34, 0.76)

0.93 (0.87, 0.99)

0.71 (0.54, 0.94)

0.93 (0.61, 1.42)

0.69 (0.41, 1.17)

0.86 (0.68, 1.09)

1.07 (0.81, 1.41)

0.73 (0.55, 0.97)

0.91 (0.84, 0.98)

0.61 (0.45, 0.83)

0.63 (0.42, 0.95)

0.94 (0.57, 1.55)0.83 (0.54, 1.28)

1.14 (0.78, 1.66)

0.50 (0.38, 0.66)

0.69 (0.53, 0.89)

0.86 (0.64, 1.16)

0.78 (0.54, 1.12)

17.42

21.38

21.89

14.87

21.65

28.07

5.54

Weight %

25.51

24.83

25.98

2.81

12.32

25.98

84.37

33.12

19.97

18.8722.50

3.51

19.31

19.60

3.77

1.25 .5 1 2

Test for subtotal effect:z = -2.79; P = 0.005





Test for subgroup differences:P = 0.007

RR (95% CI)


Favor non-users

100

100

100

100

100

Hue et al,

14 / 25

eFigure 4. Cumulative meta-analysis of the association between bisphosphonates and primary

breast cancer risk according to duration of bisphosphonate use. Square markers indicate effect

sizes of each study; horizontal lines, the 95% confidence intervals. The vertical solid line indicates

the overall pooled effect. RR, relative risk.

Rennert,34 2010_>5yr

Newcomb,33 2010_>2yr

Cardwell,17 2011_>4yr

Fournier,18 2017_<0.5yr

Rennert,34 2010_4-5yrFournier,18 2017_≥3yr

Vinogradova,19 2013_≥1yr

Fournier,18 2017_0.5-1yr

Rennert,34 2010_1-2yr

Rennert,34 2010_<1yr

Chlebowski,25 2010_<2yr

Rennert,34 2010_2-3yr

Chlebowski,25 2010_2-5yr

Newcomb,33 2010_<1yr

Cardwell,17 2011_2-3yr

Fournier,18 2017_1-3yr

Chlebowski,25 2010_>5yr

Newcomb,33 2010_1-2yr

Cardwell,17 2011_3-4yrRennert,34 2010_3-4yr

Cardwell,17 2011_1-2yr

Vinogradova,19 2013_<1yr

Study

0.77 (0.70, 0.85)

0.77 (0.69, 0.86)

0.78 (0.71, 0.86)

0.77 (0.57, 1.04)

RR (95% CI)

0.78 (0.71, 0.86)0.79 (0.71, 0.87)

0.79 (0.69, 0.89)

0.92 (0.63, 1.35)

0.81 (0.69, 0.96)

0.91 (0.72, 1.15)

0.76 (0.64, 0.90)

0.78 (0.69, 0.88)

0.78 (0.70, 0.86)

0.87 (0.73, 1.05)

0.78 (0.69, 0.87)

0.81 (0.72, 0.91)

0.78 (0.71, 0.86)

0.90 (0.84, 0.97)

0.76 (0.69, 0.85)0.76 (0.68, 0.85)

0.80 (0.69, 0.93)

0.90 (0.84, 0.97)

100

65.71

93.76

4.37

Weight %

91.2987.06

47.87

7.85

29.93

10.91

40.32

57.84

70.07

14.52

62.52

52.55

96.70

24.26

78.7074.98

35.73

21.96

1.5 1 2


Favor non-users

RR (95% CI)

Hue,14 2014_HORIZON-PFT_3-4yrHue,14 2014_FIT_3-4yr

0.78 (0.70, 0.86)

15 / 25

eFigure 5. A sensitivity analysis by excluding retrospective case-control studies. Square

markers indicate effect sizes of each study; horizontal lines, the 95% confidence intervals. The

diamond data marker indicates the summarized effect size. The vertical solid line indicates the

overall pooled effect. Note that the studies are ranked in order of their relative weights from random

effects analysis. HORIZON-PFT, Health Outcomes and Reduced Incidence with Zoledronic Acid

Once Yearly–Pivotal Fracture Trial; FIT, Fracture Intervention Trial; RR, relative risk.

Overall (I2 = 26.75%, P = 0.215)

Study



Hue et al,14 2014_FIT

Chiang et al,16 2012

Hue et al,14 2014_HORIZON-PFT

Vestergaard et al,24 2011


0.90 (0.85, 0.95)

0.98 (0.85, 1.12)

0.90 (0.72, 1.12)

1.11 (0.71, 1.73)

0.92 (0.66, 1.28)

1.15 (0.70, 1.90)

RR (95% CI)

0.86 (0.80, 0.93)

0.75 (0.63, 0.89)

13.43

6.17

1.66

2.93

1.32

Weight %

28.39

9.41

1.5 1 2

100


Favor non-users

Test for overall effect: z = -3.68; P < 0.001

RR (95% CI)

Vinogradova et al, 19 2013 0.92 (0.87, 0.97)

16 / 25

eFigure 6. A sensitivity analysis by excluding randomized controlled trials. Square markers

indicate effect sizes of each study; horizontal lines, the 95% confidence intervals. The diamond data

marker indicates the summarized effect size. The vertical solid line indicates the overall pooled

effect. Note that the studies are ranked in order of their relative weights from random effects

analysis. RR, relative risk.

Overall (I2 = 41.77%, P = 0.10)

Newcomb et al, 33 2010

Vestergaard et al, 24 2011

Chiang et al, 16 2012

Fournier et al, 18 2017

Cardwell et al, 17 2011

Rennert et al, 34 2010

Vinogradova et al, 19 2013

Chlebowski et al, 25 2010

Subtotal_Case-control (I2 = 61.79%, P = 0.07)

Subtotal_Cohort (I2 = 33.18%, P = 0.20)

Study

0.87 (0.82, 0.93)

0.70 (0.54, 0.91)

0.86 (0.80, 0.93)

RR (95% CI)

0.92 (0.66, 1.28)

0.98 (0.85, 1.12)

0.75 (0.63, 0.89)

0.80 (0.65, 0.99)

0.92 (0.87, 0.97)

0.90 (0.72, 1.12)

0.83 (0.71, 0.98)

0.87 (0.80, 0.95)

5.05

24.39

Weight %

3.43

13.67

10.09

7.47

28.99

6.91

1.5 1 2


Favor non-users

Test for subgroup differences: P = 0.37

Test for overall effect: z = -4.18;

17 / 25

eFigure 7. Forest plot for the association between bisphosphonates and the risk of

contralateral breast cancer in women with primary breast cancer. Square markers indicate

effect sizes of each study; horizontal lines, the 95% confidence intervals. The diamond data marker

indicates the summarized effect size of the observational studies included in this meta-analysis and

the vertical solid line indicates the overall pooled effect. The studies are ranked in order of their

relative weights from random effects analysis. Note that the EBCTCG’ study (Early Breast Cancer

Trialists’ Collaborative Group) is a meta-analysis of individual patient data from randomized

controlled trials of adjuvant bisphosphonate treatment in early breast cancer patients. RR, relative

risk.

Korde et al,37 2018

Monsees et al,35 2011

Overall (I2 = 0.00%, P = 0.63)

Study

Kwan et al,36 2016

RR (95% CI)

0.75 (0.47, 1.19)

0.77 (0.49, 1.19)

0.84 (0.66, 1.07)

0.96 (0.67, 1.38)

Weight %

26.87

29.52

43.61

1.5 2


Favor non-users

Test for overall effect: z = -1.43; P = 0.15

EBCTCG,43 2015 0.96 (0.74, 1.25)

18 / 25

eFigure 8. A sensitivity analysis by combining the results from studies addressing the

associations of bisphosphonates with the risk of both primary and contralateral breast cancer.

Square markers indicate effect sizes of each study; horizontal lines, the 95% confidence intervals.

The diamond data marker indicates the summarized effect size. The vertical solid line indicates the

overall pooled effect. Note that the studies are ranked in order of their relative weights from random

effects analysis EBCTCG, Early Breast Cancer Trialists’ Collaborative Group; HORIZON-PFT,

Health Outcomes and Reduced Incidence with Zoledronic Acid Once Yearly–Pivotal Fracture Trial;

FIT, Fracture Intervention Trial; RR, relative risk.

Rennert et al, 34 2010

Hue et al, 14 2014_HORIZON-PFT

Study

Hue et al, 14 2014_FIT

Chiang et al, 16 2012

Newcomb et al, 33 2010

Kwan et al, 36 2016

Fournier et al, 18 2017

EBCTCG, 43 2015

Chlebowski et al, 25 2010

Korde et al, 37 2018

Cardwell et al, 17 2011

Overall (I2 = 15.74%, P = 0.28)

Vestergaard et al, 24 2011

Vinogradova et al, 19 2013

Monsees et al, 35 2011

0.80 (0.65, 0.99)

1.15 (0.70, 1.90)

RR (95% CI)

1.11 (0.71, 1.73)

0.92 (0.66, 1.28)

0.70 (0.54, 0.91)

0.96 (0.67, 1.38)

0.98 (0.85, 1.12)

0.96 (0.74, 1.25)

0.90 (0.72, 1.12)

0.75 (0.47, 1.19)

0.75 (0.63, 0.89)

0.88 (0.84, 0.93)

0.86 (0.80, 0.93)

0.92 (0.87, 0.97)

0.77 (0.49, 1.19)

5.18

0.99

Weight %

1.25

2.22

3.36

1.88

10.65

3.46

4.74

1.18

7.33

24.09

32.38

1.29

1.5 1 2

100

Test for overall effect: z = -4.08;

19 / 25

eFigure 9. Meta-regression analyses of the association between bisphosphonates and primary

breast cancer risk according to (A) the mean age of participants, (B) the quality score, (C) the

total number of breast cancer patients and (D) the total number of bisphosphonate users in

each original study. Each circle represents an original study and the circle size is proportional to

the relative weight of the study using random-effects meta-regression analysis.

eFigure 9A

Mean Age of Participants in Each Original Study

Log

RR

54.5 63.8 70.0 73.4

0.00-0.04-0.08-0.12-0.16-0.20-0.24-0.28-0.32-0.36-0.40 60.0

Slope = 0.003P =0.65

0.080.14

eFigure 9B

Quality Score (%) of Each Original Study

Log

RR

73 75 78 81 86 89

0.00-0.04-0.08-0.12-0.16-0.20-0.24-0.28-0.32-0.36-0.40

Slope = 0.001P = 0.76

0.080.14

20 / 25

eFigure 9C

Total Number of Breast Cancer Patients in Each Original Study

Log

RR

5000 25000 37500 50000

0.00-0.04-0.08-0.12-0.16-0.20-0.24-0.28-0.32-0.36-0.40

12500200

Slope = 0.00001P = 0.11

0.080.14

eFigure 9D

Total Number of Bisphosphonate users in Each Original Study

Log

RR

200 12000 30000 42969 81329

0.00-0.04-0.08-0.12-0.16-0.20-0.24-0.28-0.32-0.36-0.40 60000

Slope = -0.00001P = 0.56

0.080.14

21 / 25

eFigure 10. Funnel plots of potential publication. White circles indicate individual studies; black

circles, trimmed studies after adjustment for potential publication bias; white diamonds, the

summarized effect estimates from the meta-analyses; black diamonds, the adjusted summary effect

estimates. The diagonal lines represent the 95% confidence interval of the summarized effect

estimate, which is indicated by the vertical solid line. RR, relative risk.

-1.0 -0.5 0.0 0.5 1.0

0.0

0.1

0.2

0.3

0.4

Stan

dard

Err

or

Log risk ratio

22 / 25

eFigure 11. Risk of bias of randomized controlled trials. (A) Judgements about each risk of bias

item for each included study; (B) Judgements about each risk of bias item presented as percentages

across all included studies.

A

B

23 / 25

APPENDICESAppendix A. PRISMA Checklist

Section/topic # Checklist item Reported on page # (or Brief description of how the criteria were handled in the meta-analysis)

TITLE

√ Title 1 Identify the report as a systematic review, meta-analysis, or both. Title page. Bisphosphonates and primary breast cancer risk: an updated systematic review and meta-analysis involving 963,995 women.

ABSTRACT

√ Structured summary 2Provide a structured summary including, as applicable: background; objectives; data sources; study eligibility criteria, participants, and interventions; study appraisal and synthesis methods; results; limitations; conclusions and implications of key findings; systematic review registration number.

Page 5-6. A structured abstract is provided on page 5-6.

INTRODUCTION

√ Rationale 3 Describe the rationale for the review in the context of what is already known. Page 7. Prevention of primary breast cancer (BCa) in women is of great public health importance. The existing results from observational epidemiologic studies focused on the association between bisphosphonates and primary BCa risk have been inconsistent.

√ Objectives 4 Provide an explicit statement of questions being addressed with reference to participants, interventions, comparisons, outcomes, and study design (PICOS).

Page 7. Does the use of bisphosphonates reduce the risk of primary breast cancer in women? The participants are women. The exposure or intervention is the use of any type of bisphosphonates. The outcome is the risk of developing primary breast cancer. The comparison is bisphosphonate users versus non-users. Study designs include cohort studies, case-control studies, and randomized controlled trials.

METHODS

√ Protocol and registration 5 Indicate if a review protocol exists, if and where it can be accessed (e.g., Web address), and, if available, provide registration information including registration number. Page 8. This meta-analysis was registered on PROSPERO (CRD42014014901).

√ Eligibility criteria 6 Specify study characteristics (e.g., PICOS, length of follow-up) and report characteristics (e.g., years considered, language, publication status) used as criteria for eligibility, giving rationale. Page 8-9. Detailed inclusion and exclusion criteria were described in the methods section.

√ Information sources 7 Describe all information sources (e.g., databases with dates of coverage, contact with study authors to identify additional studies) in the search and date last searched.

Page 8. Databases: MEDLINE, EMBASE, ProQuest, and Web of Science;Time period: the first search - through August 15, 2016; the updated search – from August 15, 2016 to June 25, 2018. We hand-searched reference sections and citation lists of the retrieved literature for additional references. If an abstract was identified, we had tried to contact with the authors to access their data.

√ Search 8 Present full electronic search strategy for at least one database, including any limits used, such that it could be repeated.

Page 8. We systematically searched several databases and conducted additional hand-searching to include all available studies. The detailed search terms are provided on page 8.

√ Study selection 9 State the process for selecting studies (i.e., screening, eligibility, included in systematic review, and, if applicable, included in the meta-analysis).

Page 9 and Figure 1. The eligibility determination for each study were performed independently by two authors (YP.L. and XS.Z.). Disagreements were resolved by discussion and consensus with a third author (HR.S.). The detailed process for study selection is provided as the flow chart (Figure 1).

√ Data collection process 10 Describe method of data extraction from reports (e.g., piloted forms, independently, in duplicate) and any processes for obtaining and confirming data from investigators.

Page 9. The data extraction, and quality assessment for each study were performed independently by two authors (YP.L. and XS.Z.). Disagreements were resolved by discussion and consensus with a third author (HR.S.).

√ Data items 11 List and define all variables for which data were sought (e.g., PICOS, funding sources) and any assumptions and simplifications made.

Page 9. Standard electronic forms specifically created for the present study were used to record the variables, which were described in the main text. Also see eTable 1-3, 6 and 7 in the Supplementary Materials.

√ Risk of bias in individual studies 12 Describe methods used for assessing risk of bias of individual studies (including specification of whether this was

done at the study or outcome level), and how this information is to be used in any data synthesis. Page 9-10. A quality-scoring system was used for assessding risk of bias of observational studies; and the Jadad score and Cochrane risk of bias tool were used to assess the quality of RCTs.

√ Summary measures 13 State the principal summary measures (e.g., risk ratio, difference in means). Page 9. The principal summary measures include relative risks, odds ratios and hazard ratio with 95% confidence intervals.

√ Synthesis of results 14 Describe the methods of handling data and combining results of studies, if done, including measures of consistency (e.g., I2) for each meta-analysis.

Page 10-11. We preferentially used the maximally adjusted effect estimates in this meta-analysis. The 95% confidence intervals of summarized estimates were all provided. I2 values and P values for heterogeneity were reported. The summarized results with corresponding 95% confidence intervals of sensitivity analyses were provided. The confounding RR values and the corresponding P values were provided. The E values for the pooled estimates and the lower or upper limits were provided.

24 / 25

Section/topic # Checklist item Reported on page # (or Brief description of how the criteria were handled in the meta-analysis)

√ Risk of bias across studies 15 Specify any assessment of risk of bias that may affect the cumulative evidence (e.g., publication bias, selective reporting within studies).

Page 11. We used funnel plots for asymmetry, the Begg’s rank correlation and Egger’s linear regression tests, and the Duval and Tweedie’s trim and fill method to assess the potential publication bias.

√ Additional analyses 16 Describe methods of additional analyses (e.g., sensitivity or subgroup analyses, meta-regression), if done, indicating which were pre-specified.

Page 10-11. We performed extensive sensitivity analyses, such as confounding RR, the E-value analysis, subgroup analyses, etc.

RESULTS

√ Study selection 17Give numbers of studies screened, assessed for eligibility, and included in the review, with reasons for exclusions at each stage, ideally with a flow diagram.

Figure 1. Details of the literature search process are outlined in the flow chart (Figure 1).

√ Study characteristics 18For each study, present characteristics for which data were extracted (e.g., study size, PICOS, follow-up period) and provide the citations.

Page 12 in the main text and eTable 1-3, 6 and 7 in the Supplementary Materials.

√ Risk of bias within studies 19 Present data on risk of bias of each study and, if available, any outcome level assessment (see item 12). Page 12 in the main text and eFigure 11 for RCTs and Appendix B for cohort and case-control studies in the Supplementary Materials.

√ Results of individual studies 20For all outcomes considered (benefits or harms), present, for each study: (a) simple summary data for each intervention group (b) effect estimates and confidence intervals, ideally with a forest plot.

Figure 2 in the main text and eFigure 1-8 in the Supplementary Materials.

√ Synthesis of results 21 Present results of each meta-analysis done, including confidence intervals and measures of consistency. Page 12-13. Figure 2 in the main text and eFigure 1-8 in the Supplementary Materials..

√ Risk of bias across studies 22 Present results of any assessment of risk of bias across studies (see Item 15). Page 14-15. eTable 8 and eFigure 10 in the Supplementary Materials.

√ Additional analysis 23 Give results of additional analyses, if done (e.g., sensitivity or subgroup analyses, meta-regression [see Item 16]). Page 13-14. Table 1 in the main text; eTable 4-5 and eFigure 1, 2, 5-8 in the Supplementary Materials.

DISCUSSION

√ Summary of evidence 24Summarize the main findings including the strength of evidence for each main outcome; consider their relevance to key groups (e.g., healthcare providers, users, and policy makers).

Page 15. We found a 12% risk reduction of BCa associated with bisphosphonates, which has strengthened the evidence for anticancer effects of bisphosphonates in the BCa primary prevention. However, according to the evidence synthesis, there is a general "indecisive" evidence for this association.

√ Limitations 25Discuss limitations at study and outcome level (e.g., risk of bias), and at review-level (e.g., incomplete retrieval of identified research, reporting bias).

Page 19-20. Detailed limitations of this study are provided in the Discussion section of the main text.

√ Conclusions 26 Provide a general interpretation of the results in the context of other evidence, and implications for future research.

Page 20. Our findings added to the body of evidence for the association between bisphosphonates and a decreased risk of primary breast cancer. However, we can not draw a decisive conclusion based on the current evidence. There is still a long way to go before the use of bisphosphonates as breast cancer chemoprevention strategy in routine practices.

FUNDING

√ Funding 27Describe sources of funding for the systematic review and other support (e.g., supply of data); role of funders for the systematic review.

Page 2. Funding sources are provided on the title page.

From: Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group (2009). Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement. PLoS Med 6(7): e1000097. doi:10.1371/journal.pmed1000097.

25 / 25

Appendix B. Methodological quality score of the cohort and case-control studies

Study name

Score of methodological quality

1(0/4/8)

2(0/4/7)

3(0/4/7)

4(0/10)

5(0/8)

6(0/4)

7(0/4/6)

8(0/2)

9(0/6)

10(0/4/6)

11(0/4)

12(0/4)

13(0/8)

14(0/6)

15(0/4/9)

Selection bias 40

(1-5 items)

Misclassification bias 40

(6-13 items)

Confounding bias 15

(14-15 items)Total Reporting

score (%)*

Cohort studies

Chiang 2012Taiwan, China 0 7 7 10 8 4 6 0 6 6 4 0 8 6 4 32 34 10 76 80

Cardwell 2011United Kingdom 0 7 7 10 8 4 6 2 6 6 4 4 8 6 4 32 40 10 82 86

Vestergaard 2011Denmark 0 7 7 10 8 4 6 2 6 6 0 0 8 6 4 32 32 10 74 78

Chlebowski 2010USA 0 7 7 10 8 4 4 2 6 6 0 0 8 6 9 32 30 15 77 81

Lee 2012 Taiwan, China 0 7 7 10 8 0 6 0 6 6 4 0 8 0 4 32 30 4 66 69

Fournier 2017French 4 7 7 10 8 4 4 2 6 6 4 0 8 6 9 36 34 15 85 89

Kwan 2016USA 0 7 7 10 8 4 6 2 6 6 4 0 8 6 9 32 36 15 83 87

Korde 2018USA 4 7 7 10 8 0 6 2 6 6 4 0 8 6 4 36 32 10 78 82

Case-control studies

Rennert 2010Israel 0 0 7 10 8 4 6 2 6 4 4 0 8 6 9 25 34 15 74 78

Newcomb 2010USA 0 4 7 10 8 4 4 2 6 6 0 0 8 6 4 29 30 10 69 73

Vinogradova 2013 United Kingdom 0 0 7 10 8 4 6 2 6 6 4 0 8 6 4 25 36 10 71 75

Monsees 2011USA 4 4 7 10 8 0 6 2 6 6 4 0 8 6 9 33 32 15 80 84

*The quality score of each study was presented as a percentage of the maximum score, and studies with a score more than 50% were categorized as high-quality studies.

26 / 25

dove medical press web viewbmi, smoking status, alcohol consumption, ethnicity, rheumatoid...

Documents