Cancer Incidence, Mortality, Breast Cancer Detection and Survival Among Canadian Women with Bilateral Cosmetic Breast Augmentation

ÉRIC LAVIGNE

CANCER INCIDENCE, MORTALITY, BREAST CANCER DETECTION AND SURVIVAL AMONG CANADIAN WOMEN WITH BILATERAL COSMETIC BREAST AUGMENTATION

Thèse présentée à la Faculté des études supérieures et postdoctorales de l’Université Laval dans le cadre du programme de doctorat en épidémiologie pour l’obtention du grade de Philosophiae Doctor (Ph.D.)

DÉPARTEMENT MÉDECINE SOCIALE ET PRÉVENTIVE FACULTÉ DE MÉDECINE UNIVERSITÉ LAVAL QUÉBEC

2012

Résumé

Objectifs : La popularité des implants mammaires pour fins esthétiques a rapidement augmentée dans les dernières décennies. Les objectifs de cette thèse étaient d’évaluer l’incidence de cancer, la mortalité, la détection du cancer du sein ainsi que la survie au cancer du sein chez les femmes avec des implants mammaires pour fins esthétiques. Méthodes : Cette thèse porte sur la deuxième phase d’une étude de cohorte rétrospective de 24 558 femmes ayant reçu des implants mammaires pour fins esthétiques dans les provinces de l’Ontario et du Québec entre 1974 et 1989. Ces femmes ont été comparées à un groupe de femmes ayant reçu une autre chirurgie esthétique (n = 15 893) ainsi qu’à des femmes de la population générale. Résultats : Au cours d’un suivi moyen de 24,5 ans, des taux plus élevés de suicide ont été observés chez les femmes avec une augmentation mammaire comparativement aux femmes de la population générale (Rapport Standardisé de Mortalité (RSM) = 2,00, Intervalle de confiance (IC) à 95% = 1,66-2,41) ainsi qu’aux femmes avec une autre chirurgie esthétique (Rapport de taux (RT) de mortalité = 1,43, IC à 95% = 1,02-1,99). Les implants mammaires en position rétro-glandulaire étaient associés à un taux réduit de cancer du sein comparativement aux implants en position rétro-pectorale (RT d’incidence = 0,78, IC à 95% = 0,63-0,96). Un taux 7 fois plus élevé de cancer du sein a été observé (RT d’incidence = 7,36, IC à 95% = 1,86-29,12) dans les cinq premières années après la chirurgie pour les femmes ayant des implants en position rétro-glandulaire avec enveloppe au polyuréthane comparativement aux autres femmes ayant des implants en position rétro- glandulaire sans enveloppe au polyuréthane, mais le RT d’incidence diminuait par la suite (valeur p de tendance = 0,02). Les femmes avec des implants mammaires avaient un risque plus élevé d’avoir un stade avancé au diagnostique du cancer du sein (Rapport de Cote (RC) = 1,51, IC à 95% = 1,18-1,92) ainsi qu’une réduction de la survie spécifique au cancer du sein (RT de mortalité spécifique au cancer du sein = 1.38, IC à 95% = 1.08-1.75). Conclusion : Les implants mammaires pour fins esthétiques sont associés à des risques pour la santé qui doivent être pris en considération. Davantage d’investigations sont nécessaires pour clarifier le risque de cancer associé aux implants au polyuréthane ainsi que par rapport au diagnostique et pronostic du cancer du sein chez ces femmes. iii

Abstract

Objectives: The popularity for cosmetic breast implants has been rapidly increasing in the past decades. The objectives of this thesis were to evaluate cancer incidence, mortality, breast cancer detection and survival following breast cancer diagnosis among women with cosmetic breast implants. Methods: This thesis is the second phase (follow-up) of a large Canadian retrospective cohort study of 24,558 women who received cosmetic breast implants between 1974 and 1989 in the provinces of Ontario and Quebec. These women were compared to other cosmetic surgery women (n = 15,893) and the general female population. Results: Over an average 24.5 years of follow-up, augmented women have been shown to have elevated rates of suicide relative to women in the general population (Standardized Mortality Ratio (SMR) = 2.00, 95% CI = 1.66-2.41) and compared to women seeking other cosmetic surgery (mortality Rate Ratio (RR) = 1.43, 95% CI = 1.02-1.99). Subglandular implants were associated to a reduced rate of breast cancer compared to submuscular implants (Incidence Rate Ratio (IRR) = 0.78, 95% CI = 0.63-0.96). We observed a 7-fold increased rate (IRR = 7.36, 95% CI = 1.86-29.12) of breast cancer in the first five years after the date of surgery for polyurethane-coated subglandular implant women compared to other women with subglandular implants without polyurethane coating, but this IRR decreased progressively over time (p value for trend = 0.02). Women who received cosmetic breast implants were also more likely to have an advanced stage at breast cancer diagnosis (Odds ratio = 1.51, 95 % CI = 1.18-1.92) and to have poorer breast cancer- specific survival (Overall breast cancer-specific mortality hazard ratio: 1.38, 95 % CI = 1.08-1.75). Conclusion: Cosmetic breast implants are associated with health risks that need to be acknowledged. Further investigations are needed for clarifying the cancer risk associated with polyurethane implants and for pursuing investigations on breast cancer diagnosis and prognosis among these women.

Avant-propos

Cette thèse de doctorat en épidémiologie s’inscrit à l’intérieur d’un grand projet de recherche portant sur les effets à long terme des implants mammaires sur la santé mis sur pied par le Dr. Jacques Brisson en collaboration avec le Dr. Eric J. Holowaty et le Dr. Yang Mao. Il s'agit d'un projet ayant été subventionné par l’Agence de la santé publique du Canada. De plus, un support financier m’a été accordé par l’Unité de recherche en santé des populations (URESP), Action cancer Ontario et l’Agence de la santé publique du Canada afin que je puisse poursuivre mes études doctorales. Cette thèse de doctorat représente la deuxième phase (suivi) d’une étude de cohorte rétrospective portant sur des femmes ayant reçus des implants mammaires pour fins esthétiques entre les années 1974 et 1989 dans les provinces de l’Ontario et de Québec. La première phase portait sur l’évaluation de l’incidence du cancer, de la mortalité, de la détection du cancer du sein ainsi que la survie au cancer du sein chez ces femmes. Cette deuxième phase porte sur un suivi additionnel de 10 ans.

J’ai eu un rôle de premier plan dans les étapes suivantes pour cette deuxième phase de ce projet de recherche : développement du protocole de recherche incluant la revue de la littérature, le développement des questions de recherche et le plan méthodologique, la collecte des données, l’analyse des données, l’interprétation des résultats ainsi que la rédaction des articles scientifiques. D’ailleurs, quatre articles sont issus de cette thèse de doctorat. Ces articles sont les suivants :

1. Pan SY, Lavigne E, Holowaty EJ, Villeneuve PJ, Xie L, Morrison H, et al. Canadian breast implant cohort: Extended follow-up of cancer incidence. Int J Cancer 2012 Oct 1;131(7):E1148-57. 2. Holowaty EJ, Lavigne E, Pan SY, Villeneuve PJ, Xie L, Morrison H, Brisson J. Cosmetic breast augmentation and mortality: an update of a Canadian cohort. 2012. (Cet article sera soumis pour publication). 3. Lavigne E, Holowaty EJ, Pan SY, Xie L, Villeneuve PJ, Morrison H, et al. Do breast implants adversely affect prognosis among those subsequently diagnosed with breast cancer? Findings from an extended follow-up of a Canadian cohort. Cancer Epidemiol Biomarkers Prev 2012 Oct;21(10):1868-76. 4. Lavigne E, Holowaty EJ, Pan SY, Villeneuve PJ, Morrison H, Brisson J. Breast cancer detection and survival among women with cosmetic breast implants: a v systematic review and meta-analysis of observational studies. 2012. (Cet article sera soumis pour publicaiton au British Medical Journal).

Aknowledgements

I would first like to thank my research director, Dr. Jacques Brisson, for his support, his advices, his availability and his help throughout this scientific journey. He has been a mentor for me by communicating his knowledge and experience for which I will be forever grateful. Finally, I would also like to thank him for providing me the skills and knowledge to accomplish myself either professionally than personally.

This project was made possible because of the dedicated work of several collaborators. I would like to thank Drs Yang Mao and Eric J. Holowaty for their devoted contribution to this project and for their advices and help throughout my studies. I would also like to thank Drs Sai Yi Pan, Howard Morrison, Paul J. Villeneuve, Mrs. Lin Xie and Drs Ken C. Johnson and Dean Fergusson for their scientific contribution and for their collaboration in the writing of the different manuscripts included in this thesis. I am also thankful to Sylvie Bérubé and Caty Blanchette from Quebec and Gemma Lee and Susitha Wanigaratne from Ontario, for their help in the design and conduct of this study.

My sincere thanks go also to all the people who have supported me throughout these years. First, I would like to thank all my friends for their great support and encouraging comments. Thanks for your friendship and for being there for me. I am also grateful for the unconditional love that my parents, Murielle and Michel, have giving me throughout my life. Thank you for supporting me every day and for making me who I am today. I give thanks with all my heart to family members who have been there for me in the good and not so good times. Thanks to Françoise and Michel, my parents-in-law, for their supporting comments and for being in my life. Finally, I would like to thank my beloved Myriam for her unconditional support, her useful advices, her love and for making me a better person every day. I love you with all my heart.

vii

Table of contents Résumé...... ii Abstract ...... iii Avant-propos...... iv Aknowledgements...... vi Table of contents ...... vii List of tables...... ix List of figures ...... xi

Chapter 1: General introduction ...... 12 Introduction ...... 13 State of knowledge...... 17 Objectives...... 27 Material and Methods ...... 28

Chapter 2: Canadian Breast Implant Cohort: extended follow-up of cancer incidence ...... 51 Résumé...... 52 Abstract ...... 53 Introduction ...... 54 Material and methods...... 56 Results ...... 60 Discussion ...... 63 Acknowledgements ...... 68 References ...... 69

Chapter 3: Cosmetic breast augmentation and mortality: an update of a Canadian cohort...... 79 Résumé...... 80 Abstract ...... 81 Introduction ...... 82 Methods...... 83 Results ...... 86 Discussion ...... 87 References ...... 94

Chapter 4: Do breast implants adversely affect prognosis among those subsequently diagnosed with breast cancer? Findings from an extended follow-up of a Canadian cohort...... 102 Résumé...... 103 Abstract ...... 104 Introduction ...... 105 Methods...... 107 Results ...... 110 Discussion ...... 112 viii

Acknowledgements ...... 116 References ...... 117

Chapter 5: Breast cancer detection and survival among women with cosmetic breast implants: a systematic review and meta-analysis of observational studies ...... 128 Résumé...... 129 Abstract ...... 131 Introduction ...... 132 Materials and methods ...... 134 Results ...... 138 Discussion ...... 140 References ...... 145 Appendix ...... 159

Conclusion ...... 160

References ...... 164 ix

List of tables Chapter 1

Table 1. Least Significant Relative Risk (LSRR) comparing cancer incidence (overall and site-specific) between cosmetic breast implant women and the female general population applying the methods proposed by Walter………………………...... 42 Table 2. Least Significant Relative Risk (LSRR) for mortality (overall and cause specific) between cosmetic breast implants and female general population applying the methods proposed by Walter………………………………………………………………………...43 Table 3. Least Significant Relative Risk (LSRR) for cancer incidence (overall and site- specific) among breast implant women vs. other cosmetic surgery women applying the methods proposed by Signorini for Poisson regression models...... 44 Table 4. Least Significant Relative Risk (LSRR) for mortality (overall and cause specific) among breast implant women vs. other cosmetic surgery women applying the methods proposed by Signorini for Poisson regression models...... 45 Table 5. Least Significant Relative Risk (LSRR) for breast cancer incidence among cosmetic breast implant women according to specific implant characteristics applying the methods proposed by Signorini for Poisson regression models...... 46 Table 6. Least significant odds ratio (LSOR) for stage of breast cancer at diagnosis among cosmetic breast implant women vs. other cosmetic surgery women applying the methods proposed by Hsieh for logistic regression models...... 47 Table 7. Least significant odds ratio (LSOR) for stage of breast cancer at diagnosis among cosmetic breast implant women according to specific implant characteristics applying the methods proposed by Hsieh for logistic regression models...... 48 Table 8. Least significant hazard ratio (LSHR) to investigate whether there is a differential breast cancer-specific survival between the implant women and other plastic surgery women applying the methods proposed by Hsieh and Lavori and Shoenfeld for Cox Proportional Hazards regression models...... 49 Table 9. Least significant hazard ratio (LSHR) to investigate whether there is a differential breast cancer-specific survival among cosmetic breast implant women according to specific implant characteristics when applying the methods proposed by Hsieh and Lavori and Shoenfeld for Cox Proportional Hazards regression models...... 50

Chapter 2

Table 1. Frequency distribution for selected characteristics of women who received breast implants and women who received other cosmetic surgeries, Canadian Breast Implant Cohort Study……………………………………………………………………………….73 Table 2. Standardized incidence ratios (SIRs) for selected cancers based on general population cancer incidence rates (1974-2007) among breast implant and other cosmetic surgery women with comparisons with previous follow-up……………………………….74 Table 3. Incidence rate ratios (IRRs) for selected cancers between breast implant and other cosmetic surgery women with comparisons with previous analysis……………………….75 Table 4. Incidence rate ratios (IRRs) and 95% confidence intervals (CIs) of breast cancer for selected breast implant characteristics among implant women………………………...76 x

Chapter 3

Table 1. Frequency distribution for selected characteristics of women who received breast implants and women who received other cosmetic surgeries, Canadian Breast Implant Cohort Study……………………………………………………………………………….98 Table 2. Standardized mortality ratios (SMRs) for selected causes of death based on general female population mortality rates (1974-2007) among breast implant and other cosmetic surgery women, compared with previous analysis………………………………99 Table 3. Mortality rate ratios (RRs) for selected causes of death between breast implant and other cosmetic surgery women with comparisons with previous analysis……………….100 Table 4. Standardized mortality ratios (SMRs) for suicide death based on general female population mortality rates (1974-2007) among breast implant and other cosmetic surgery women, and incidence rate ratios (RRs) of suicide death for breast implant vs. other cosmetic surgery women by selected surgery characteristics…………………………….101

Chapter 4

Table 1. Frequency distribution for selected characteristics of women diagnosed with breast cancer who received breast implants and other cosmetic surgeries, Canadian Breast Implant Cohort Study……………………………………………………………………..123 Table 2. Odds ratios (ORs) and 95% confidence intervals (CIs) for selected characteristics of incident cases of breast cancer comparing breast implant women to other cosmetic surgery women……………………………………………………………………………125 Table 3. Odds ratios (ORs) and 95% confidence intervals (CIs) for stage distribution of breast cancer for selected breast implant characteristics………………………………….126

Chapter 5

Table 1. Characteristics of 12 studies selected for quantitative analysis to evaluate the association between cosmetic breast implants and stage distribution of breast cancer...... 151 Table 2. Random effects overall odds ratio for the association between cosmetic breast implants and stage distribution of breast cancer stratified by variables potentially related to study quality………………………………………………………………………………154 Table 3. Characteristics of 5 studies selected for quantitative analysis to evaluate the association between cosmetic breast implants and breast cancer-specific survival following breast cancer diagnosis...... 155

List of figures Chapter 2

Figure 1. Cumulative breast cancer incidence curves for time since index surgery comparing breast implant with other cosmetic surgery women……………………………77 Figure 2. Incidence rate ratios (IRRs) and 95% confidence intervals (CIs) to evaluate the trend in breast cancer risk for women who received subglandular polyurethane coated breast implants relative to other women who received subglandular implants, by time since surgery……………………………………………………………………………………...78

Chapter 4

Figure 1. Breast cancer-specific survival curves comparing breast implant with other cosmetic surgery women………………………………………………………………….126

Chapter 5

Figure 1. Flowchart of the meta-analysis search strategy and process of selecting scientific articles on the association between cosmetic breast implants and stage distribution of breast cancer and the association between cosmetic breast implants and survival following breast cancer diagnosis...... 156 Figure 2. Forest plot with study-specific and random effects overall odds ratio (OR) for the association between cosmetic breast implants and stage distribution of breast cancer...... 157 Figure 3. Forest plot with study-specific and random effects overall hazard ratio (HR) for the association between cosmetic breast implants and breast cancer-specific survival...... 158

Chapter 1: General introduction

Introduction

The demand for cosmetic breast implants among women has been rapidly increasing since the first mammary prosthesis was inserted in a woman’s breasts in the early 1960’s (1). For instance, in the US, cosmetic breast augmentation was the most commonly performed surgical cosmetic procedure in 2010 with 296,000 surgeries performed (2), an increase of approximately 800 % compared with the early 90’s and an increase of about 50 % compared with the year 2000 (3). It is estimated that 100,000 to 200,000 women have breast implants in Canada (4). Although very popular, there remains considerable controversy about the long-term health effects of breast implants.

Cosmetic breast augmentation has been the subject of numerous investigations in the past two decades. Concerns have been raised concerning the possible carcinogenic effect of cosmetic breast implants, specifically for breast cancer (5). Most of the epidemiological studies published have shown no carcinogenic effect of breast augmentation on breast cancer (6-32) and even showing for some a lower than expected incidence of breast cancer (9;14;16;22). In recent years, the concern of the carcinogenic effect of breast implants has focused on the fact that breast cancer development might be affected by specific implant characteristics: type of implant (saline and silicone-gel-filled implants (SGFIs)), implant placement (submuscular and subglandular), implant fill volume and implant envelope (polyurethane coated or not). For the latter characteristic, it has been reported that polyurethane could degrade into significant quantities of 2,4- and 2,6-toluene-diisocyanate diamines (TDA’s) which has been shown to cause cancer in laboratory animals and is recognized by the International Agency for Research on Cancer as an animal carcinogen (33;34).

However, very few studies to date have been able to investigate breast cancer incidence by implant characteristics. For the studies that investigated this issue, results have been inconclusive due either to their small number of cancer incident cases and insufficient amount of follow-up time (9;12;14;18;28;35). Extended follow-up time is particularly relevant given that immunologic changes and deterioration of the implant capsule may only occur over an extended period of time (36;37). Moreover, the existing literature suggests 14 that the expected latency period between first exposure to carcinogenic agents and solid tumor development is lengthy (38). In this thesis, we aim to evaluate if breast cancer risk is affected by specific implant characteristics, especially for implants with polyurethane coated envelope.

Furthermore, some studies have evaluated the relationship between breast implants and the incidence of cancers at sites other than the breast (8;9;16;18;22;26;27;29;30;39). The findings from these studies have been largely negative, though excesses have been inconsistently reported for some types of cancers including the brain (25), lung (22;25;26), vulva (16) and cervix (39). Although these excesses appear largely attributable to life style characteristics rather than the presence of the implant itself, taken as a whole, the epidemiological evidence for risk of cancer at body sites other than the breast needs to be further clarified. Moreover, a recent report by the FDA as well as review articles recommended that the risk of hematopoietic malignancies among women who have cosmetic breast implants be further investigated (39-41). One of the aims of this thesis is to examine non-breast cancer risk among breast implant women.

In addition to the question of cancer risks, some investigations reported lower breast cancer and overall mortality rates among implant women (15;42-45) while others have shown increased risks of overall mortality (46-48) compared to women in the general population or to women with other cosmetic surgeries. Furthermore, a number of studies have shown higher deaths from suicide relative to the general female population (42-49). Some studies have also suggested that the excess in suicide risk may change with length of time since surgery (42;46) and age at which surgery was performed (42;43;46). A number of studies have also reported higher deaths related to respiratory diseases (46-48), lung cancer deaths (46;48) and deaths from motor vehicle accidents (42) and several other types of injuries (44;46-48) among cosmetic breast implant women relative to the general female population. However, these reports have been severely limited by a small number of deaths for these causes which makes it difficult to draw any solid conclusions. In this thesis, we will assess mortality experience, specifically deaths from suicide, among cosmetic breast implant women. 15

Another concern is that breast augmentation may be associated with an advanced stage of breast cancer at diagnosis. It’s been reported that cosmetic breast implants, especially implants placed under the breast glands (subglandular placement), are radiopaque and may obscure the visualization of breast tissue with mammography which may delay the detection of breast cancer and affect prognosis of the disease (50-53). While a few studies have performed such evaluations, their findings have been mixed with some reporting that women with breast augmentation may be more likely to be diagnosed with advanced cancers (51;54-56) while others have reported no such difference (7;8;12;14;17;18;20;30;32;57-65). These conflicting results may be explained by the small number of incident breast cancer cases in these studies, especially among women with breast implants, which could have limited statistical power to clearly evaluate this question. In particular, across studies, the number of incident breast cancer cases among implant subjects varies from 7 to 182 (7;8;12;14;17;18;20;30;32;51;54-65).

In addition to the concern of possible advanced stage of breast cancers among augmented women, breast cancer related survival has been of interest (56;58-61;64). If diagnosis of breast cancer is associated with more advanced tumors among augmented women, this could translate into poorer survival. To date, all published studies reported no statistically significant differences in breast cancer-specific survival when comparing augmented women with breast cancer to non-augmented women with breast cancer (56;58-61;64). However, the small numbers of incident breast cancer cases and insufficient follow-up time after diagnosis in these studies may have limited the statistical power necessary to detect a difference in survival. Additionally, no study has evaluated breast cancer survival according to implant characteristics. In this thesis, we will examine whether cosmetic breast implants can impair the ability to identify breast cancer at an early stage and affect prognosis of the disease. We will also examine whether the stage distribution and prognostic factors of breast cancer among women with implants are affected by implant characteristics and examine survival rate patterns among breast implant women who develop breast cancer.

This thesis is the second phase (follow-up) of a large Canadian retrospective cohort study evaluating long term health effects of women who received cosmetic breast implants between 1974 and 1989 in the provinces of Ontario and Quebec. This project is conducted 16 by Laval University’s Population Health Research Unit (URESP), Cancer Care Ontario and the Public Health Agency of Canada for which three scientific papers were previously published based on the results of the first phase (9;43;56). This first phase identified incident cancers and deaths through linkage to Canadian registries up to December 31, 1997. The second phase of this project is based on an additional 10 years of follow-up for cancer incidence, mortality and survival analyses by identifying incident cancers and deaths through linkage to provincial registries (Quebec and Ontario). With the largest cohort to date evaluating long term health effects of cosmetic breast implants, the additional follow- up time is expected to substantially increase statistical power as more cancer cases and deaths will be identified.

I have been involved in all stages of the second phase of this cohort study, namely the literature review, development of the research questions, the writing of the research protocol, the data collection, the data analyses and the writing of the manuscripts.

State of knowledge

Cosmetic breast implants

Breast implants have been a popular type of cosmetic surgery since the early 1960s (66). Cosmetic breast implants were first used in the United States in 1962 (67), and in Canada in 1969 (66). Also known as augmentation mammaplasty, breast implant surgery is an elective surgical procedure to enhance the size of a woman’s breast by insertion of a mammary prosthesis behind each breast (68). They are the most common cosmetic procedure in the Unites-States (69;70). They are principally used for cosmetic purposes among women who have dissatisfaction with breast size and/or shape (71), but also used for breast reconstruction in the case of a mastectomy. There are mainly two types of cosmetic breast implants; silicone gel-filled implants (SGFIs) and saline-filled implants. SGFIs are composed of a silicone elastomer shell of varying thickness that encases different volumes of silicone gel and oil. Saline-filled implants consist of a single lumen enclosed by a silicone elastomer envelope. These implants are inflated with sterile saline via a self-sealing valve through a closed system into the recipient (68;72).

Implants vary in size (implant fill volume), shape, and external surfaces (i.e. smooth, textured, polyurethane foam-covered), and can be inserted either superficial to the muscles, on top of the pectoralis muscle and directly under the breast glands (subglandular placement), covering the chest wall or under the pectoralis major muscle (submuscular placement) (68;73). Implants placed under the breast glands (subglandular placement) are considered to be less painful post- surgery as only the skin and fat are cut, with a quicker recovery (74). Subglandular implant placement is considered in the case of a breast with adequate body fat to minimize the palpability of the implant, a breast with good skin elasticity and when the elimination of any pectoralis muscle incision is requested (68). Implants placed under the pectoralis major muscle (submuscular placement) are considered to be more painful post surgery as the skin, fat and muscle are cut and the muscle is stretched, with a slightly longer recovery (75). Although the recovery is longer, submuscular implant placement has certain benefits; implants are less palpable, there is a possibility of less capsular contracture and easier imaging of the breast 18 tissue with mammography (68). The site of implantation, which is usually decided by the plastic surgeon in collaboration with the recipient, depends on the body type, the thickness of the breast tissue and its ability to adequately cover the breast implant and the patient’s preferences (68).

Implants covered with polyurethane foam became popular in 1981 (66) since they reduced the incidence of capsular contracture so frequently encountered with smooth or textured SGFIs (52;76-79). However, polyurethane foam-covered (PU) breast implants were withdrawn from the market in April, 1991 when the United States FDA found that polyurethane could degrade into significant quantities of 2,4- and 2,6-toluene-diisocyanate diamines (TDAs) (33) which has been shown to cause cancer in laboratory animals and is recognized by the International Agency for Research on Cancer as an animal carcinogen (33;34).

Since their first use in the 1960’s, cosmetic breast implants, particularly silicone gel-filled implants (SGFIs), have been subject to controversies. Concern over silicone as a potential carcinogen largely began when the results of a 2-year toxicity and carcinogenicity study by Dow Corning Corporation (1987) of medical-grade silicone gel found that 23% of rats developed sarcomas at the implant site, most of which were fibrosarcomas (80). On January 6, 1992, Health Canada asked manufacturers to stop the sale of silicone gel-filled implants (SGFIs) in Canada until further studies could be done (81). A similar decision was made in the United States, and silicone gel-filled breast implants were removed from general sale (81). However, because of the lack of studies showing carcinogenic effect of silicone, between the year 2000 and October 2006, silicone implants had been available again in Canada through special approval by Health Canada (82). Finally, on October 20th 2006, Health Canada reapproved silicone gel-filled breast implants for general cosmetic use in Canada (83).

Characteristics of women with cosmetic breast implants

Certain studies have been able to characterize the reasons why women have cosmetic breast surgery. Motivations for cosmetic breast surgery include improving self-esteem due to dissatisfaction with body image and with breast size and/or shape and increasing sexual attractiveness and responsiveness (71;84;85). Studies have also shown that between 3% and 19

15% of patients undergoing cosmetic surgery may also suffer from excessive dissatisfaction with body image consistent with the psychiatric diagnosis of body dysmorphic disorder (86) which would also be a reason to have breast augmentation (71).

Literature has also been able to describe inherent characteristics of women having cosmetic breast implants. It is suggested that women with implants are more likely to have had more sexual partners, be middle to upper middle socioeconomic status, be younger at their first pregnancy, use oral contraceptives, have a history of terminated pregnancies, be more frequent users of alcohol and tobacco, have higher divorce rate and be below average body weight (68;84).

In the epidemiological studies assessing the health effects of breast implants, women with implants were compared either to general population estimates or to a control group consisting of women who received other elective cosmetic surgery. These other surgeries include chemical peel or dermabrasion, coronal brow lift (eyebrow and forehead lift), otoplasty (ear surgery), rhinoplasty (nose surgery), rhytidectomy (face lift) or blepharoplasty (eyelid surgery). Studies using other cosmetic surgery women as a control group have supported the notion that these patients are a more appropriate comparison group than women in the general population to study the health effects associated with breast implants because women in the control group tend to be more similar in terms of sociodemographic and lifestyle factors than women in the general population (87;88). Understanding this, it is then important to specify when reporting statistically significant results of health effects associated with cosmetic breast implants the comparison group used because there could be discrepancy between the ones using general population estimates and the ones using a more comparable comparison group.

Cancer incidence among augmented women

Understanding that cosmetic breast augmentation has been subject to controversies regarding their possible carcinogenic effect (72;89-93), numerous investigations in the past two decades have been conducted. No epidemiological study to date was able to establish that cosmetic breast implants are associated with increased breast cancer incidence (6-32). However, some studies have found statistically significant lower than expected breast 20 cancer risk among women with breast implants when compared with general population estimates (9;14;16;22) or women having other cosmetic surgeries (9). Possible biological mechanisms have been suggested in the literature to explain this finding. It seems that the presence of breast implants could enhance the immune system, whereby carcinogens and transformed cells would be more easily destroyed (14). The weight and volume of breast implants is suggested to compress the glandular tissue resulting in a decreased blood supply that may reduce the rate of cell proliferation (14;94). It is also important to mention that the decreased risk of breast cancer incidence among augmented women reported in the literature could also be due to the fact that these women have smaller breasts prior to augmentation which could make them less likely to develop breast cancer.

Nonetheless, in recent years, breast cancer incidence among augmented women has continued to be studied in an attempt to clearly evaluate variations of breast cancer risk, if any exist, according to specific implant characteristics. Common implant characteristics include the type of implant (saline and silicone-gel-filled implants (SGFIs)), the placement of the implant (submuscular and subglandular), the implant fill volume and the implant envelope (polyurethane coated or not). Polyurethane foam-covered breast implants is of particular interest here because of their potential carcinogenic effect (33;91-93).

However, most of the epidemiological investigations reporting on breast implants and breast cancer risk have not been able to conduct subgroup analyses due principally to their inability to collect specific implant characteristics. For the studies that did report subgroup analyses, results have been deemed inconclusive due to important limitations such as their relatively small sample sizes and their limited number of breast cancer cases which lacked the statistical power needed to conduct subgroup analyses by specific implant characteristics (9;12;14;18;28;35). Another identified shortcoming of past studies include an insufficient amount of follow-up time as most studies followed women for less than 30 years and only a few had a follow-up time over 30 years (18;22;26). Considering that immunologic changes and deterioration of the implant capsule may only occur over an extended period of time (36;37), and that the expected latency period between first 21 exposure to carcinogenic agents and solid tumor development is lengthy, it is of great importance to have an extended follow-up period (9;38).

To our knowledge, only one scientific article, Brisson et al., which was one of the scientific papers published from the first phase of this project, was able to report a statistically significant 2 fold increase of breast cancer risk among women whose implants were inserted subglandularly with a polyurethane envelope compared to women with the same site of implantation without such envelope (9). However, no other statistically significant associations were reported for breast cancer risk according to implant characteristics in the first phase of this project. Moreover, to date, there is no other study that has been able to report convincing subgroup analyses due to the reasons mentioned above to compare breast cancer risk between saline implants and SGFI’s, subglandular and submuscular implant placement, between different fill volumes of implants and between polyurethane coated implants and non-coated implants. It is then still uncertain whether there are variations in breast cancer risk according to specific implant characteristics due to the lack of scientific evidence. Given that a certain number of women received breast implants at a young age, continued follow-up is important to examine associations between breast implants and the incidence of breast cancer up to postmenopausal years when breast cancer incidence is greater. In this thesis, we propose to examine with a large cohort adding 10 more years of follow-up for cancer incidence whether cancer risk among cosmetic breast implant women is affected by the following implant characteristics: implant type (saline and silicone-gel- filled implants (SGFIs)), implant placement (submuscular and subglandular), implant envelope (polyurethane coated or not) and fill volume.

A certain number of epidemiological investigations have been able to evaluate the relationship between breast implants and the incidence of non-breast cancers (8;9;16;18;22;26;27;29;30;39). The findings from these studies have been largely negative when women with implants are compared to other women that had cosmetic surgery (8;9;18) or to general population estimates (9;16;18;30). However, statistically significant increased risk for some types of cancer including the brain (25), lung (22;25;26), vulva (16) and cervix (39) have been inconsistently reported among women with cosmetic breast 22 implants when compared with general populations estimates. Nonetheless, these increases in risk, appear largely attributable to life style characteristics rather than the implant itself (39). For instance, studies reporting these results have not used women with other cosmetic surgeries as control group which would have been more appropriate according to the rationale given above. In one of our previous papers, results showed that women with implants were not at an increased risk of these types of cancers compared to a more comparable control group (9). Moreover, a recent report by the FDA as well as review articles recommended that the risk of hematopoietic malignancies, especially for anaplastic large T cell lymphomas, among women who have cosmetic breast implants be further investigated (39-41). Taken as a whole, the epidemiological evidence for risk of cancer at body sites other than the breast needs to be further clarified (39). One of the aims of this thesis is to clarify the question of non-breast cancer risk among breast implant women using as recommended a comparison group of women with other cosmetic surgeries.

Mortality and women with cosmetic breast implants

Assessment of mortality experience among cosmetic breast implant women has been investigated in past years. Conflicting results have been reported for overall mortality because some studies reported a statistically significant reduction in overall mortality compared with general population estimates (42;43;45), others reported statistically significant increased risk of overall mortality compared to general population estimates (46-48) and one study showed no difference (44). However, among studies that used a control group that consisted of women who received other plastic surgery, no difference of overall mortality risk has been reported (42;43;45-47).

Breast cancer mortality has also been studied when we consider that the implants might induce carcinogenic activity in the breast and result in death. Studies have consistently reported that women with cosmetic breast implants are not at an increased risk of breast cancer deaths compared to women in the general population or women with other cosmetic surgeries (42-48).

Another cause of death that has been of concern among women with implants is death related to respiratory diseases such as lung cancer, chronic obstructive pulmonary diseases or pneumonia. In fact, three studies have shown that women with implants have a higher statistically significant risk of death from respiratory diseases than women in the general population (46-48) while two others have shown a trend towards higher risk of death from respiratory diseases without being significant (42;43). Here again, studies that have used a more comparable control group in parallel or not to the comparisons with general population estimates showed no difference of risk of deaths by respiratory diseases (42;43;45-47).

Psychological considerations concerning women who have cosmetic breast implants have been of interest when several cohort studies reported 2 to 3-fold higher suicide rates among women with implants compared with general population estimates (42-49). Despite the consistency of these findings, some studies have been severely limited by small sample sizes and the fact that they did not use a more comparable comparison group. For example, the small sample sizes make it difficult to characterize suicide risk by age group and time since surgery was performed and the comparison with general population estimates is affected by residual confounding due to the fact that women with implants are not similar in terms of sociodemographic and lifestyle factors to women in the general population. However, no studies using a comparison group of women who have had other cosmetic surgeries reported this association (42;43;45;47).

As mentioned earlier, it is suggested that women with implants may suffer from excessive dissatisfaction consistent with the psychiatric diagnosis of body dysmorphic disorder (95;96). Understanding this, it would not be the implant itself that would cause suicide, but rather psychological disorders prior to breast augmentation. It is then recommended that women interested in having breast augmentation undergo mental health evaluations before getting the surgery. In this thesis, we aim at clarifying mortality patterns among women with cosmetic breast implants. 24

Detection of breast cancer and survival among augmented women

There has been some concern that cosmetic breast implants may delay the detection of breast cancer and impair the ability to identify breast cancer at an early stage (50-53). The diagnosis of more advanced cancers can lead to poorer prognosis and consequently poorer survival post-breast cancer diagnosis.

It is reported that cosmetic breast implants are radiopaque which may obscure the visualization of breast tissue with mammography and result in a delayed detection of breast cancer (50-53). In fact, the amount of parenchymal breast tissue obscured at mammography by the implant is known to be between 22 and 83 % (97). Consequently, breast implants will hinder visualization of breast tissue and affect mammography detection of breast cancers (31). Insufficient compression of the breast to visualize the parenchyma and the production of implant-related artifacts on the film can also make it difficult to interpret mammographic exams in women with augmented breasts (52;98). There has been a few reports that showed that the presence of breast implants increases the false-negative rate (poorer sensitivity) of mammograms compared with non-augmented breasts (32;51;54;64), but does not increase the false-positive rate (poorer specificity) (62). While specialized radiographic techniques have been developed for women with breast implants to improve visualization which involve displacing the implant posteriorly against the chest wall and pulling breast tissue over and in front of the implant (53;99-101), there is still one-third of the breast that is not adequately visualized leading to an increase of false-negative mammograms (52).

Most of the studies that evaluated the detection of breast cancer among women with cosmetic breast implants have compared the stage distribution of breast cancer at diagnosis to a control group that consisted of women with other types of cosmetic surgery with breast cancer (8;17;18;56) or to other non-augmented women with breast cancer (7;12;14;20;30;32;51;54;55;57-65). Results have been mixed with some reporting that women with breast augmentation may have up to a 3-fold increased risk to be diagnosed 25 with advanced cancers compared with non-augmented women (51;54-56) while others have reported no such difference (7;8;12;14;17;18;20;30;32;57-65). These conflicting results can be explained by the small number of incident breast cancer cases in these studies which limit statistical power to clearly evaluate this association. It is of great importance to pursue this evaluation in order to give clear information to women who are interested in breast augmentation and who are at a high risk of developing breast cancer. In this thesis, we will examine whether breast implants impair the ability to identify breast cancer at an early stage.

Furthermore, it has been suggested that specific breast implant characteristics might affect the detection of breast cancer (102). Specifically, implants placed under the breast glands (subglandular placement), because of their proximity with breast tissue, are suspected to obstruct mammographic visualization of the breast more so than submuscular placement (50;103). However, to date, no study has been able to report increased stage at diagnosis of breast cancer among women with subglandular implants compared with women whose implants are placed submuscularly. This can also be due to the small number of breast cancer incident cases which limits the capacity to perform subgroup analyses. In this thesis, we will examine whether the stage distribution of breast cancer among women with implants is affected by implant characteristics such as the placement of the implant.

In addition to the concern of advanced stage at diagnosis of breast cancer among augmented women, breast cancer related survival subsequent to the diagnosis of breast cancer has been investigated in the past years (56;58-61;64). The idea is that the possible advanced stage of breast cancer in the augmented women could translate into poorer survival. However, all studies published to date reported no statistically significant differences in breast cancer survival when comparing augmented women with breast cancer to non-augmented women with breast cancer (56;58-61;64). The small numbers of incident breast cancer cases and insufficient amount of follow-up time in these studies are suspected to limit statistical power to clearly compare survival rate patterns among these two groups of women. To provide further insight about length of survival post-breast cancer diagnosis, 26 we will compare survival rates of augmented women with non-augmented women post- breast cancer diagnosis.

Survival rate patterns following breast cancer diagnosis according to breast implant characteristics are also of interest if any difference exists in terms of breast cancer stage at diagnosis among augmented women according to these characteristics. To our knowledge, no study to date, even the first phase of this project which is probably due to limited statistical power for these analyses, has evaluated this association. In this thesis, we will compare survival rates following breast cancer diagnosis among women with cosmetic breast implants by specific implant characteristics (e.g. survival rate post breast cancer diagnosis of submuscular vs. subglandular placement).

Objectives

In order to investigate the above research gaps, this thesis seeks to fill critical voids in the current literature as several hypotheses regarding outcomes among cosmetic breast implant women have yet to be adequately examined. Specifically, the objectives of the proposed investigation are as follows:

1. To evaluate the association between cancer incidence, overall and site-specific, and cosmetic breast implants. 2. To examine whether breast cancer risk among cosmetic breast implant women is associated to the following implant characteristics: implant type (saline and silicone-gel-filled implants (SGFIs)), implant placement (submuscular and subglandular), implant envelope (polyurethane coated or not) and fill volume. 3. To assess the association of mortality, specifically deaths from suicide and other causes, with cosmetic breast implants. 4. To examine whether cosmetic breast implants are associated with more advanced stage at diagnosis of breast cancer. 5. To examine whether characteristics of cosmetic breast implants are associated with stage at breast cancer diagnosis focusing on implant type (saline and silicone-gel- filled implants (SGFIs)), implant placement (submuscular and subglandular), implant envelope (polyurethane coated or not) and fill volume. 6. To examine whether cosmetic breast implants are associated with poorer survival rates following breast cancer diagnosis. 7. To examine whether characteristics of cosmetic breast implants are associated with poorer survival rates following breast cancer diagnosis focusing on the effect of implant type (saline and silicone-gel-filled implants (SGFIs)), implant placement (submuscular and subglandular), implant envelope (polyurethane coated or not) and fill volume.

Material and Methods

Study design, study population and selection criteria

As mentioned earlier, this thesis is based on an extension of the largest retrospective cohort study carried-out to date to evaluate long term health effects among women with cosmetic breast implants (9;43;56).

The study population consists of women, 18 years of age or older, who were residents of the province of Ontario or Quebec in Canada and who underwent bilateral cosmetic breast augmentation in their province of residence between January 1, 1974, and December 31, 1989. A control cohort (comparison group) was assembled consisting of women who received other common elective cosmetic surgeries. These women received other cosmetic surgeries, not billable to the publicly funded health insurance plans of Ontario or Quebec, which included the following: chemical peel or dermabrasion, coronal brow lift (eyebrow and forehead lift), otoplasty (ear surgery), rhinoplasty (nose surgery), rhytidectomy (face- lift), or blepharoplasty (eyelid surgery). Controls were frequency matched to the breast implant recipients by year of entry into the cohort, province of residence and by surgeon. The matching ratio would be called a two-to-one (2:1) implant to other plastic surgery patient, but in practice, there was a slight departure from this ratio. This is due to the fact that recruitment was done based on a priori expected case numbers, and some women who received breast implants did not fulfill the requisite inclusion criteria.

Eligible subjects in Ontario were identified through plastic surgeons that performed cosmetic breast implant surgeries in this province between January 1, 1974, and December 31, 1989. Investigations revealed that plastic surgeons that performed cosmetic breast implant surgeries also performed most of the other cosmetic surgeries. In total, 133 plastic surgeons ever performed augmentation mammaplasty during the study period where nearly 75 % of these procedures were performed by 24 plastic surgeons.

Breast implant recipients and controls in the province of Quebec were identified from MED-ECHO files, a computerized system that documents all hospital discharges and day surgeries that occur in Quebec and from records of plastic surgeons in private plastic surgery clinics for the period between 1974 and 1989. The implant and control cohorts were recruited from among women operated on by approximately 100 surgeons who have practiced in Quebec between 1974 and 1989.

Ineligible subjects, for both the implant and control cohorts, were women who had undergone any previous major breast surgery, including reduction mammaplasty, breast lift, and breast cancer surgery. Women who had received other types of silicone or artificial implants, or had a male genotype, or had a history of cancer (excluding nonmelanoma skin cancer) of any site before breast implant surgery were also excluded. No data were collected on other cosmetic procedures performed subsequent to the initial breast implant or on cosmetic procedures. Limited data were collected on breast implant revision surgeries, where applicable.

In total, the cohort consists of 40,451 women; 24,558 received breast implants (7,153 women from Ontario and 17,405 from Quebec). The other cosmetic surgeries control group consists of 15,893 women (4,418 from Ontario and 11,475 from Quebec).

Ascertainment of exposure and comparison groups

For the purpose of the analyses, implant subjects are the exposed group and are by definition women who received bilateral cosmetic breast augmentation. There are two comparison groups. One consists of a cohort of women who received other common elective cosmetic surgeries (unexposed group) and the other consists of women in the general population. Information on year of surgery, age at surgery, personal identifying information (used only for linkage purposes) and verification of eligibility criterias for both the implant subjects and the comparison cohort and information on implant characteristics such as the type of implant, implant envelope, fill volume and site of implantation for the implant subjects was collected by review of medical (hospital or private clinic) records of all women (exposed and unexposed) in the cohorts. This extensive data collection was carried-out by the 30

Ontario and Quebec research teams in phase I of the study using a standardized data collection approach and by validating the retrieved implant characteristics information with breast implant manufacturers. General female population rates for the provinces of Ontario and Quebec will be obtained from provincial vital and cancer registries (unpublished mortality and cancer tabulations, Chronic Diseases Surveillance and Monitoring Division, Public Health Agency of Canada, Ottawa, 2011).

Ascertainment of incident cancers

Incident cases of cancer for the first phase of this project who were diagnosed from the date of surgery until December 31, 1997 were identified by linking personal identifying information of the cohort members to the Canadian Cancer Registry (CCR) (104;105). This computer-accessible population-based national registry is managed by Statistics Canada through collaboration with provincial and territorial cancer registries and captures all cancer cases that occur in Canadians and are diagnosed in Canada or in approximately 20 states in the United-States. The cohort was also linked to cancer incidence data before the index date of surgery, the earliest being year 1969, to exclude women diagnosed with cancer before their surgeries.

Incident cases of cancers for the second phase of this project who were diagnosed between January 1, 1998 and December 31, 2006 (Quebec) or December 31, 2007 (Ontario) were identified by linking to provincial registries, namely the Ontario Cancer Registry (OCR) (106) for the Ontario cohort and the Quebec Tumor Registry (QTR) for the Quebec cohort (107). These provincial cancer registries collect information on cancer cases diagnosed in the province corresponding to the cancer registry. Cancer cases were classified according to the International Classification for Diseases, 9th revision (ICD-9) (108). Where no link was found each patient was assumed to be cancer free at the end of follow-up.

Ascertainment of mortality

The follow-up of the cohort for mortality outcomes started from the date of the index surgery. The cohort was linked in the first phase of this project to the Canadian Mortality Database (CMDB) to determine vital status from the date of the index surgery until 31

December 31, 1997 (105). The CMDB, maintained by Statistics Canada, contains death data from 1950 (109). For out of country deaths, only those that occur in the United States are reported as Canada currently receives abstracted death data from approximately 20 states. Previous studies suggest that the number of deaths missed in cohorts linked with the

CMDB is quite small (110).

The cohort was linked to provincial registries in the second phase of this project to identify mortality cases with cause of death for the period between January 1, 1998 and December 31, 2007 for the Quebec cohort and between January 1, 1998 and December 31, 2006 for the Ontario cohort. Provincial registries used for this linkage were the Ontario Mortality Database (OMDB) for the province of Ontario and the mortality file of Quebec held by the Quebec Institute of Statistics. Causes of death were classified according to the International Classification for Diseases, 9th revision (ICD-9) and according to the ICD 10th revision for deaths that occurred after January 1, 2000 in the province of Quebec (108;111).

Ascertainment of breast cancer cases prognostic factors

In order to assess the concern that breast implants may be associated with more advanced stage at diagnosis of breast cancer and cause poorer survival, after completion of the linkage with national and provincial registries, medical records of breast cancer cases in the cohort were reviewed both for confirmation of the diagnosis of breast cancer as well as to provide information about stage of the breast cancer at diagnosis, tumor size, axillary lymph node involvement, the existence of distant metastasis and tumor histology. The stage of th breast cancer at diagnosis was classified according to the TNM 6 edition (112;113). The grouped TNM stage in these data included the pathologic stage group, augmented by the clinical stage group when the pathologic stage was not recorded.

Statistical analysis

Objective 1 To evaluate the first objective of this thesis, person-years of follow-up were calculated for each woman in the breast implant and other cosmetic surgery cohorts from 1 year after the date of surgery until the earliest of date of death, date of cancer diagnosis, December 31, 32

2006 (Quebec cohort) or December 31, 2007 (Ontario cohort). The first year of follow-up was excluded from analysis, consistent with other investigations (8;39), to reduce the influence that pre-existing disease at the time of surgery may have had on our comparisons. The numbers of person-years and incident cases of cancer was tabulated across strata defined by implant or surgical control group, province of residence at the time of implant (Quebec or Ontario), attained age (18–24, 25–29, 30–34, . . ., 75–79, ≥ 80 years), calendar period of follow-up (1974–1977, 1978–1981, . . .,1994–1997, 1998-2001, 2002-2007), period of surgery (1974–1979, 1980–1984, 1985–1989), age at surgery (18–<30, 30–<40, ≥ 40 years) and time since surgery (1-<5, 5-<10, 10-<15, 15-<20 and ≥20 years). Attained age, calendar period of follow-up and time since surgery were time-dependent variables because women would contribute person-years to different categories within these variables as they are followed over time. In contrast, women would contribute person-years to only one level of the classification variables of period of surgery and age at surgery. The DATAB module in the Epicure software program was used to tabulate person-years of follow-up (114).

We compared overall and site-specific cancer incidence rates for both the breast implant patients and the other cosmetic surgeries group with those for the general population. Cancer incidence rates for the provinces of Ontario and Quebec was obtained from provincial registries as described above. The expected numbers of incident cancers in the cohort was estimated by multiplying the tabulated person-years of follow-up by the corresponding overall and site-specific cancer rate observed in the general population according to province (Ontario or Quebec), age (by 5-year age intervals), and follow-up interval (1974–1977, 1978–1981, . . ., 1994–1997, 1998-2001, 2002-2007). Differences in cancer incidence rates between the implant and surgical control cohorts relative to the general population was evaluated by calculating the standardized incidence ratio (SIR), which is the ratio of the observed-to-expected number of incident cancers (115). The 95 percent confidence intervals were calculated for the SIR by assuming that the observed number of incident cancers followed a Poisson distribution, using formulae detailed elsewhere (115). All the p values reported were two sided.

Comparisons of site-specific incident cancers between the implant recipients and the other plastic surgery patients, rather than the general population, were done using multivariable Poisson regression models using the incidence rate ratios (IRR) as the measure of association (116). We used Cox proportional hazards regression models to evaluate cumulative incidence of breast cancer over the follow-up period (117). The potential confounding influence of the following factors was evaluated: attained age, province of residence, follow-up interval, age at surgery, year of surgery and time since surgery. Confounding was evaluated by a backward deletion approach (118). The analyses were done with SAS, version 9.2 (119).

Objective 2 In order to assess the second objective of this thesis, we restricted the analyses to include only those women who received breast implants with their corresponding person-years of follow-up by using multivariable Poisson regression models using the IRR as the measure of association. These analyses were conducted to identify possible differences in breast cancer incidence according to the following implant characteristics: type of implant (silicone gel-filled implants (SGFIs) and saline), envelope (polyurethane-coated or not), subglandular/submuscular placement, and fill volume. For implant fill volume, women were categorized based on the quartiles of the observed frequency distribution of the mean value of the right and left implants (<175, 175–<200, 200–<225 and ≥ 225 cc) (9). Confounding was assessed with the same approach mentioned above. These analyses were done with SAS, version 9.2 (119).

Objective 3 To evaluate the third objective of this thesis, person-years of follow-up were calculated for each woman in the breast implant and other cosmetic surgery cohorts from 1 year after the date of surgery until the earliest of date of death, December 31, 2006 (Ontario) or December 31, 2007 (Quebec). We excluded the first year of follow-up from the analyses to reduce the influence that pre-existing disease at the time of surgery may have had on our comparisons. The numbers of person-years and deaths were tabulated across strata defined by implant or surgical control group, province of residence at the time of implant (Quebec 34 or Ontario), attained age (18–24, 25–29, 30–34, . . ., 75–79, ≥ 80 years), calendar period of follow-up (1974–1977, 1978–1981, . . .,1994–1997, 1998-2001, 2002-2007), period of surgery (1974–1979, 1980–1984, 1985–1989), age at surgery (18–<30, 30–<40, ≥ 40 years) and time since surgery (1-<5, 5-<10, 10-<15, 15-<20 and ≥20 years). Attained age, calendar period of follow-up and time since surgery were time-dependent variables because women would contribute person-years to different categories within these variables as they were followed over time. In contrast, women would contribute person-years to only one level of the classification variables of period of surgery and age at surgery. The DATAB module in the Epicure software program was used to tabulate person-years of follow-up (114).

The comparison of overall and cause-specific mortality incidence rates for both the breast implant patients and the other cosmetic surgeries group was done with those for the general population. Mortality rates for the provinces of Ontario and Quebec were obtained from provincial vital statistics registries as described above. The expected numbers of deaths in the cohort were estimated by multiplying the tabulated person-years of follow-up by the corresponding overall and cause-specific mortality rates observed in the general population according to province (Ontario or Quebec), age (by 5-year age intervals), and calendar period of follow-up (1974–1977, 1978–1981, . . ., 1994–1997, 1998-2001, 2002-2007). Mortality risk for the implant and surgical control cohorts relative to the general population was evaluated by calculating the standardized mortality ratio (SMR), which is the ratio of the observed-to-expected number of deaths (115). The 95 percent confidence intervals were calculated for the SMR by assuming that the observed number of incident deaths followed a Poisson distribution, using formulae detailed elsewhere (115).

Comparisons of cause-specific deaths, particularly suicide deaths, between the implant recipients and the other plastic surgery patients, rather than the general population, was done using multivariable Poisson regression models using the mortality rate ratio (RR) as the measure of association. The influence of confounding factors was evaluated following the approach mentioned above. These analyses were done with SAS, version 9.2 (119).

Objective 4 The assessment of objective 4 was done by comparing augmented women with breast cancer with other plastic surgery women with breast cancer and by performing multivariable logistic regression models and multinomial logit models to identify possible differences in breast cancer stage at diagnosis and prognosis factors between the two groups. Specifically, breast cancer stage at diagnosis (TNM stage I, II, III/IV, or unknown), tumor size (<21, 21-≤50, > 50 mm, or unknown), lymph node involvement (yes, no, or unknown) and tumor histology (infiltrating ductal carcinoma, lobular carcinoma, or other) were used as the primary outcome variables. The influence of the following confounding variables was evaluated: age at index surgery (18 - <25, 25 - <30, 30 - <35, 35 - <40, 40 - <45, or ≥45 years), province of residence (Ontario or Quebec), calendar period of index surgery (1974-1977, 1978-1981, 1982-1985, or 1986-1989), age at diagnosis (<45, 45 - <50, 50 - <60, or ≥60 years) and period of breast cancer diagnosis (<1990, 1990-1997, or ≥ 1998). Because Ontario started their organized breast cancer screening program in 1990 and Quebec started their organized breast cancer screening program in 1998, period of breast cancer diagnosis was grouped into three calendar periods (<1990, 1990-1997, and ≥ 1998). Evaluation of confounding in the multivariable logistic models was done following the same approach discussed earlier. These analyses were done with SAS, version 9.2 (119).

The evaluation of detection of breast cancer in terms of stage at diagnosis among women with breast implants was also done by ways of a systematic review and meta-analysis. Specifically, a systematic search of the literature was done using specific keywords to identify all relevant scientific articles. Eligible studies were used for a quantitative meta- analysis in order to calculate a pooled effect of the association between cosmetic breast implants and breast cancer stage at diagnosis. The meta-analysis was conducted using the Stata software, version 11 (120).

Objective 5 36

Restricted analyses to include only those women who received breast implants and developed breast cancer was performed to identify possible differences in breast cancer stage at diagnosis in terms of implant characteristics; type of implant (silicone gel-filled implants (SGFIs) and saline), envelope (polyurethane-coated or not), subglandular/submuscular placement, and fill volume. The later variable was categorized based on the categorization used in the previous scientific article in the first phase of this study that assessed these associations(<200 cc, ≥ 200 cc) (56). Crude and adjusted odds ratios (ORs) and their 95% confidence intervals (CIs) were calculated using logistic regression models and multinomial logit models. The influence of the confounding variables mentioned in the statistical analyses of objective 5 was evaluated with the same approach mentioned earlier. These analyses were done with SAS, version 9.2 (119).

Objective 6 To compare breast cancer-specific survival between the implant and other plastic surgery patients we applied the Cox proportional hazards model adjusting for factors such as age at diagnosis, period of diagnosis and province of residency. Another model was adjusted for stage of breast cancer at diagnosis for comparison purposes. The assumption of proportionality was evaluated by inspecting plots of the log negative log survival curves in proportional hazards models. Survival curves to investigate differential survival between groups were produced using Cox proportional hazards model after adjusting for covariates. Cox proportional hazards model analyses were done using SAS, version 9.2 (119).

A systematic review and meta-analysis was also conducted to evaluate breast cancer- specific survival among cosmetic breast implant women. A systematic search of the literature was done using specific keywords to identify all relevant scientific articles. Eligible studies were used for a quantitative meta-analysis in order to calculate a pooled effect of the association between cosmetic breast implants on breast cancer-specific survival. The meta-analysis was conducted using the Stata software, version 11 (120).

Objective 7 37

Breast cancer-specific survival among breast implant women was evaluated comparing survival according to specific implant characteristics; type of implant (silicone gel-filled implants (SGFIs) and saline), envelope (polyurethane-coated or not), subglandular/submuscular placement, and fill volume (<200 cc, ≥ 200 cc). The Cox proportional hazards model was used adjusting for factors mentioned above. The assumption of proportionality was evaluated by inspecting plots of the log negative log survival curves in proportional hazards models. These analyses were done with SAS, version 9.2 (119).

Study power

To evaluate the statistical power, we calculated for all of the objectives the least significant detectable measures of association with α = 0.05 (two sided test) and β = 0.20 (study power of 80 %) used in all of the calculations. For internal analysis (implant women vs. other cosmetic surgery women) using Poisson, Cox and Logistic regression models, we used the software Power Analysis and Sample Size (PASS), 2008 (121). When the implant cohort was compared with the female general population, we used the Least Significant Relative Risk (LSRR) method proposed by Walter (122). Power calculations were first performed before conducting all of the analyses and then performed again after data collection. The following power calculations describe the latter.

Objectives 1 to 3 Study power analyses were first done for objectives 1 and 3 when the cosmetic breast implant cohort was compared with the female general population for mortality and cancer incidence. The Least Significant Relative Risk (LSRR) method proposed by Walter has been used to evaluate the relative risk which it is possible to demonstrate as statistically significant with a given person-years of follow-up and the disease/death rate in the general female population for SIR’s and SMR’s used in objective 1 and 3 (122). Person-years of follow-up (PYOF) and disease/death rate were obtained from the main analyses presented in chapters 2 and 3. Table 1 and 2 describe the LSRR for the association between cosmetic breast implants and overall and site specific cancer incidence and mortality applying the methods proposed by Walter, with PYOF of 552,726 for cancer analyses and 571,683 for 38 mortality analyses and specified cancer and mortality incidence rates in the reference population (Canadian general population, unpublished mortality and cancer tabulations, Chronic Diseases Surveillance and Monitoring Division, Public Health Agency of Canada, Ottawa, 2011). The incidence rates were the ones expected in the implant cohort. They were obtained by multiplying the tabulated person-years of follow-up by the corresponding rate observed in the general female population according to province (Ontario or Quebec), age (by 5-year age intervals), and calendar period of follow-up (1974–1977, 1978–1981, . . ., 1994–1997, 1998-2001, 2002-2007). It is shown in Table 1 and 2 that the LSRR for overall mortality are 0.90 and 1.10 and are 0.92 and 1.09 for overall cancer incidence.

Power calculations of objectives 1, 2 and 3 when using Poisson regression models for the internal analysis were done with the PASS software using methods proposed by Signorini (123). We calculated the least significant relative risks (LSRR) for cancer incidence and mortality when comparing cosmetic breast implant women to other cosmetic surgery women. Parameters used for calculation were a cancer incidence or mortality rate for the other cosmetic surgery group (reference group) specified in tables 3 and 4, a proportion of women with implants (proportion of exposed) of 61 %, an over-dispersion parameter of 1, a sample size of 40,451 and a mean follow-up time of 23.7 (cancer incidence) and 24.5 (mortality). The different parameters used in these calculations were obtained from the analyses presented in chapters 2 and 3. The LSRR for cancer incidence and mortality are detailed in tables 3 and 4. For instance, the LSRR for overall cancer incidence are 0.90 and 1.10 and they are 0.88 and 1.12 for overall mortality.

The LSRR were calculated for a Poisson regression model for breast cancer incidence when comparing augmented women for specific implant characteristics. Calculations attained best study power for the analysis by site of implantation with LSRR of 0.80 and 1.25 and attained worst study power for the analysis by the type of implant with LSRR of 0.49 and 2.05 (Table 5). Parameters used for calculations according to each implant characteristic were an over-dispersion parameter of 1, a mean follow-up time of 23.7 and a sample size, a proportion of exposed and a breast cancer incidence rate for the reference group obtained from chapter 2. 39

Objective 4 To evaluate statistical power for the analysis in objective 4 when using Logistic regression models, we used the PASS software applying methods proposed by Hsieh (124). The least significant odds ratios (LSOR) were calculated to evaluate the likelihood of an advanced stage at breast cancer diagnosis when comparing women with cosmetic breast implants to other cosmetic surgery women. The following parameters were used: the total number of women with implants and with other cosmetic surgery specified in the table, an estimated probability that women with other cosmetic surgery will have stage 3/4 or stage 2 breast cancers relative to stage 1 breast cancers and an estimated proportion of the sample size in the exposed group (cosmetic breast implant women). The different parameters were obtained from the analyses presented in chapter 4. The LSOR for investigating stage 3/4 breast cancers vs. stage 1 are 0.34 and 2.11 (Table 6).

Objective 5 To calculate the LSOR for the analysis in objective 5 when using Logistic regression models, we used the PASS software applying methods proposed by Hsieh (124). This analysis focuses on the likelihood of advanced stage at breast cancer diagnosis among cosmetic breast implant women according to specific implant characteristics. Calculations attained best study power for the analysis by the fill volume with LSOR of 0.36 and 2.11 and attained worst study power for the analysis of polyurethane coating with LSRR of < 0.01 and 3.61 (Table 7). Parameters used for these calculations are the number of women with implants with breast cancer specified in the table, a probability that women in the reference group according to the specified implant characteristic will have stage 3/4 breast cancers relative to stage 1/2 and a proportion of the sample size in the exposed group according to the implant characteristic. The different parameters were obtained from the analyses presented in chapter 4.

Objectives 6 40

The assessment of statistical power to detect the least significant hazard ratios (LSHR) for the objective 6 has been conducted with the PASS software using methods proposed by Hsieh and Lavori (2000) and Shoenfeld (1983) for Cox Proportional Hazards regression models (125;126). The LSHR are 0.52 and 1.92 when investigating whether there is a differential breast cancer-specific survival between the implant patients with breast cancer and the other plastic surgery patients with breast cancer (Table 8). Parameters used for this calculation were the number of breast cancer cases among both the implant women and the other cosmetic surgery women (N = 853), a proportion of subjects in which breast cancer- specific mortality occurs (proportion of non-censored subjects) of 9%, a proportion (p = 48%) of the sample size in the exposed group (cosmetic breast implant women with breast cancer) and an estimation of the standard deviation of the independent variable (implants vs. other cosmetic surgery) of 0.5. Estimations were based on the previous reporting of this study (56). The different parameters were obtained from the analyses presented in chapter 4. Additionally, the standard deviation was estimated using the following formula proposed by the PASS software: √ p (1 – p) (121).

Objective 7 Statistical power to detect the least significant hazard ratios (LSHR) for the objective 7 has been conducted with the PASS software using methods proposed by Hsieh and Lavori (2000) and Shoenfeld (1983) for Cox Proportional Hazards regression models (125;126). This analysis investigates whether there is a differential breast cancer-specific survival between augmented women with breast cancer focusing on specific implant characteristics. Calculations attained best study power for the analysis by the fill volume with LSHR of 0.49 and 2.03 and attained worst study power for the analysis by polyurethane coating with LSHR of 0.23 and 4.26 (Table 9). Parameters used for these calculations were the number of breast cancer cases among the augmented women specified in the table, a proportion of breast cancer-specific mortality specified in the table, a proportion of the sample size in the exposed group according to the implant characteristic evaluated and an estimated standard deviation for each implant characteristic using the formula described above. The different parameters were obtained from the analyses presented in chapter 4.

Analyses of study power show that for most objectives of this thesis it is reasonable to test the proposed hypotheses. Although this study does not have adequate power to test certain hypotheses, it can only produce suggestive information to guide future research.

Ethical considerations

Ethics approval for the study was granted by the University of Toronto’s Office of Research Ethics, the ethics committee of the Centre Hospitalier Affilié universitaire de Québec’s (CHA) Saint-Sacrement Hospital and the Ethics Committee for Clinical Research of Laval University.

Table 1. Least Significant Relative Risk (LSRR) comparing cancer incidence (overall and site-specific) between cosmetic breast implant women and the female general population applying the methods proposed by Walter (122). Cancer site Incidence Ratea LSRRb All sites 388.23 (0.92, 1.09) Stomach 4.66 (0.35, 1.96) Colorectal 36.44 (0.74, 1.30) Pancreas 6.36 (0.43, 1.80) Lung 46.97 (0.77, 1.27) Malignant melanoma 9.38 (0.51, 1.64) Breast 138.68 (0.86, 1.15) Genital 54.92 (0.78, 1.24) Cervix 13.28 (0.58, 1.53) Endometrial 21.38 (0.66, 1.40) Ovary 15.78 (0.59, 1.52) Bladder 6.93 (0.45, 1.76) Kidney 8.03 (0.48, 1.70) Nervous system 5.93 (0.41, 1.83) Brain 6.18 (0.42, 1.81) Thyroid 13.15 (0.58, 1.53) Lymphohematopoietic 25.62 (0.69, 1.37) Non-Hodgkin’s lymphoma 13.57 (0.59, 1.52) Leukemia 6.53 (0.43, 1.79) Other cancer sites not listed above 31.37 (0.72, 1.33) a. Expected incidence rates per 100,000 person-years for (general female population 18 years of age or older) the referent group and obtained by multiplying the tabulated person-years of follow-up by the corresponding rate observed in the general female population according to province, age and calendar period of follow-up. b. Parameters used for calculations: Total person-years of follow-up = 552,726; α = 0.05 (two sided test); β = 0.20 (study power of 80 %).

Table 2. Least Significant Relative Risk (LSRR) for mortality (overall and cause specific) between cosmetic breast implants and female general population applying the methods proposed by Walter (122). Cause of death Mortality Ratea LSRRb All causes 288.79 (0.90, 1.10) Infectious diseases 4.28 (0.34, 1.98) Endocrine diseases 9.90 (0.54, 1.60) Mental disorders 4.12 (0.33, 2.00) Circulatory diseases 60.36 (0.80, 1.23) Coronary heart 30.81 (0.72, 1.32) Cerebrovascular 13.52 (0.60, 1.51) Respiratory diseases 15.28 (0.62, 1.47) Digestive diseases 10.49 (0.55, 1.58) Genitourinary diseases 3.42 (0.28, 2.12) Cancer 137.35 (0.86, 1.15) Breast 30.49 (0.72, 1.32) Brain 4.18 (0.34, 2.00) Genital 13.43 (0.59, 1.51) Colorectal 13.74 (0.60, 1.50) Bronchus and lung 35.23 (0.74, 1.30) Lymphohematopoietic 24.77 (0.69, 1.36) Injuries 23.94 (0.69, 1.37) Suicide 9.84 (0.54, 1.60) Motor vehicle accidents 6.04 (0.43, 1.80) Other 8.06 (0.49, 1.68) Other deaths 19.65 (0.66, 1.41) a. Expected incidence rates per 100,000 person-years for (general female population 18 years of age or older) the referent group and obtained by multiplying the tabulated person-years of follow-up by the corresponding rate observed in the general female population according to province, age and calendar period of follow-up. b. Parameters used for calculations: Total person-years of follow-up = 571,683; α = 0.05 (two sided test); β = 0.20 (study power of 80 %).

Table 3. Least Significant Relative Risk (LSRR) for cancer incidence among breast implant women vs. other cosmetic surgery women applying the methods proposed by Signorini (123) for Poisson regression models.a Cancer site Incidence Rateb LSRR All sites 342.03 (0.90, 1.10) Stomach 5.05 (0.41, 2.22) Colorectal 30.00 (0.71, 1.39) Pancreas 6.17 (0.45, 2.06) Lung 46.82 (0.76, 1.30) Malignant melanoma 7.85 (0.50, 1.90) Breast 128.12 (0.84, 1.17) Genital 48.50 (0.76, 1.30) Cervix 11.21 (0.50, 1.72) Endometrial 19.90 (0.65, 1.50) Ovary 13.73 (0.60, 1.63) Bladder 5.05 (0.41, 2.22) Kidney 8.41 (0.51, 1.86) Nervous system 7.56 (0.49, 1.93) Brain 7.29 (0.48, 1.95) Thyroid 8.69 (0.52, 1.84) Lymphohematopoietic 22.15 (0.67, 1.47) Non-Hodgkin’s lymphoma 12.62 (0.58, 1.66) Leukemia 4.49 (0.39, 2.32) Other cancer sites not listed above 5.89 (0.44, 2.09) a. Parameters used for calculations: N = 40,451; α = 0.05 (two sided test); Cancer incidence rate for the reference (other cosmetic surgery) group specified in the table; φ (over-dispersion parameter) = 1; Proportion in the exposed (implant women) group = 61 %; Mean follow-up time (years) = 23.7; β = 0.20 (study power of 80 %). b. Incidence rates per 100,000 person-years for the referent group (other cosmetic surgery women).

Table 4. Least Significant Relative Risk (LSRR) for mortality (overall and cause specific) among breast implant women vs. other cosmetic surgery women applying the methods proposed by Signorini (123) for Poisson regression models.a Cause of death Mortality Rateb LSRRb All causes 235.55 (0.88, 1.12) Infectious diseases 3.50 (0.34, 2.54) Endocrine diseases 2.16 (0.24, 3.26) Mental disorders 2.16 (0.24, 3.26) Circulatory diseases 47.16 (0.76, 1.30) Coronary heart 23.45 (0.68, 1.44) Cerebrovascular 12.67 (0.59, 1.65) Respiratory diseases 9.97 (0.55, 1.76) Digestive diseases 4.04 (0.37, 2.39) Genitourinary diseases 1.08 (0.09, 5.20) Cancer 123.17 (0.85, 1.18) Breast 22.10 (0.67, 1.46) Brain 7.55 (0.50, 1.90) Genital 11.05 (0.56, 1.71) Colorectal 8.09 (0.51, 1.87) Bronchus and lung 40.70 (0.75, 1.32) Lymphohematopoietic 9.43 (0.50, 1.78) Injuries 27.22 (0.70, 1.41) Suicide 13.48 (0.60, 1.62) Motor vehicle accidents 5.93 (0.45, 2.05) Other 7.28 (0.49, 1.92) Other deaths 9.16 (0.53, 1.79) a. Parameters used for calculations: N = 40,451; α = 0.05 (two sided test); Cancer incidence rate for the reference (other cosmetic surgery) group specified in the table; φ (over-dispersion parameter) = 1; Proportion in the exposed (implant women) group = 61 %; Mean follow-up time (years) = 24.5; β = 0.20 (study power of 80 %). b. Incidence rates per 100,000 person-years for the referent group (other cosmetic surgery women).

Table 5. Least Significant Relative Risk (LSRR) for breast cancer incidence among cosmetic breast implant women according to specific implant characteristics applying the methods proposed by Signorini (123) for Poisson regression models. Implant characteristics Incidence Rate* LSRR

Type of filla 79.56 (0.49, 2.05) Fill volumeb 70.92 (0.75, 1.31) Site of implantationc 86.30 (0.80, 1.25) Polyurethane coatingd 71.86 (0.61, 1.54) a. Parameters used for calculations: N = 16,990; α = 0.05 (two sided test); Women with silicone breast implants (exposed group) are compared with women with saline implants (reference group); Breast cancer incidence rate for the reference group (women with saline implants) specified in the table; φ (over-dispersion parameter) = 1; Proportion in the exposed (silicone implants) group = 95 %; Mean follow-up time (years) = 23.7; β = 0.20 (study power of 80 %). b. Parameters used for calculations: N = 24,397; α = 0.05 (two sided test); Women with breast implants fill volume ≥ 200 cc (exposed group) are compared with women with fill volume < 200 cc (reference group). Breast cancer incidence rate for the reference group (women with fill volume < 200 cc) specified in the table; φ (over-dispersion parameter) = 1; Proportion in the exposed (fill volume ≥ 200 cc) group = 50 %; Mean follow-up time (years) = 23.7; β = 0.20 (study power of 80 %). c. Parameters used for calculations: N = 21,774; α = 0.05 (two sided test); Women with subglandular implant placement (exposed group) are compared with women with submuscular implant placement (reference group); Breast cancer incidence rate for the reference group (women with submuscular implant placement) specified in the table; φ (over-dispersion parameter) = 1; Proportion in the exposed (submuscular placement) group = 63 %; Mean follow-up time (years) = 23.7; β = 0.20 (study power of 80 %). d. Parameters used for calculations: N = 18,465; α = 0.05 (two sided test); Women with breast implants with polyurethane coating (exposed group) are compared with breast implant women without polyurethane coating (reference group); Breast cancer incidence rate for the reference group (women without polyurethane coating) specified in the table; φ (over-dispersion parameter) = 1; Proportion in the exposed (polyurethane coating) group = 14 %; Mean follow-up time (years) = 23.7; β = 0.20 (study power of 80 %). * Incidence rates per 100,000 person-years for the reference group.

Table 6. Least significant odds ratio (LSOR) for stage of breast cancer at diagnosis among cosmetic breast implant women vs. other cosmetic surgery women applying the methods proposed by Hsieh (124) for Logistic regression models. Stage at diagnosis LSOR

IIa (0.64, 1.57) III/IVb (0.34, 2.11) a. Parameters used for calculations: N = 636; α = 0.05 (two sided test); Logistic regression of a response variable (Stage II vs. Stage I breast cancers) on a binary independent variable (Implant women vs. oth er cosmetic surgery women); Probability that women with other cosmetic surgery will have stage II breast cancers relative to stage I breast cancers) = 0.48; Proportion of N in the exposed (cosmetic breast implants) group = 45 %; β = 0.20 (study power of 80 %). b. Parameters used for calculations: N = 390; α = 0.05 (two sided test); Logistic regression of a response variable (Stage III/IV vs. Stage I breast cancers) on a binary independent variable (Implant women vs. other cosmetic surgery women); Probability that women with other cosmetic surgery will have stage III/IV breast cancers relative to stage I breast cancers) = 0.13; Proportion of N in the exposed (cosmetic breast implants) group = 47 %; β = 0.20 (study power of 80 %).

Table 7. Least significant odds ratio (LSOR) for stage of breast cancer at diagnosis among cosmetic breast implant women according to specific implant characteristics applying the methods proposed by Hsieh (124) for Logistic regression models. Implant characteristics LSOR

Type of filla (0.19, 3.00) Fill volumeb (0.36, 2.11) Site of implantationc (0.35, 2.15) Polyurethane coatingd (<0.01, 3.61) a. Parameters used for calculations: N = 280; α = 0.05 (two sided test); Logistic regression of a binary response variable (Stage III/IV vs. Stage I/II breast cancers) on a binary independent variable (silicone implants vs. saline implants); Probability that women with saline implants will have stage III/IV breast cancers relative to stage I/II breast cancers) = 0.14; Proportion of N in the exposed (silicone implants) group = 82 %; β = 0.20 (study power of 80 %). b. Parameters used for calculations: N = 339; α = 0.05 (two sided test); Logistic regression of a binary response variable (Stage III/IV vs. Stage I/II breast cancers) on a binary independent variable (fill volume ≥ 200 cc vs. fill volume < 200 cc); Probability that women with fill volume < 200 cc will have stage III/IV breast cancers relative to stage I/II breast cancers) = 0.16; Proportion of N in the exposed (fill volume ≥ 200 cc) group = 50 %; β = 0.20 (study power of 80 %). c. Parameters used for calculations: N = 300; α = 0.05 (two sided test); Logistic regression of a binary response variable (Stage III/IV vs. Stage I/II breast cancers) on a binary independent variable (submuscular vs. subglandular); Probability that women with submuscular implant placement will have stage III/IV breast cancers relative to stage I/II breast cancers) = 0.18; Proportion of N in the exposed (subglandular implant placement) group = 42 %; β = 0.20 (study power of 80 %). d. Parameters used for calculations: N = 237; α = 0.05 (two sided test); Logistic regression of a binary response variable (Stage III/IV vs. Stage I/II breast cancers) on a binary independent variable (polyurethane coating vs. without polyurethane coating); Probability that women without polyurethane coating implants will have stage III/IV breast cancers relative to s tage I/II breast cancers) = 0.15; Proportion of N in the exposed (polyurethane coating implants) group = 11 %; β = 0.20 (study power of 80 %).

Table 8. Least significant hazard ratio (LSHR) to investigate whether there is a differential breast cancer-specific survival between the implant women and other plastic surgery women applying the methods proposed by Hsieh and Lavori (125) and Shoenfeld (126) for Cox Proportional Hazards regression models.* Breast cancer death rate LSHR

0.09 (0.52, 1.92) * Parameters used for calculations: N = 834; α = 0.05 (two sided test); Cox regression for breast cancer specific survival on a binary independent variable (cosmetic breast implants vs. other cosmetic surgery); Overall event (breast cancer specific mortality) rate specified in the table; Proportion o f N in the exposed (cosmetic breast implants) group = 48 %; Estimated standard deviation of binary independent variable = 0.5; β = 0.20 (study power of 80 %).

Table 9. Least significant hazard ratio (LSHR) to investigate whether there is a differential breast cancer-specific survival among cosmetic breast implant women according to specific implant characteristics when applying the methods proposed by Hsieh and Lavori (125) and Shoenfeld (126) for Cox Proportional Hazards regression models. Implant characteristics Breast cancer death rate LSHR

Type of filla 0.21 (0.41, 2.44) Fill volumeb 0.19 (0.49, 2.03) Site of implantationc 0.18 (0.45, 2.20) Polyurethane coatingd 0.17 (0.23, 4.26) a. Parameters used for calculations: N = 284; α = 0.05 (two sided test); Cox regression for breast cancer specific survival on a binary independent variable (silicone implants vs. saline implants); Overall event (breast cancer specific mortality) rate specified in the table; Proportion of N in the exp osed (silicone implants) group = 79 %; Estimated standard deviation for binary independent variable = 0.41; β = 0.20 (study power of 80 %). b. Parameters used for calculations: N = 331; α = 0.05 (two sided test); Cox regression for breast cancer specific survival on a binary independent variable (fill volume ≥ 200 cc vs. fill volume < 200 cc); Overall event (breast cancer specific mortality) rate specified in the table; Proportion of N in the exposed (fill volume ≥ 200) group = 49 %; Estimated standard deviation for binary independent variable = 0.5; β = 0.20 (study power of 80 %). c. Parameters used for calculations: N = 295; α = 0.05 (two sided test); Cox regression for breast cancer specific survival on a binary independent variable (subglandular implant placement vs. submuscular implant placement); Overall event (breast cancer specific mortality) rate specified in the table; Proportion of N in the exposed (subglandular implant placement) group = 42 %; Estimated standard deviation for binary independent variable = 0.5; β = 0.20 (study power of 80 %). d. Parameters used for calculations: N = 230; α = 0.05 (two sided test); Cox regression for breast cancer specific survival on a binary independent variable (polyurethane coating vs. without polyurethane coating); Overall event (breast cancer specific mortality) rate specified in the table; Proportion of N in the exposed (polyurethane coating implants) group = 11 %; Estimated standard deviation for binary independent variable = 0.31; β = 0.20 (study power of 80 %).

Chapter 2: Canadian Breast Implant Cohort: extended follow- up of cancer incidence

Résumé

Objectifs : Les implants mammaires pour fins esthétiques ne sont pas associés à une augmentation de risque de cancer du sein, mais les variations du risque selon les caractéristiques des implants ne sont pas encore bien comprises. De plus, le risque de cancer à d’autres sites que le sein doit être clarifié. Cette étude a pour objet de combler ces lacunes. Méthodes : Cette étude présente une analyse de 10 ans de plus de suivi d’une étude de cohorte de femmes ayant reçu soit des implants mammaires pour fins esthétiques (n=24 558) ou une autre chirurgie esthétique (15 893). Plus de 70% de la cohorte a été suivi pour plus de 20 ans. L’incidence du cancer parmi le groupe avec implants mammaires a été comparée au groupe contrôle avec des modèles de Poisson multivariés ainsi qu’aux femmes de la population générale en utilisant le Rapport Standardisé d’Incidence (RSI). Résultats : Les femmes avec des implants mammaires avaient des taux réduits de cancers du sein et de l’endomètre comparativement aux femmes avec une autre chirurgie esthétique. Les implants mammaires en position rétro-glandulaire étaient associés à un taux réduit de cancer du sein comparativement aux implants en position rétro-pectorale (Rapport de Taux d’incidence = 0,78, IC à 95% = 0,63-0,96). Un taux 7 fois plus élevé de cancer du sein a été observé (RT d’incidence = 7,36, IC à 95% = 1,86-29,12) dans les cinq premières années après la chirurgie pour les femmes ayant des implants en position rétro-glandulaire avec enveloppe au polyuréthane comparativement aux femmes ayant des implants en position rétro-glandulaire sans polyuréthane, mais le RT d’incidence diminuait progressivement avec le temps (valeur p de tendance = 0,02). Nous n’avons observé aucune augmentation du risque de formes de cancer plus rares parmi les femmes avec implants mammaires. Conclusion : Une réduction de l’incidence de cancer du sein a été observée pour les femmes avec implants mammaires en position rétro-glandulaire relativement à la position rétro-pectorale. L’augmentation possible de l’incidence de cancer du sein pour les implants avec enveloppe au polyuréthane peu après la chirurgie nécessite d’être clarifiée.

Abstract

Objectives: Cosmetic breast implants are not associated with increased breast cancer incidence, but variations of risk according to implant characteristics are still poorly understood. As well, the assessment of cancer risk for sites other than breast needs to be clarified. The purpose of this study was to fill these research gaps. Methods: This study presents an extended analysis of 10 more years of follow-up of a large Canadian cohort of women who received either cosmetic breast implants (n=24,558) or other cosmetic surgery (15,893). Over 70% of the implant cohort was followed for over 20 years. Cancer incidence among implant women was compared to those of controls using multivariate Poisson models and the general female population using the Standardized Incidence Ratios (SIRs). Results: Women with breast implants had reduced rates of breast and endometrial cancers compared with other surgery women. Subglandular implants were associated to a reduced rate of breast cancer compared to submuscular implants (Incidence Rate Ratio (IRR) = 0.78, 95% CI= 0.63-0.96) and this reduction persisted over time. We observed a 7-fold increased rate (IRR = 7.36, 95% CI= 1.86-29.12) of breast cancer in the first five years after the date of surgery for polyurethane-coated subglandular implant women compared to other women with subglandular implants without polyurethane coating, but this IRR decreased progressively over time (p value for trend = 0.02). We also observed no increased risk of rarer forms of cancer among augmented women. Conclusion: A reduction in breast cancer incidence was observed for women with subglandular implants relative to women with submuscular implants. Possible increase of breast cancer incidence shortly after breast augmentation with polyurethane implants needs to be verified.

Introduction

Cosmetic breast implants have been the subject of numerous investigations of the long term risk of mortality and cancer incidence (1-3). Early concerns focused on their potential carcinogenic effect, especially for breast cancer, because of the possible link between silicone and such disease (4). Consequently, silicone gel-filled breast implant (SGFIs) were removed from the market in the United-States and Canada in the early 1990s, but were reapproved for general cosmetic use in both countries in 2006 because later studies showed no carcinogenic effect of silicone (1-3;5). Indeed, a recent report from the U.S. Food and Drug Administration concluded that cosmetic breast implants are not associated with increased breast cancer incidence (3). However, examining long term health risks associated with these devices is still important as adverse health effects may only occur after a long latency period.

More recently, breast implants have been studied in an attempt to evaluate variations of breast cancer risk according to specific implant characteristics. Results across epidemiological studies that collected information on implant characteristics have been inconclusive (6-11). This paucity of evidence may be due to the relatively limited number of breast cancer cases across studies. Common implant characteristics include the type of implant (saline or silicone gel-filled implants (SGFIs)), the placement of the implant (submuscular or subglandular), the implant fill volume and the implant envelope (polyurethane coated or not). Polyurethane foam-covered breast implants were withdrawn from the market in 1991, both in the United-States and Canada, when a report showed that polyurethane could degrade into significant quantities of 2,4-diaminotoluene (2,4-toluene diamine) (TDA) (12) which has been recognized as an animal carcinogen and potential human carcinogen (12). However, polyurethane implants are still used in Europe and South America. Little is known to date about the long term health effects in humans of such implants. Evaluation of the potential cancer risk associated with these devices, especially in the subglandular position because of proximity to breast tissue, is important. Furthermore, the concern raised in our earlier study of a possible two-fold increased risk of breast cancer for women with polyurethane implants in the subglandular position needs to be clarified over a longer period of follow-up (7).

A number of epidemiological investigations have evaluated the relationship between breast implants and the incidence of cancer at sites other than the breast (1;6;7;13-18). The findings from these studies have been largely negative when women with implants were compared with women who had other cosmetic surgeries (6;7;17) or with women from the general population (6;7;14;16). The absence of a positive association could be due to lack of effect but also it could be explained by methodological limitations such as the small number of identified incident cancers and/or short follow-up (18). For instance, few studies had a follow-up time over 30 years (6;13;15). However, some studies reported an increased risk for some types of cancer when compared with general population estimates including the brain (19), lung (13;15;19;20), vulva (14) and cervix (18). The assessment of cancer incidence at sites other than the breast was reported in our earlier study (7). Although elevated or reduced risks for some types of cancers were observed, these results were not statistically significant. Moreover, a recent report by the FDA as well as review articles recommended that the risk of hematopoietic malignancies among women who have cosmetic breast implants be further investigated (18;21;22). Taken as a whole, the epidemiological evidence for risk of cancer at body sites other than the breast, especially for hematopoietic malignancies, needs to be further clarified.

In this updated analysis, 10 more years of follow-up have been added to the largest cohort study carried-out to date on cosmetic breast implants. The considerable number of additional incident breast cancer cases provides opportunities to evaluate breast cancer risk according to implant characteristics over a much longer period of time. Additionally, the assessment of non-breast cancer risk will confirm or verify several of the associations that were observed in our previous analyses. Finally, this analysis provides some comparisons of current results with those of our previous publication of cancer incidence among augmented women (7).

Material and methods

Study design, study population and selection criteria

The study population was described in detail in our previous publication (7). The cohort consisted of women, 18 years of age or older, who were residents of the province of Ontario or Quebec, in Canada, and who underwent bilateral cosmetic breast augmentation (implant group) or received other common elective cosmetic surgeries (controls/comparison group) in their province of residence between January 1, 1974, and December 31, 1989. Other cosmetic surgeries included the following: chemical peel or dermabrasion, coronal brow lift (eyebrow and forehead lift), otoplasty (ear surgery), rhinoplasty (nose surgery), rhytidectomy (face-lift), or blepharoplasty (eyelid surgery). Implant women were frequency matched to other plastic surgery patients by year of entry into the cohort, province of residence and by surgeon.

Ineligible subjects, for both the implant and control groups, were women who had undergone any previous major breast surgery, including reduction mammaplasty, breast lift, and breast cancer surgery. We also excluded women who received other types of silicone or artificial implants, or had a male genotype, or had a history of cancer (excluding nonmelanoma skin cancer) before surgery. Frequency of women excluded for various reasons is documented in our earlier paper (7).

In total, the cohort consisted of 40,451 women: 24,558 received cosmetic breast implants (7,153 women from Ontario and 17,405 from Quebec) and 15,893 women (4,418 from Ontario and 11,475 from Quebec) received other common elective cosmetic surgeries. Information on year of surgery, age at surgery, personal identifying information (used only for linkage purposes) and verification of eligibility criteria for both the implant subjects and the controls and information on implant characteristics such as the type of implant, implant envelope, fill volume and site of implantation, was collected by review of medical (hospital or private clinic) records of all women (implant and control) in the cohort.

The implant and control cohorts were compared to the general population of women. General female population rates of cancer incidence and mortality for the provinces of Ontario and Quebec were obtained from provincial vital and cancer registries (unpublished mortality and cancer tabulations, Chronic Diseases Surveillance and Monitoring Division, Public Health Agency of Canada, Ottawa, 2011).

Ascertainment of outcomes

Incident cases of cancer and deaths that occurred from the date of surgery until December 31, 2006 (Quebec) or December 31, 2007 (Ontario) were identified by linking personal identifying information (surname, given and maiden names, mother’s name, father’s name, birth date, residential address and health insurance number) of the cohort members to national and provincial cancer and mortality registries. Specifically, in our previous follow- up (7), cohort members were linked to the Canadian Cancer Registry (CCR) (23) and the Canadian Mortality Database (CMDB) (24) until December 31, 1997. These national registries are managed by Statistics Canada through collaboration with provincial and territorial cancer registries and capture all cancer cases and deaths that occur in Canada and in approximately 20 states in the United-States. The cohort was also linked to cancer incidence data before the index date of surgery, the earliest being 1969. This enabled us to exclude women diagnosed with cancer before their index cosmetic surgeries. Secondly, for this extended follow-up, incident cases of cancer who were diagnosed between January 1, 1998 and December 31, 2006 (Quebec) or December 31, 2007 (Ontario) were identified by linking to provincial registries, namely the Ontario Cancer Registry (OCR) (25) for the Ontario cohort and the Quebec Tumor Registry (QTR) for the Quebec cohort (26). These provincial cancer registries collect information on cancer cases diagnosed in the province corresponding to the cancer registry. The cohort was also linked to provincial mortality databases to identify mortality cases for the period between January 1, 1998 and December

31, 2006 for the Quebec cohort using the mortality file of Quebec held by the Quebec Institute of Statistics and between January 1, 1998 and December 31, 2007 for the Ontario cohort with the use of the Ontario Mortality Database (OMDB) provided by the Registrar General of Ontario. Linkage of the Quebec cohort to the QTR and mortality file was conducted using a deterministic approach while the linkage of the OCR and OMDB to the Ontario cohort was conducted using a probabilistic record linkage system (27). Where no link was found each patient was assumed to be cancer free and alive at the end of follow- up.

Statistical analysis

Person-years of follow-up were calculated for each woman in the breast implant and other cosmetic surgery cohorts from 1 year after the date of surgery until the earliest of date of death, date of cancer diagnosis, December 31, 2006 (Quebec cohort) or December 31, 2007 (Ontario cohort). The first year of follow-up was excluded from analysis, consistent with other investigations (17;18), to reduce the influence that pre-clinically detectable cancers at the time of index cosmetic surgery may have had on our comparisons. The numbers of person-years and incident cases of cancer were tabulated across strata defined by study group (implant or surgical control group), province of residence at the time of index cosmetic surgery (Quebec or Ontario), attained age (18–24, 25–29, 30–34, . . ., 75–79, ≥ 80 years), calendar period of follow-up (1974–1977, 1978–1981, . . .,1994–1997, 1998-2001, 2002-2007), period of surgery (1974–1979, 1980–1984, 1985–1989), age at surgery (18– <30, 30–<40, ≥ 40 years) and time since surgery (1-<5, 5-<10, 10-<15, 15-<20, 20-<25, ≥25 years). Attained age, follow-up interval and time since surgery were time-dependent variables because women would contribute person-years to different categories within these variables as they were followed over time. In contrast, women would contribute person- years to only one level of the classification variables period of surgery and age at surgery. The DATAB module in the Epicure software program was used to calculate person-years of follow-up (28).

The expected numbers of incident cancers in the cohort and the other cosmetic surgeries group were estimated by multiplying the tabulated person-years of follow-up by the

59 corresponding overall and site-specific cancer rate observed in the general population according to province (Ontario or Quebec), age (by 5-year age intervals), and calendar period of follow-up (1974–1977, 1978–1981, . . ., 1994–1997, 1998-2001, 2002-2007). Differences in cancer incidence rates between the implant and surgical control cohorts relative to the general population were evaluated by calculating the standardized incidence ratio (SIR), which is the ratio of the observed-to-expected number of incident cancers (29). For the comparison with general population estimates, person-years contributed for the period after 1998 were reduced by an interprovincial migration rate according to province, attained age and calendar period of follow-up on the basis of migration rates observed through active follow-up of the Canadian population (30). This was done to account for interprovincial mobility. This approach has been previously applied to reduce the impact of losses to follow-up in a cohort study (31). The 95 percent confidence intervals were calculated for the SIR by assuming that the observed number of incident cancers followed a Poisson distribution, using formulae detailed elsewhere (29). All the p values reported are two sided.

Comparisons of site-specific incident cancer rates between the implant recipients and the other plastic surgery patients, rather than the general population, were done using multivariate Poisson regression models using incidence rate ratios (IRR) as the measure of association (32). We used Cox proportional hazards regression models to evaluate cumulative incidence of breast cancer over the follow-up period (33). The potential confounding influence of the following factors was evaluated: linear and quadratic attained age components, province of residence, calendar period of follow-up, age at surgery, year of surgery and time since surgery. Confounding was examined by a backward deletion approach (34). Specifically, we first adjusted for all potential confounders and then removed one by one in a stepwise manner the least significant confounding variables until the total proportional change in incidence rate ratio estimates compared with those of the fully adjusted model was less than 10 %. Covariates that were not confounders, but increased the precision of the estimates were kept in the final model. To evaluate whether the incidence rate ratio differed by province, a test of homogeneity was conducted by including in the Poisson regression model a first-order interaction term of province and implant status. The 2

60 provinces were deemed to have different risk estimates if the interaction term was found to be statistically significant based on a two-tailed alpha of <5%. P values for trend of incidence rate ratio over time since surgery were computed, where applicable, using the median time since surgery value for each category as a continuous variable. We included in the regression model a first-order interaction term of this continuous time since surgery variable and the main exposure variable of interest. There was a trend of increasing (or decreasing) incidence rate ratio if the interaction term was found to be statistically significant based on a two-tailed alpha of <5%. For instance, if the main exposure variable is study group (implant vs. controls), a positive and statistically significant interaction term indicates that the incidence rate ratio comparing implant women to controls increases with time since surgery.

Analyses including only women who received breast implants were performed using multivariate Poisson regression models to assess associations of implant characteristics to breast cancer incidence. The following implant characteristics were evaluated: type of implant (silicone gel-filled implants (SGFIs) or and saline), envelope (polyurethane-coated or not), subglandular or submuscular placement, and fill volume. For implant fill volume, women were categorized based on the quartiles of the observed frequency distribution of the mean value of the right and left implants (<175, 175–<200, 200–<225 and ≥ 225 cc) (7). Confounding, trend and interaction were assessed with the same approaches mentioned above. Analyses were done with SAS, version 9.2 (35).

Results

A total of 581,331 and 374,996 person–years of follow-up were accrued in the breast implant (n=24,558) and the other cosmetic surgery (n=15,893) cohorts, respectively (Table 1). The total amount of person-years accrued when the interprovincial migration correction was applied reached 577,257 and 372,532 respectively for the implant group and the other plastic surgery patients (data not shown). The mean duration of follow-up was about the same in the two study groups; 23.7 years for the implant cohort and 23.6 years for the control cohort. Specifically, more than 70% of the women in both the breast implant and other cosmetic surgery cohort were followed for at least 20 years. As reported in our

61 previous publication, most of women in the breast implant cohort (65.6%) received implants filled with silicone gel (7). The site of implantation was more frequently submuscular (56%) than subglandular (32.6%) (7). Few recipients received implants with a polyurethane foam covered envelope (10.5 %); of those who did, most came from the province of Quebec (7).

A total of 1,521 and 1,220 incident cancers were identified among implant women and other cosmetic surgery women, respectively (Table 2). Comparisons with general female population estimates showed that the observed number for cancers of all sites was significantly lower than the expected number in both the implant cohort (SIR = 0.71, 95% CI = 0.67-0.75) and other cosmetic surgery cohort (SIR = 0.79, 95% CI = 0.74-0.83). Statistically significant reductions in rate of breast cancer were observed in both the implant women (SIR = 0.54, 95% CI= 0.49-0.59) and the control group (SIR = 0.88, 95% CI = 0.80-0.96). As well, significantly lower than expected rates for stomach, colorectal, endometrial, ovary, lymphohematopoietic cancers and all other cancer sites combined were observed among implant women compared to the general female population. There were also reduced risks of colorectal, endometrial, lymphohematopoietic cancers and all other cancer sites combined for the other cosmetic surgery cohort relative to general female population estimates.

Internal comparisons revealed that compared with other cosmetic surgery women, those with breast implants had significantly reduced rates for cancers of all sites (IRR = 0.88, 95% CI= 0.82-0.95), breast (IRR = 0.60, 95% CI= 0.53-0.69), overall genital (IRR = 0.77, 95% CI= 0.63-0.95) and endometrial (IRR = 0.55, 95% CI= 0.38-0.78) cancers (Table 3). However, when removing breast and endometrial cancers from all sites combined, there were little or no differences between implant women and other cosmetic surgery women for overall cancer incidence (IRR = 1.08, 95% CI= 0.94-1.23) (data not shown). As well, when removing endometrial cancers from overall genital cancers, little or no differences were seen between implant women and those with other cosmetic surgeries for overall genital cancer incidence (IRR = 0.92, 95% CI= 0.70-1.20) (data not shown). Breast cancer cumulative incidence is shown in figure 1. After 30 years of follow-up, breast cancer risk

62 for the implant patients reached 2.3 % compared to 3.7 % for other plastic surgery women. Moreover, incidence rate ratios for different lengths of follow-up remained steadily around 0.60 (p value for trend in IRR over time since surgery = 0.95) (data shown in footnote of figure 1).

Table 4 presents breast cancer incidence among implant women according to specific implant characteristics. Results show that women whose implants were inserted in the subglandular position had a significantly reduced rate of breast cancer compared to those whose implants were inserted submuscularly (IRR = 0.78, 95% CI= 0.63-0.96). As well, women who received polyurethane coated implants had a non-statistically significant elevated IRR of 1.22 (95% CI= 0.84-1.77) for breast cancer when compared with implant women without such coating. There was no statistically significant difference in breast cancer rates for type of implant and fill volume. Results did not change when we mutually adjusted for implant characteristics in the multivariate models (results not shown).

We further investigated the pattern of breast cancer risk for subglandular implants relative to submuscular implants over a long period of time since surgery. This analysis revealed incidence rate ratios of 0.68 (0.31 – 1.51) for 1 to <5 years after surgery, 1.03 (0.58 – 1.83) for 5 to <10 years, 0.64 (0.42 – 0.98) for 10 to <15 years and 0.80 (0.61 – 1.06) for over 15 years after surgery (p value for trend in IRR over time since surgery = 0.86). This suggests that the reduction in breast cancer incidence among subglandular implants relative to those with submuscular implants can be observed many years after receiving the surgery.

Further analyses for polyurethane-coated implants in the subglandular position relative to women who received other subglandular implants, by time since surgery, was undertaken (Figure 2). The results indicate a statistically significant decreasing monotonic trend in the incidence rate ratio according to time since surgery (p value for trend in IRR over time since surgery = 0.02). Specifically, the IRR of breast cancer incidence among those with subglandular polyurethane-coated implants compared to other women with subglandular implants decreased from 7.36 (95% CI= 1.86-29.12) for 1 to 5 years after surgery to 0.69 (95% CI = 0.29–1.60) for follow-up of more than 15 years after surgery. An analysis for

63 polyurethane-coated implants in the submuscular position was not possible because only 132 women received such devices and consequently few breast cancer cases were observed among these women.

Discussion

In this extended follow-up of Canadian women with cosmetic breast implants, with at least 20 years of follow-up for more than 70% of cohort members, we observed that breast cancer incidence among implant women continued to be lower than the other plastic surgery cohort over this long period of follow-up. Our extended analysis also allowed us to more confidently report the absence of increased risk of rarer forms of cancer among women with cosmetic breast implants. Moreover, we observed a reduced rate of breast cancer for women with subglandular implants relative to women with submuscular implants that persisted over a long period of follow-up. Finally, we observed a 7-fold increased rate of breast cancer soon after the index date of surgery for women with polyurethane covered subglandular implants that decreased progressively over follow-up.

Findings were somewhat different when the implant cohort was compared to the general female population rather than women with other cosmetic surgeries. The observed reduction in rates for overall cancers in the implant group compared with general female population estimates is consistent with our previous work (7) and previous investigations (10;16;17;36). However, similar patterns of reduced rates for colorectal, overall genital, lymphohematopoietic cancers and all other cancer sites combined were also seen for the control group compared with women in the general population. Thus, some reductions in rates for site-specific cancers seen in the implant cohort the other cosmetic surgery women may be due to the fact that augmented women have different risk factor profile for cancer than the general female population, including the fact that they are more likely to be white and of higher socioeconomic status (37). Therefore, the observed reductions compared with the general population may be overestimated. This highlights the importance of using a comparable control population, such as women with other cosmetic surgeries who have similarities with breast implant women in terms of sociodemographic and lifestyle characteristics (38), and not the general population as a reference group for these analyses.

Although there is a concern for a possible link between breast implants and anaplastic large T- cell lymphomas of the breast (40;41), we could not confirm this association because no diagnoses of anaplastic large T-cell lymphomas (ICD-9: 200.6) occurred throughout the follow-up in both the implant subjects and the other cosmetic surgery group (Result not shown).

Women with cosmetic breast implants have lower rates of overall cancers and reduced rates of breast, overall genital and endometrial cancers relative to other cosmetic surgery women. However, the observed reduction in rates for all cancer sites combined and overall genital cancers were explained by the reduced rates of breast and endometrial cancers respectively. Our study confirmed the findings of our previous report and other publications of a reduced rate of breast cancer for implant women compared with other surgery women (1;7;14). Additionally, we have showed that breast cancer incidence among augmented women remained lower over a long period of time compared to those with other surgeries. Although studies using other cosmetic surgery women as a control group have supported the notion that these patients are a more appropriate comparison group for augmented women (39), there are still important differences between the two groups. For example, implant patients are more likely than other plastic surgery patients to be white, have earlier age at first birth and be thin (39). Furthermore, women with a family history of breast cancer may have elected not to receive breast implants, as these devices may interfere with the detection of breast cancer (40). Therefore, observed differences in breast cancer incidence rates between these two groups may be explained partly by differences in risk factors. In fact, this argument can be supported by the fact that we have found similar patterns of risk for breast and endometrial cancers, two cancers that share many of the same risk factors (41).

Possible biological mechanisms have been suggested in the literature to explain the reduced breast cancer risk among augmented women. It has been suggested that the presence of breast implants could enhance the immune system, whereby carcinogens and transformed cells would be more easily destroyed (10). Further, the weight and volume of breast implants may compress the glandular tissue resulting in a decreased blood supply that may reduce the rate of cell proliferation (10). However, the decreased risk of breast cancer

65 among augmented women might also be due to smaller native breasts (and thus less breast tissue) prior to augmentation which could make these women less likely to develop breast cancer (2). Others argue that the exclusion of women with prevalent tumors as a result of the pre-surgery screening examination may also be a possible mechanism (15). However, this seems unlikely as we have observed persistently lower breast cancer risk over a long period of time after surgery among augmented women relative to the other surgery group.

Breast implants can be placed in the subglandular position, which is on top of the pectoralis muscle and directly under the breast glands, or in the submuscular position, which is under the pectoralis major muscle (38). Our analysis according to specific implant characteristics has shown, for the first time, a statistically significant decrease in breast cancer incidence for subglandular breast implants compared with submuscular implants, confirming the non- significant pattern we reported in our previous follow-up (7). As well, the rates of breast cancer incidence for women with subglandular implants remained lower even over a long period of follow-up. This result may be attributable in part to variation in one of our previous explanations of possible effects of breast implants on the immune system and blood flow in the breast gland (10). This needs to be further studied.

In our previous publication, we had identified the 7-fold increase in breast cancer rate shortly after insertion of subglandular polyurethane coated implants, but the insufficient amount of follow-up time limited the evaluation of a possible trend according to time since surgery (7). Scientific literature has shown that a polyurethane envelope begins to biodegrade approximately 2 years after augmentation surgery has been performed (42). The biodegradation of the polyurethane envelope by body fluids will result in the break-down product 2,4-toluenediamine (TDA) (12) which is recognized as an animal carcinogen and a potential human carcinogen (12). A possible explanation of our finding is that the biodegradation product of polyurethane could act as a tumor promoter. In fact, 2,4- toluenediamine (TDA) has been previously shown to stimulate hepatic cellular proliferation and promote mutated cells in rats (43;44). Therefore, this could explain the sudden increase in breast cancer rate that coincides with the time when the polyurethane envelope biodegrades. In fact, early investigations reported that the polyurethane foam biodegrades

66 rapidly after implantation (45;46). However, subsequent analyses suggested that polyurethane foam implants biodegrades through a slow process (47) and that large amounts of unbroken polyurethane foam still remains 9 years after implantation (48). Therefore, we believe further assessments on the rate of biodegradation of polyurethane among augmented women are needed in order to clarify our findings. Additionally, to our knowledge, no epidemiological study has been able to provide any confirmation of a tumor promotion effect of TDA. Furthermore, two occupational studies of workers exposed to polyurethane over long periods did not show any increase in cancers of any type (49;50). Finally, our estimates were based only on a small number of incident cases which increases the possibility that the observed results could be due to chance. However, given the large increase in the incidence rate ratio observed shortly after implantation and the fact that polyurethane coated implants are still in use, it is critical to pursue investigations in order to clarify the potential tumor promotion effect of TDA.

Some limitations of our study need to be acknowledged. For instance, no information was available throughout the follow-up if augmented women had their implants removed or if women with other cosmetic surgery had breast implants following their initial procedure. This misclassification bias would only lead to an underestimation of the measures of association. Residential mobility and the resulting loss to follow-up may contribute to lower-than-expected incidence rates among the breast implant and other plastic surgery women when compared with general female population estimates. However, losses to follow-up were minimized with respect to mobility by linking the cohort members to national cancer and mortality databases for the follow-up prior to 1998. Additionally, we accounted for interprovincial migration for the second phase of this study. In fact, by applying the correction for interprovincial migration, the total amount of person-years for the implant and other surgery women combined went from 956,327 to 949,789. Thus, this correction for interprovincial migration had little impact on expected numbers of cases and the resulting incidence rate ratio estimates in the comparison of implant cohort to the general population were more conservative. It is possible that some cancers and deaths may have been missed, especially where data were linked to provincial registries because, as time passes, more people are expected to move and be diagnosed in other provinces or

67 other countries. For this reason, we evaluated whether there were differences in completeness of ascertainment of cancer cases and deaths comparing ascertainment using national registries (as in our initial follow-up (7)) to that using provincial registries (as in our extended follow-up). The two approaches could be compared because they were used independently for a common period which extended from 1995 to 1997 for both the Quebec and Ontario cohorts. The assessment of possible differences in completeness of ascertainment revealed that less than 7% of cancer cases identified in the national linkage could not be identified in the provincial linkages and less than 4% of cancer cases were missing in the national linkage compared with the provincial linkages. Similarly, 13% of deaths were missing in the provincial linkage compared with the national linkage and less than 6% of deaths were missing in the national linkage compared with the provincial linkage. Overall, this results in a net missingness of cancer cases of 3% and a net missingness of deaths of 7%. It should be noted that the comparison of implant and control cohorts should not be materially affected by interprovincial migration or limits in linkages because these factors are expected to be comparable for the two cohorts. Therefore, we know of no reason why incomplete ascertainment would be differential between the implant subjects and other cosmetic surgery women, a condition necessary to bias risk estimates generated from the internal comparison.

Strengths of our study include the largest sample size to date, the cohort design, the long follow-up period, the detailed information on implant characteristics and the fact that we used both other cosmetic surgery women and general female population estimates as the comparison groups. Moreover, we excluded the first year after surgery in the follow-up of this cohort, which is consistent with approaches undertaken by our previous study (7) and other investigations (39).

In conclusion, this study found significant decreases in incidence rates for breast and endometrial cancers among augmented women compared to other cosmetic surgery women and these reductions persisted for more than 20 years after surgery. No increased incidence of rarer forms of cancers, including hematopoietic cancers, was seen among augmented women compared with other plastic surgery women. As well, for the first time, a persistent

68 statistically significant reduction in breast cancer incidence was observed for women with subglandular implants relative to women with submuscular implants. Finally, our study shows that women with subglandular polyurethane covered implants may have an increase in breast cancer rate for the first few years after breast augmentation surgery that decreases with increasing follow-up, suggesting a possible tumor promotion effect on the breast tissue by the biodegradation products of polyurethane. Additional information regarding the possible increase of breast cancer incidence among polyurethane augmented women shortly after surgery is needed.

Acknowledgements

This study was supported by the Public Health Agency of Canada. We thank Dr. Yang Mao for extensive contribution to this project. We also thank the plastic surgeons in Ontario and Quebec for their participation. We also thank Sylvie Bérubé and Caty Blanchette from Quebec and Gemma Lee and Susitha Wanigaratne from Ontario, for their help in the design and conduct of this updated study. We also thank Drs Robert Dales and Scott Weichenthal from Health Canada for their comments on an earlier version of this manuscript and Mr. Robert Semenciw for helping in acquiring general female population cancer and mortality rates.

References

(1) Mclaughlin JK, Lipworth L, Murphy DK, Walker PS. The safety of silicone gel- filled breast implants. A review of the epidemiologic evidence. Ann Plast Surg 2007;59(5):569-80.

(2) Deapen DM. Breast implants and breast cancer: a review of incidence, detection, mortality, and survival. Plast Reconstr Surg 2007;120(Suppl. 1):70S-80S.

(3) U.S.Food and Drug Administration Center for Devices and Radiological Health. FDA Update on the Safety of Silicone Gel-Filled Breast Implants. 2011.

(4) Lilla JA, Vistnes LM. Long-term study of reactions to various silicone breast implants in rabbits. Plast Reconstr Surg 1976;57(5):637-49.

(5) Health Canada. Drugs and health products: Notice of decision for silicone gel- filled breast implants. Health Canada 2009. Available from: URL: http://www.hc- sc.gc.ca/dhp-mps/prodpharma/sbd-smd/phase1-decision/md im/nd_ad_2006_lowbleed_69312-end.php. 2009.

(6) Friis S, Holmich LR, Mclaughlin JK, Kjøller K, Fryzek JP, Henriksen TF, Olsen JH. Cancer risk among Danish women with cosmetic breast implants. Int J Cancer 2006;118(4):998-1003.

(7) Brisson J, Holowaty EJ, Villeneuve PJ, Xie L, Ugnat A-M, Latulippe L, Mao Y. Cancer incidence in a cohort of Ontario and Quebec women having bilateral breast augmentation. Int J Cancer 2005;118:2854-62.

(8) Mellemkjaer L, Kjoller K, Friis S, Mclaughlin JK, Høgsted C, Winther JF, Breiting V, Krag C, Krüger Kjaer S, Blot WJ, Olsen JH. Cancer occurence after cosmetic breast implantation in Denmark. Int J Cancer 2000;88:301-6.

(9) Deapen DM, Brody GS. Augmentation mammaplasty and breast cancer: a 5-year update of the Los Angeles study. Plast Reconstr Surg 1992;89(4):660-5.

(10) Deapen DM, Bernstein L, Brody GS. Are breast implants anticarcinogenic ? A 14-year follow-up of the Los Angeles study. Plast Reconstr Surg 1997;99(5):1346-53.

(11) Cohney BC, Cohney TB, Hearne VA. Augmentation mammaplasty - a further review of 20 years using the polyurethane-covered prosthesis. J Long Term Eff Med Implants 1992;1(3):269-79.

(12) Expert Panel on the Safety of Polyurethane-covered Breast Implants. Safety of polyurethane-covered breast implants. CMAJ 1991;145(9):1125-32. 70

(13) Lipworth L, Tarone RE, Friis S, Ye W, Olsen JH, Nyren O, Mclaughlin JK. Cancer among scandinavian women with cosmetic breast implants: A pooled long-term follow-up study. Int J Cancer 2009;124:490-3.

(14) Deapen DM, Hirsch EM, Brody GS. Cancer risk among Los Angeles women with cosmetic breast implants. Plast Reconstr Surg 2007;119:1987-92.

(15) Mclaughlin JK, Lipworth L, Fryzek JP, Weimin Y, Tarone RE, Nyren O. Long- term cancer risk among Swedish women with cosmetic breast implants: An update of a nationwide study. J Natl Cancer Inst 2006;98(8):557-60.

(16) Pukkala E, Boice Jr.JD, Hovi S-L, Hemminki E, Asko-Seljavaara S, Keskimäki I, Mclaughlin JK, Pakkanen M, Teppo L. Incidence of breast cancer and other cancers among Finnish women with cosmetic breast implants, 1970-1999. J Long Term Eff Med Implants 2002;12(4):251-3.

(17) Brinton LA, Lubin J, Burich MC, Colton T, Brown SL, Hoover RN. Breast cancer following augmentation mammoplasty (United States). Cancer Causes Control 2000;11:819-27.

(18) Brinton LA. The relationship of silicone breast implants and cancer at other sites. Plast Reconstr Surg 2007;120(Suppl. 1):94S-102S.

(19) Mclaughlin JK, Nyren O, Blot WJ, Yin L, Josefsson S, Fraumeni JF Jr, Adami HO. Cancer risk among women with cosmetic breast implants: a population based cohort study in Sweden. J Natl Cancer Inst 1998;90(2):156-8.

(20) Lipworth L, Tarone RE, Mclaughlin JK. Breast implants and lymphoma risk: A review of the epidemiologic evidence through 2008. Plast Reconstr Surg 2009;123(3):790-3.

(21) Jewell M, Spear SL, Largent J, Oefelein MG, Adams WP. Anaplastic large T-cell lymphoma and breast implants: a review of the literature. Plast Reconstr Surg 2011;128(3):651-61.

(22) U.S.Food and Drug Administration. Anaplastic Large Cell Lymphoma (ALCL) In Women with Breast Implants: Preliminary FDA Findings and Analyses. Center for Devices and Radiological Health; 2011.

(23) Band PR, Gaudette LA, Hill GB, Holowaty EJ, Huchcroft SA, Johnston GM, Illing EMM, Mao Y, Semenciw RM. The making of the Canadian cancer registry: cancer incidence in Canada and its regions, 1969 to 1988. Ottawa, ON; 1993.

(24) Howe GR, Lindsay J. A generalized iterative record linkage computer system for use in medical follow-up studies. Comput Biomed Res 1981;14(4):327-40.

(25) Brenner DR, Tammemägi MC, Bull SB, Pinnaduwaje D, Andrulis IL. Using cancer registry data: agreement in cause-of-death data between the Ontario Cancer Registry and a longitudinal study of breast cancer patients. Chronic Dis Can 2009;30(1):16-9.

(26) Ministère de la Santé et des Services Sociaux du Québec. Fichier des tumeurs du Québec. Ministère de la Santé et des Services Sociaux du Québec 2010. Available from: URL: http://msssa4.msss.gouv.qc.ca/santpub/tumeurs.nsf/cat?OpenView. 2010.

(27) Matchware Technologies I. Automatch generalised record linkage system, version 4.2: user's manual. Kennebunk, Maine: MatchWare Technologies, Inc.; 1998.

(28) EPICURE: risk regression and data analysis software [computer program]. Seattle: Hirosoft International Corporation; 1996.

(29) Breslow NE, Day NE. Statistical methods in cancer research. Vol. II, The design and analysis of cohort studies. Lyon, France: International Agency for Research on Cancer, 1987.

(30) Bernard A, Finnie R, St-Jean B. Interprovincial mobility and earnings. Ottawa: Statistics Canada; 2008.

(31) Phillips N, Coldman A. Comparison of nonbreast cancer incidence, survival and mortality between breast screening program participants and nonparticipants. Int J Cancer 2007;122(1):197-201.

(32) Kleinbaum DG, Kupper LL, Muller KE, Nizam A. Applied regression analysis and other multivariable methods, 3rd ed Brooks/Cole Publishing Company, 1998.

(33) Kleinbaum DG, Klein M. Survival analysis: a self leaning text, 2 ed. Atlanta, GA: Springer, 2005.

(34) Rothman KJ, Greenland S, Lash TL. Modern epidemiology, 3rd ed. Philadelphia: Lippincott, Williams & Wilkins, 2008.

(35) SAS, ed 9.2 [computer program]. Cary, NC, USA: SAS Institute; 2008.

(36) Kern KA, Flannery JT, Kuehn PG. Carcinogenic potential of silicone breast implants: a Connecticut statewide study. Plast Reconstr Surg 1997;100(3):737-47.

(37) Kjoller K, Holmich LR, Fryzek JP, Jacobsen PH, Friis S, McLaughlinJK, Lipworth L, Henriksen TF, Jorgensen S, Bittmann S, Olsen JH. Characteristics of women with cosmetic breast implants compared with women with other types of cosmetic surgery and populationbased controls in Denmark. Ann Plast Surg 2003;50:6-12.

(38) Gladfelter J. Breast augmentation 101: Understanding cosmetic breast mammaplasty. Plastic Surgical Nursing 2007;27(3):136-45.

(39) Brinton LA, Brown SL, Colton T, Burich MC, Lubin J. Characteristics of a population of women with breast implants compared with women seeking other types of plastic surgery. Plast Reconstr Surg 2000;105(3):919-29.

(40) Hayes HJr, Vandergrift J, Diner WC. Mammography and breast implants. Plast Reconstr Surg 1988;82(1):1-8.

(41) Healey CS, Ahmed S, ANECS, AOCS Management Group, O'Mara TA, Ferguson K, Lambrechts D, Garcia-Dios DA, Vergote I, Amant F, NSECG, Howarth K, et al. Breast Cancer Susceptibility Polymorphisms and Endometrial Cancer Risk: a Collaborative Endometrial Cancer Study. Carcinogenis 2011;32(12):1862-6.

(42) Vazquez G, Pellon A. Polyurethane-Coated Silicone Gel Breast Implants Used for 18 Years. Aesth Plast Surg 2007;31:330-6.

(43) Shanmugam K, Subrahmanyam S, Tarakad SV, Kodandapani N, Stanly DF. 2,4- Toluene Diamines-Their carcinogenicity, biodegradation, analytical techniques and an approach towards development of biosensors. Anal Sci 2001;17(12):1369- 74.

(44) Cunningham ML, Foley J, Maronpot RR, Matthews HB. Correlation of hepatocellular proliferation with hepatocarcinogenicity induced by the mutagenic noncarcinogen:carcinogen pair--2,6- and 2,4-diaminotoluene. Toxicol Appl Pharmacol 1991;107(3):562-7.

(45) Smahel J. Tissue reaction to breast implants coated with polyurethane. Plast Reconstr Surg 1978;61:80-5.

(46) Slade LC, Peterson HD. Disappearance of the polyurethane cover of the Ashley Natural Y prosthesis. Plast Reconstr Surg 1982;70:379-83.

(47) Hester TR. The polyurethane-covered mammary prosthesis: facts and fiction. Persp Plast Surg 1988;2:135-69.

(48) Szycher M, Siciliano AA. Polyurethane-covered mammary prosthesis: a nine-year follow-up assessment. J Biomaterials App 1991;5:282-322.

(49) Sorahan T, Pope D. Mortality and cancer morbidity of production workers in the UK flexible polyurethane foam industry. Br J Indus Med 1993;50:528.

(50) Hagmar L, Welinder H, Mikoczy Z. Cancer incidence and mortality in the Swedish polyurethane foam manufacturing industry. Br J Indus Med 1993;50:537.

Table 1. Frequency distribution for selected characteristics of women who received breast implants and women who received other cosmetic surgeries, Canadian Breast Implant Cohort Study.

Characteristics Implant patients Control patients Length of follow-up (years), N (%) <1 28 (0.1) 21 (0.1) 1 to <5 173 (0.7) 138 (0.8) 5 to <10 267 (1.1) 227 (1.4) 10 to <15 400 (1.6) 342 (2.2) 15 to <20 5,356 (21.8) 3,016 (19.0) 20 to <25 8,359 (34.0) 5,834 (36.7) ≥25 9,975 (40.6) 6,315 (39.7) Year of surgery, N (%) 1974 - 1977 4,726 (19.2) 3,011 (19.0) 1978 - 1981 5,750 (23.4) 3,766 (23.7) 1982 - 1985 6,685 (27.2) 4,706 (29.6) 1986 - 1989 7,397 (30.1) 4,410 (27.7) Age at surgery (years), N (%) 18 to <25 3,665 (14.9) 3,481 (21.9) 25 to <30 5,961 (24.3) 3,064 (19.3) 30 to <35 6,868 (28.0) 2,828 (17.8) 35 to <40 4,195 (17.1) 2,357 (14.8) 40 to <45 2,068 (8.4) 1,562 (9.8) ≥45 1,801 (7.3) 2,601 (16.4)

Mean age at surgery (SD), years 32.2 (7.8) 33.5 (10.4) Mean duration of follow up (range), years 23.7 (0.1–34.0) 23.6 (0.0–33.9) Total person-years of follow up1 581,331 374,996 Total number of women 24,558 15,893 1Person-years were accrued from the date of surgery until the earliest date of cancer diagnosis, death, December 31, 2006 (Quebec) or December 31, 2007 (Ontario).

Table 2. Standardized incidence ratios (SIRs)1 for selected cancers based on general population cancer incidence rates (1974-2007) among breast implant and other cosmetic surgery women with comparisons with previous follow-up2.

Implant patients Control patients O bs. cases SIR SIR O bs. cases SIR SIR Cancer site ICD-9 1974 – 2007 1974 – 1997 1974 – 2007 1974 – 1997 Follow-up Follow-up Follow-up Follow-up interval interval interval interval All sites 140-208 (excl. 173) 1,521 0.71* 0.75* 1,220 0.79* 0.81* Stomach 151 12 0.47* 0.68 18 0.85 0.79 Colorectal 153-154 151 0.75* 0.79 107 0.66* 0.61* Pancreas 157 33 0.94 1.22 22 0.74 1.33 Lung 162.2-5, .8, .9 271 1.04 1.09 167 0.85* 1.11 Malignant melanoma 172 56 1.08 1.29 28 0.84 0.79 Breast 174 414 0.54* 0.57* 457 0.88* 0.64* Genital 179-184 195 0.64* 0.78* 173 0.81* 0.87* Cervix 180 61 0.83 0.96 40 0.85 0.80* Endometrial 182 52 0.44* 0.53* 71 0.82 0.91 Ovary 183.0 66 0.76* 0.80 49 0.79 0.70 Bladder 188 33 0.86 0.88 18 0.59* 0.64 Kidney 189 32 0.72 0.71 30 0.90 0.74 Nervous system 191, 192 27 0.83 0.65 27 1.14 0.88 Brain 191 25 0.73 0.65 26 1.06 0.80 Thyroid 193 61 0.84 0.73 31 0.68* 0.42* Lymphohematopoietic 200-208 106 0.75* 0.69* 79 0.74* 0.68* Non-Hodgkin’s lymphoma 200, 202 63 0.84 0.75 45 0.81 0.78 Leukemia 204-206, 207.0, .2, .8, 208 27 0.75 0.68 16 0.57* 0.66 Other cancer sites not listed above 130 0.75* 0.87 63 0.47* 0.60* 1The SIR is the ratio of the observed to expected cases; the expected number of cases was estimated by applying age, period and province (Ontario or Quebec) specific cancer incidence rates to the corresponding number of person-years of follow-up observed in the cohort. 2Previous follow-up interval standardized incidence ratios (SIRs) for selected cancers based on general population cancer incidence rates (1974-1997). *Denotes a statistically significant difference based on a two-tailed alpha of less than 5 %.

Table 3. Incidence rate ratios (IRRs)1 for selected cancers between breast implant and other cosmetic surgery women with comparisons with previous analysis2. Internal comparison Internal comparison Cancer site ICD-9 1974 – 2007 Follow-up interval 1974 – 1997 Follow-up interval IRR 95 % CI IRR 95 % CI All sites 140-208 (excl. 173) 0.88 0.82-0.95 0.91 0.81-1.02 Stomach 151 0.54 0.26-1.14 0.93 0.34-2.52 Colorectal 153-154 1.14 0.88-1.46 1.22 0.81-1.84 Pancreas 157 1.27 0.73-2.20 0.94 0.45-1.95 Lung 162.2-5, .8, .9 1.18 0.97-1.44 0.93 0.69-1.26 Malignant melanoma 172 1.35 0.85-2.13 1.69 0.88-3.23 Breast 174 0.60 0.53-0.69 0.64 0.53-0.79 Genital 179-184 0.77 0.63-0.95 0.93 0.70-1.24 Cervix 180 0.94 0.63-1.40 1.00 0.62-1.61 Endometrial 182 0.55 0.38-0.78 0.63 0.37-1.09 Ovary 183.0 0.95 0.65-1.38 1.11 0.64-1.91 Bladder 188 1.54 0.86-2.76 1.37 0.56-3.35 Kidney 189 0.79 0.48-1.31 1.02 0.43-2.39 Nervous system 191, 192 0.69 0.40-1.17 0.66 0.28-1.54 Brain 191 0.67 0.38-1.16 0.74 0.31-1.75 Thyroid 193 1.23 0.80-1.90 1.66 0.80-3.46 Lymphohematopoietic 200-208 1.02 0.76-1.36 0.97 0.61-1.54 Non-Hodgkin’s lymphoma 200, 202 1.03 0.70-1.52 0.97 0.53-1.76 Leukemia 204-206, 207.0, .2, .8, 208 1.34 0.72-2.51 0.94 0.39-2.25 Other cancer sites not listed above 1.30 0.76-2.22 1.35 0.89-2.04 1The IRRs estimates were derived using Poisson multivariate regression model and were adjusted for attained age, calendar period and province of residence. 2Previous analysis incidence rate ratios (IRRs) for selected cancers (1974-1998).

Table 4. Incidence rate ratios (IRRs)1 and 95% confidence intervals (CIs) of breast cancer for selected breast implant characteristics among implant women.

Implant characteristics Person-years Cases IRR 95% CI

Type of fill Silicone 356,975 284 1.0 - Saline 20,489 11 0.74 0.40-1.37 Saline and silicone 100,944 68 0.91 0.66-1.25 Unknown 78,393 51 0.74 0.56-0.96 Polyurethane coating No 361,817 260 1.0 - Yes 45,377 35 1.22 0.84-1.77 Unknown 149,607 119 0.99 0.78-1.24 Fill volume (cc) <175 147,957 95 1.0 - 175 to < 200 138,256 108 1.21 0.92-1.59 200 to <225 143,413 121 1.31 1.00-1.72 ≥225 123,026 88 1.18 0.88-1.60 Unknown 4149 2 0.66 0.16-2.68 Site of implantation Submuscular 178,446 154 1.0 - Subglandular 308,488 211 0.78 0.63-0.96 Unknown 66,263 48 0.74 0.54-1.03 1Incidence rate ratios estimates were adjusted for attained age, calendar period and province of residence.

Figure 1. Cumulative breast cancer incidence curves1 for time since index surgery comparing breast implant with other cosmetic

surgery women.

Implant women

Other cosmetic surgery women Cumulative (%) Cumulative incidence

Time since surgery (Years)

1Cumulative incidence curves were adjusted for attained age, calendar period and province of residence using Cox proportional hazards model. 2Incidence rate ratios and respective confidence intervals for different length of follow-up after index date of surgery: 1 to 5 years, 0.61 (0.38 – 0.99), 5 to 10 years, 0.57 (0.41 – 0.81), 10 to 15 years, 0.69 (0.52 – 0.91), 15 to 20 years, 0.58 (0.44 – 0.76) and ≥ 20 years, 0.61 (0.48 – 0.78). 3P value for trend in IRR over time since surgery = 0.95.

Figure 2. Incidence rate ratios1 (IRRs) and 95% confidence intervals2 (CIs) to evaluate the trend3 in breast cancer risk for women who received subglandular polyurethane coated breast implants relative to other women who received subglandular implants, by time since surgery.

10.00

9.00

8.00 7.36 7.00

6.00

5.00 IRR and 95 and % CI IRR 4.00

3.00

2.00 1.37 1.27 1.00 0.69 0.00 1 to 5 5 to 10 10 to 15 ≥ 15

Time since surgery (years)

1Incidence rate ratios estimates were adjusted for attained age, calendar period and province of residence. 2Incidence rate ratios and respective confidence intervals: 7.36 (1.86 – 29.12), 1.37 (0.55 – 3.45), 1.27 (0.60 – 2.67) and 0.69 (0.29 – 1.60). 3P value for trend in IRR over time since surgery = 0.02.

Chapter 3: Cosmetic breast augmentation and mortality: an update of a Canadian cohort

Résumé

Objectifs : Il a été démontré que les femmes ayant des implants mammaires pour fins esthétiques ont des taux élevés de suicide, mais le risque de suicide selon le temps après la chirurgie et selon l’âge à la chirurgie nécessite d’être clarifié. De plus, plusieurs autres causes de décès parmi ces femmes doivent être clarifiées. Le but de cette étude était de combler ces lacunes. Méthodes : Cette étude présente une analyse de 10 ans de plus de suivi d’une grande étude de cohorte de femmes ayant reçu soit des implants mammaires pour fins esthétiques (n=24 558) ou une autre chirurgie esthétique (15 893). Plus de 70% de la cohorte a été suivi pour plus de 20 ans. La mortalité chez les femmes avec des implants mammaires a été comparée au groupe contrôle en utilisant des modèles de Poisson multivariés (Rapport de Taux de mortalité (RT)) ainsi qu’aux femmes de la population générale en utilisant le Rapport Standardisé de Mortalité (RSM). Résultats : Des taux plus élevés de suicide ont été observés chez les femmes avec une augmentation mammaire comparativement aux femmes de la population générale (RSM = 2,00, IC à 95% = 1,66-2,41) ainsi qu’aux femmes avec une autre chirurgie esthétique (RT de mortalité = 1,43, IC à 95% = 1,02-1,99). Nos résultats suggèrent une tendance croissante du RT de mortalité selon le temps après la chirurgie. Les femmes ayant reçu une augmentation mammaire en bas âge avaient le plus haut RT de mortalité de suicide (RT de mortalité = 1,92, IC à 95% = 1,16-3,16). Aucune différence de risque de décès de plusieurs autres causes n’a été observée lorsque les femmes avec implants mammaires ont été comparées aux femmes avec une autre chirurgie esthétique. Conclusion : Les taux de suicide chez les femmes avec une augmentation mammaire demeurent élevés 20 ans ou plus après la chirurgie. Davantage d’investigations sont nécessaires afin d’élucider l’augmentation possible du risque de suicide selon le temps après la chirurgie et la variation du risque selon l’âge à la chirurgie.

Abstract

Objectives: Cosmetic breast implant women have been shown to have elevated rates of suicide, but suicide risk over time after implantation and according to age at surgery needs to be clarified. As well, clarification in risk for several other causes of death is needed. The purpose of this study is to fill these research gaps. Methods: This study presents an extended analysis of 10 more years of follow-up of a large Canadian cohort of women who received either cosmetic breast implants (n=24,558) or other cosmetic surgery (15,893). Over 70% of the cohort was followed for over 20 years. Mortality among implant women was compared to those of controls using multivariate Poisson models (mortality Rate Ratio (RR)) and the general female population using Standardized Mortality Ratios (SMRs). Results: Augmented women have been shown to have elevated rates of suicide relative to the general female population (SMR = 2.00, 95% CI= 1.66-2.41) and compared to women seeking other cosmetic surgery (RR = 1.43, 95% CI= 1.02-1.99). Our results are suggestive of an increasing trend in the mortality RR of suicide as time since implantation increases. Women who had breast augmentation at a young age were seen to have the highest suicide mortality ratio (RR = 1.92, 95% CI = 1.16-3.16). No differences were seen for several other causes of death when comparing augmented women to other cosmetic surgery women. Conclusions: Rates of suicide among augmented women remain elevated 20 years or more after surgery. Further work is needed to elucidate possible increase in suicide risk as time since implantation increases and variation in risk according to age at surgery.

Introduction

Mortality patterns among women with cosmetic breast implants have been reported by a number of investigators over the last decade (1-8). Given higher suicide rates observed among augmented women compared with women in the general population (1-8), psychological considerations have been of much interest (9-12). Some studies have suggested that the excess in suicide risk may change with length of time since surgery (1;6) and age at which surgery was performed (1;3;6). However, these reports have been severely limited by a small number of suicide deaths which makes it difficult to draw any solid conclusions. Given the limited evidence, suicide risk patterns over time following breast augmentation surgery and age at surgery still need to be clarified.

Studies have consistently reported that women with cosmetic breast implants are not at an increased risk of breast cancer mortality compared to women in the general population or to women with other cosmetic surgeries (1;3-8). However, clarification is needed for several other causes of death. For instance, three reports showed that women with implants have higher risks of death from respiratory diseases than women in the general population (6-8). Additionally, Lipworth et al. (6) and Koot et al. (8) reported an increased risk of lung cancer death relative to the general female population, but other studies showed no difference in risk (1;3;5;7). There is also a concern of an elevated risk of death from motor vehicle accidents (1) and several other types of injuries (4;6-8). Therefore, clarification is needed in order to understand mortality risk among women seeking breast augmentation.

In this report, we present an updated analysis after adding 10 more years of follow-up to our Canadian cohort of women with cosmetic breast implants (3). We aim to provide more evidence on the risk of suicide among women with cosmetic breast implants over a much longer period of time. Additionally, we provide further insights on several other causes of death among these women.

Methods

Study design, study population and selection criteria

The study population was detailed in our previous publication (3) and described here briefly. The cohort includes women, 18 years of age or older, who underwent bilateral cosmetic breast augmentation in the provinces of Ontario or Quebec, in Canada, between January 1, 1974, and December 31, 1989. A comparison group was assembled for the same time frame and includes women who received other common elective cosmetic surgeries: chemical peel or dermabrasion, coronal brow lift (eyebrow and forehead lift), otoplasty (ear surgery), rhinoplasty (nose surgery), rhytidectomy (face-lift), or blepharoplasty (eyelid surgery). Controls were identified through frequency matching to the breast implant recipients by year of first eligible plastic surgery, province of residence and surgeon. Eligibility criteria are detailed in our previous publication (3).

The cohorts include 40,451 women: 24,558 received breast implants (7,153 women from Ontario and 17,405 from Quebec) and 15,893 women (4,418 from Ontario and 11,475 from Quebec) received other common elective cosmetic surgeries. For both cohorts, information on year of surgery, age at surgery, personal identifying information (used only for linkage purposes) and verification of eligibility criteria was collected by review of medical (hospital or private clinic) records of all women in the cohort.

Mortality in the implant and other cosmetic surgery cohort was also compared to that of the general female population. Female mortality rates for the provinces of Ontario and Quebec were obtained from provincial vital registries (unpublished document, Chronic Disease Surveillance and Monitoring Division, Public Health Agency of Canada, Ottawa, 2011).

Ascertainment of outcomes

In our previous analysis (3), the mortality experience that occurred from the date of surgery until December 31, 1997 was reported. Mortality was assessed by linking personal identifying information of cohort members to the Canadian Mortality Database (CMDB) (13). Additional details are provided in our previous paper regarding the linkage with the CMDB (3). Mortality experience for the extended follow-up among the Quebec cohort was assessed for the period between January 1, 1998 and December 31, 2007 by linking cohort member identifiers using a deterministic linkage approach with the mortality file of Quebec held by the Quebec Institute of Statistics. Linkage of the Ontario cohort with the Ontario Mortality Database (OMDB) was performed at Cancer Care Ontario, covering the period between January 1, 1998 and December 31, 2006 using a computerized probabilistic record linkage system (14). Date of death and underlying cause of death, based on the International Classification of Diseases (ICD), 9th and 10th Revisions, were extracted from the databases. ICD 9 codes were converted to ICD 10 codes using a valid documentation (15). Where no link was found each patient was assumed to be alive at the end of follow- up.

Statistical analysis

The calculation of person-years of follow-up for each cohort member was done from one year after the date of surgery until the earliest of date of death or December 31, 2007. The first year of follow-up was excluded from the analyses to reduce the influence that pre- existing disease at the time of surgery may have had on our comparisons. Numbers of person-years and deaths were tabulated across strata defined by study group (implant women or other cosmetic surgery women), province of residence at the time of index surgery (Quebec or Ontario), attained age (18–24, 25–29, 30–34, . . ., 75–79, ≥ 80 years), calendar period of follow-up (1974–1977, 1978–1981, . . .,1994–1997, 1998-2001, 2002- 2007), period of surgery (1974–1977, 1978-1981, 1982–1985, 1986–1989), age at surgery (18–<30, 30–<40, ≥ 40 years) and length of follow-up (1-<5, 5-<10, 10-<15, 15-<20 and ≥20 years). Attained age, calendar period of follow-up and length of follow-up were time- dependent variables because women would contribute person-years to different categories

85 within these variables as they were followed over time. In contrast, women would contribute person-years to only one level of the classification variables period of surgery and age at surgery. The DATAB module in the Epicure software program was used to calculate person-years of follow-up (16).

Initially, overall and cause-specific mortality rates for both the breast implant patients and the other cosmetic surgery group were compared with those for the general population. Mortality rates for the provinces of Ontario and Quebec were obtained from provincial vital statistics registries as described above. The expected numbers of deaths in the cohort were estimated by multiplying the tabulated person-years of follow-up by the corresponding overall and cause-specific mortality rate observed in the general female population according to province (Ontario or Quebec), age (by 5-year age intervals), and calendar period of follow-up (1974–1977, 1978–1981, . . ., 1994–1997, 1998-2001, 2002-2007). Mortality rate for the implant and other cosmetic surgery women relative to the general population was evaluated by calculating the standardized mortality ratio (SMR), which is the ratio of the observed-to-expected number of deaths (17) For the comparison with general female population estimates, person-years contributed for the period after 1998 were reduced by interprovincial migration rates according to province, attained age and calendar period of follow-up on the basis of migration rates observed through active follow-up of the Canadian population (18). This was done to account for interprovincial mobility. This approach has been previously applied to reduce the impact of losses to follow-up in a cohort study (19). The SMRs and their 95% confidence intervals (CIs) were calculated for overall mortality and cause-specific deaths, assuming a Poisson distribution.

Comparisons of cause-specific deaths, between the implant recipients and the other plastic surgery group were done using multivariate Poisson regression models using the mortality rate ratio (RR) as the measure of association (20). The influence of confounding factors was evaluated using a backward deletion approach (21). P values for trend were computed using the median time since surgery value for each category as a continuous variable and including a first-order interaction term of time since surgery using the median values and the main exposure variable of interest. The strength of association (mortality rate ratio)

86 comparing implant to other surgery group was found to vary with time since surgery if the interaction term was statistically significant based on a two-tailed alpha of <5%. These analyses were done with SAS, version 9.2.

Results

The total number of person-years accrued by the breast implant cohort (n=24,558) was 599,992; for the other cosmetic surgery cohort (n=15,893), person-years accrued reached 389,199 (Table 1). The total amount of person-years when the interprovincial migration correction was applied decreased to 596,219 and 386,881 for the implant group and the other plastic surgery patients respectively. Given this extended follow-up, more than 70% of the women in both the breast implant and other cosmetic surgery cohorts were followed for at least 20 years. As well, the number of deaths identified has more than doubled in both the implant and other cosmetic surgery women compared with our previous publication (3). This represents a total of 1179 deaths (480 in previous follow-up) among implant women and a total of 874 deaths (383 in previous follow-up) among other cosmetic surgery women.

The comparisons with the general female population (Table 2) showed a statistically significant reduction of overall mortality rate in both the implant cohort (SMR = 0.71, 95% CI = 0.67-0.76) and the other cosmetic surgery cohort (SMR = 0.60, 95% CI = 0.56-0.64). As well, significant lower than expected rates of mortality for endocrine diseases, mental disorders, circulatory diseases, respiratory diseases, digestive diseases, genitourinary diseases, overall cancers and other causes of death combined were observed among both the implant women and the other cosmetic surgery women relative to general female population estimates. However, augmented women had a statistically significant 17% (SMR = 1.17, 95 % CI = 1.02-1.33) increase in rate of lung cancer mortality compared to general female population estimates. Increased rates of suicide death were observed in both the augmented women (SMR = 2.00, 95% CI= 1.66-2.41) and the other plastic surgery women (SMR = 1.41, 95% CI= 1.05-1.86).

Internal comparisons revealed that women with breast implants have a significantly increased rate of suicide (RR = 1.43, 95% CI= 1.02-1.99) relative to other cosmetic surgery women (Table 3). Conversely, implant women were found to have a significantly lower rate of brain cancer mortality (RR = 0.47, 95% CI= 0.26-0.86). Colorectal cancer mortality was also higher among augmented women relative to women with other plastic surgery (RR = 1.67, 95% CI= 1.08-2.59).

Further analyses for suicide according to surgery characteristics were undertaken (Table 4). The results for implant women relative to females in the general population are suggestive of an increasing trend in suicide rate ratio according to the time since surgery. Specifically, the SMR for suicide increased from 1.66 (95% CI= 0.93-2.73) for 1 to 5 years after surgery, to 2.42 (95% CI = 1.61-3.49) for follow-up of 15 to 20 years after surgery. The SMR for the follow-up interval of more than 20 years after surgery was 2.15 (95% CI = 1.38-3.20). As well, the internal comparison revealed that the RR was highest for the interval of 20 years or more after surgery (RR = 2.29, 95% CI = 0.98-5.31) (p value for trend in the RR over time since surgery = 0.19).

Women who had breast augmentation at a young age (18 to <30 years of age) were found to have the highest rate ratio of suicide compared with either women in the general population (SMR = 2.63, 95% CI = 1.99-3.40) or those with other cosmetic surgeries (RR = 1.92, 95% CI = 1.16-3.16) (p value for trend in the RR over age at surgery = 0.24). Additionally, women that underwent breast implantation in the period of 1986 to 1989 had the greatest increase in the rate of suicide when compared to general female population estimates (SMR = 2.59, 95% CI = 1.91-3.30) and other cosmetic surgery women (RR = 2.92, 95% CI = 1.22-6.96).

Discussion

This extended follow-up of our Canadian cohort of women with cosmetic breast implants shows a reduced mortality rate for several causes of death among these women. In contrast, augmented women have been found to have persistently elevated rates of suicide either compared to women in the general population or other cosmetic surgery women. An

88 increasing trend in the suicide rate ratio as time since implantation increases, was observed. Women who had breast augmentation at a young age had the highest suicide rate ratio either compared to the general female population or those with other cosmetic surgeries. However, these findings were not statistically significant.

In our cohort, a reduced rate for overall mortality was found among women who received breast implants when compared with the general population. This finding is consistent with previous cohort studies including our previous follow-up (1-3;5). We also showed that women with cosmetic breast implants have lower rates of several specific causes of death such as endocrine diseases, mental disorders, circulatory diseases, respiratory diseases, digestive diseases, genitourinary diseases, overall cancers and other causes of death combined. Similar mortality patterns were observed among women with other cosmetic surgeries relative to the general female population, suggesting similarities with breast implant women in terms of sociodemographic and lifestyle characteristics (12;22). The reduced rates in both groups seem consistent with the fact that women undergoing cosmetic procedures have a different risk factor profile than the female general population (23). Women undergoing cosmetic procedures are recognized to be of higher socioeconomic status which is correlated with improved health status (24). This highlights the realization that women with other cosmetic surgeries are a more appropriate control group when analyzing mortality in augmented women.

An increased rate of mortality from lung cancer was observed among augmented women compared with general female population estimates which may reflect previously reported higher smoking rates among these women compared with women in the general population (23). In fact, an excess of deaths compared with the general population for lung cancer and nonmalignant respiratory diseases was seen in a previous Swedish study and authors attributed this result to differences in smoking habits (6). Interestingly, we observed a non-statistically significant increase of lung cancer mortality when comparing implant women to other cosmetic surgery women which is consistent with a previous report (1). However, this study also reported similar rates of smoking between cosmetic breast implant women and other cosmetic surgery women (1). Moreover, in our study, the increase in the rate of lung cancer

89 mortality varied substantially according to time since surgery (data not shown). Thus, the association of breast implantation to increased lung cancer mortality rate remains uncertain.

A statistically significant increased rate for colorectal cancer mortality and a significant decrease in brain cancer mortality were observed when comparing augmented women to the control group. However, we found in our recent extended follow-up of cancer incidence (25) that there were no differences in brain cancer incidence (Incidence rate ratio = 0.67, 95% CI: 0.38–1.16) and colorectal cancer incidence between these two groups (Incidence rate ratio = 1.14, 95% CI: 0.88–1.46), which is consistent with previous investigations (26-29). Therefore, these findings for brain cancer and colorectal cancer mortality should be tempered.

Our finding of an increased suicide rate among women with breast augmentation relative to women in the general population is in agreement with previous epidemiological studies and our earlier paper (1-8). Specifically, studies have consistently reported that the rate of suicide among these women is two- to threefold higher than expected in the general female population (1-8). The excess rate of suicide may reflect increased prevalence of underlying psychiatric problems prior to breast augmentation surgery and other risk factors for suicide among a subset of women seeking cosmetic breast augmentation (30;31). In fact, women seeking cosmetic breast augmentation have been found to have higher frequencies of psychiatric treatments and psychiatric hospital admissions before surgery, as well as varying degrees of depression, anxiety and low self-esteem compared to women not seeking augmentation (7;30). Moreover, elevated frequencies of ongoing psychiatric treatments and lower health-related quality of life in the dimension of distress after breast augmentation have also been reported among these women (9;10;32). In recent years, some authors have raised the issue of body dysmorphic disorder (BDD) as a possible explanation of the increased rate of suicide among augmented women (33;34). Specifically, BDD is a psychiatric diagnosis characterized by excessive dissatisfaction of body image that can lead to substantial distress (9) and suicidal ideation (11). In fact, the mean annual suicidal attempt rate in the U.S. population among individuals with BDD is 3 to 12 times higher than that in the general population (11). Studies have also shown that between 3% and 15% of patients undergoing cosmetic surgery have some form of BDD (35) and more than 90%

90 of them report either no change or a worsening of their symptoms following cosmetic surgery (9;11;12). Consequently, BDD may be a contraindication to some forms of cosmetic surgery, including augmentation mammoplasty (9;11). Thus, it is possible that a subset of women in our study suffers from BDD and this may explain the increased rate of suicide among these women.

Our finding of an increasing trend in the rate ratio (SMR) of suicide with time since surgery compared with the general female population is consistent with a previous Swedish study. Compared to population rates, Lipworth et al. (6) reported an excess rate of suicide based on 24 observed deaths which became apparent only 10 years after breast augmentation surgery and continued to increase thereafter. Compared to population rates, a study by Brinton et al. (1) based on 29 suicide deaths among women with breast implants also showed that the higher risk of suicide became apparent only 10 years after breast augmentation and remained steady thereafter. However, the increasing trend in the rate of suicide according to time since surgery for the comparison with women in the general population needs to be interpreted with caution as SMRs were used for this analysis. Thus, no proper statistical value, such as the p value for trend, could be calculated to evaluate the trend in the rate ratio (SMR) of suicide according to time since surgery.

Internal comparisons revealed a significant increased rate of suicide among implant women compared with other cosmetic surgery women. As well, we have observed that the excess rate of suicide among implant women compared with those with other cosmetic surgeries tends to increase with time since surgery and was highest for the interval of 20 years or more after surgery. However, this result and the p value for trend in the RR over time since surgery did not reach statistical significance. A possible explanation for the increased rate of suicide among augmented women and for the suggestive increase in the suicide rate ratio over time since surgery may be linked with the complications associated with the implant itself, which may contribute to increased despair and consequently increased risk of suicide (36). Local complications over time associated with the implant include implant rupture, serious infection, hematoma and capsular contracture (37). In fact, hematomas and serious infections can occur shortly after breast augmentation surgery (38) while capsular

91 contracture has been shown to be present in approximately 8 % of women 6 years after implantation (37) and in about 15% to 20% of women over a longer period of time (39). Brinton et al. (1) previously reported an increased rate of suicide among augmented women compared with surgical controls (RR = 2.58, 95% CI = 0.9-7.8), but results were not statistically significant. In addition, their evaluation of the suicide rate ratio when comparing augmented women to surgical controls over time since surgery did not provide conclusive results (1). As well, there is evidence that the time since breast augmentation surgery negatively affects patient’s satisfaction (40).

We also observed that women who had breast augmentation at a young age had the highest increase in suicide rate compared to either other cosmetic surgery women or women in the general population. This finding is somewhat different than reported in our previous follow-up and the results observed in previous studies for which the highest increase in suicide rate was seen for women undergoing breast augmentation after 40 years of age (1;3;6). However, analyses according to age group in previous reports were limited by small numbers of suicide death. Therefore, we believe that our finding which is based on the largest number of suicide deaths observed to date provides important information on the risk of suicide according to age at surgery.

Risk of death associated with motor vehicle accidents among augmented women is an emerging issue (1). Although this issue is important, the number of deaths on which the estimations were based was quite small, which increases the likelihood that the results may be due to chance. In our study, no differences were found for motor vehicle accidents and other types of fatal injuries combined when comparing implant women to either the general female population or other cosmetic surgery women.

Loss to follow-up and interprovincial residential mobility could affect our comparisons with the general population. However, this was minimized by linking the cohort members to national mortality databases for the follow-up prior to 1998. In addition, we accounted for interprovincial migration rates for the second phase of this study. There is still a possibility that some deaths were missed, especially for the linkage with provincial

92 databases because, as time passes, more people are expected to move. For this reason, we evaluated the differences in the completeness of ascertainment of deaths comparing ascertainment using national registries (as in our initial follow-up (3)) to that using provincial registries (as in our extended follow-up) using a common period of linkage which extended from 1995 to 1997 for both the Quebec and Ontario cohorts. This showed that less than 13% of deaths identified in the national linkage were not identified in the provincial linkages and less than 6% of deaths were missing in the national linkage compared with the provincial linkages. This results in a net missingness of deaths of 7%. Linkages to both the national and provincial registries were done without knowledge of the cosmetic surgery received. Thus, we know of no reason why incomplete ascertainment would be differential between implant women and controls.

Strengths of our study include the largest sample size and the long follow-up time for a study evaluating long terms effects of cosmetic breast implants. The cohort design and the fact that we used two comparison groups are also strengths of our study. Moreover, we excluded the first year after surgery in the follow-up of this cohort, which is consistent with approaches undertaken previously (1;5).

In conclusion, augmented women have been shown to have elevated rates of suicide relative to women in the general population and to women seeking other cosmetic surgery. As well, our results are suggestive of a trend of increasing excess rate of suicide with increasing time since implantation. We also observed that women who had breast augmentation under age 30 had the highest increase in suicide rate compared to either other cosmetic surgery women or women in the general population. In our study, we showed reduced mortality rates among augmented women compared with the general female population for endocrine diseases, mental disorders, circulatory diseases, respiratory diseases, digestive diseases, genitourinary diseases, overall cancers and other causes of death combined which seems to be correlated with the improved health status of women undergoing cosmetic surgery. We believe that it is prudent for plastic surgeons to refer breast implant patients for a psychiatric consultation before surgery, if patients report a history of psychopathology or if the plastic surgeon suspects such. Long term monitoring

93 for post implant psychiatric conditions also seems warranted among women with cosmetic breast implants.

References

(1) Brinton LA, Lubin JH, Murray MC, Colton T. Mortality rates among augmentation mammoplasty patients: an update. Epidemiology 2006;17(2):162-9.

(2) Rubin JP, Landfair AS, Shestak K, Lane D, Valoski A, Chang Y, et al. Health characteristics of postmenopausal women with breast implants. Plast Reconstr Surg 2010;125(3):799-810.

(3) Villeneuve PJ, Holowaty EJ, Brisson J, Xie L, Ugnat A-M, Latulippe L, et al. Mortality among Canadian women with cosmetic breast implants. Am J Epidemiol 2006;164(4):334-41.

(4) Pukkala E, Kulmana I, Hovi S-L, Hemminki E, Keskimäki I, Pakkanen M, et al. Causes of death among Finnish women with cosmetic breast implants, 1971-2001. Ann Plast Surg 2003;51(4):339-42.

(5) Brinton LA, Lubin JH, Burich C. Mortality among augmentation mammoplasty patients. Epidemiology 2001;12(3):321-6.

(6) Lipworth L, Nyren O, Ye W, Fryzek JP, Tarone RE. Excess mortality from suicide and other external causes of death among women with cosmetic breast implants. Ann Plast Surg 2007;59(2):119-23.

(7) Jacobsen PH, Holmich LR, Mclaughlin JK, Johansen C, Olsen JH, Kjøller K, et al. Mortality and suicide among Danish women with cosmetic breast implants. Arch Intern Med 2004;164(22):2450-5.

(8) Koot VCM, Peeters PHM, Granath F, Grobbee DE, Nyren O. Total and cause- specific mortality among Swedish women with cosmetic breast implants: prospective study. BMJ 2003;326(7388):527-8.

(9) Crerand CE, Infield AL, Sarwer DB. Psychological considerations in cosmetic breast augmentation. Plastic Surgical Nursing 2009;29(1):49-57.

(10) Lipworth L, Kjoller K, Holmich LR, Friis S, Olsen JH, Mclaughlin JK. Psychological characteristics of Danish women with cosmetic breast implants. Ann Plast Surg 2009;63(1):11-4.

(11) Sarwer DB, Brown GK, Evans DL. Cosmetic breast augmentation and suicide. Am J Psychiatry 2007;164(7):1006-13.

(12) Sarwer DB. The psychological aspects of cosmetic breast augmentation. Plast Reconstr Surg 2007;120(Suppl. 1):110S-7S.

(13) Band PR, Gaudette LA, Hill GB, Holowaty EJ, Huchcroft SA, Johnston GM, et al. The making of the Canadian cancer registry: cancer incidence in Canada and its regions, 1969 to 1988. Ottawa, ON; 1993.

(14) Matchware Technologies I. Automatch generalised record linkage system, version 4.2: user's manual. Kennebunk, Maine: MatchWare Technologies, Inc.; 1998.

(15) Anderson RN, Minimo AM, Hoyert DL, Rosenberg HM. Comparability of cause of death between ICD-9 and ICD-10. Preliminary estimates. National Vital Statistics Reports 2001;49(2).

(16) EPICURE: risk regression and data analysis software [computer program]. Seattle: Hirosoft International Corporation; 1996.

(17) Breslow NE, Day NE. Statistical methods in cancer research. Vol. II, The design and analysis of cohort studies. Lyon, France: International Agency for Research on Cancer; 1987.

(18) Bernard A, Finnie R, St-Jean B. Interprovincial mobility and earnings. Ottawa: Statistics Canada; 2008.

(19) Phillips N, Coldman A. Comparison of nonbreast cancer incidence, survival and mortality between breast screening program participants and nonparticipants. Int J Cancer 2007;122(1):197-201.

(20) Kleinbaum DG, Kupper LL, Muller KE, Nizam A. Applied regression analysis and other multivariable methods. 3rd ed. Brooks/Cole Publishing Company; 1998.

(21) Rothman KJ, Greenland S, Lash TL. Modern epidemiology. 3rd ed. Philadelphia: Lippincott, Williams & Wilkins; 2008.

(22) Gladfelter J. Breast augmentation 101: Understanding cosmetic breast mammaplasty. Plastic Surgical Nursing 2007;27(3):136-45.

(23) Kjoller K, Holmich LR, Fryzek JP, Jacobsen PH, Friis S, McLaughlinJK, et al. Characteristics of women with cosmetic breast implants compared with women with other types of cosmetic surgery and populationbased controls in Denmark. Ann Plast Surg 2003;50:6-12.

(24) Cook RR, Perkins LL. The prevalence of breast implants among women in the United States. Curr Top Microbiol Immunol 1996;210:419-25.

(25) Pan SY, Lavigne E, Holowaty EJ, Villeneuve PJ, Xie L, Morrison H, et al. Canadian Breast Implant Cohort: extended follow-up of cancer incidence. 2012. Ref Type: Unpublished Work

(26) Friis S, Holmich LR, Mclaughlin JK, Kjøller K, Fryzek JP, Henriksen TF, et al. Cancer risk among Danish women with cosmetic breast implants. Int J Cancer 2006;118(4):998-1003.

(27) Lipworth L, Tarone RE, Friis S, Ye W, Olsen JH, Nyren O, et al. Cancer among scandinavian women with cosmetic breast implants: A pooled long-term follow- up study. Int J Cancer 2009;124:490-3.

(28) Mclaughlin JK, Lipworth L, Fryzek JP, Weimin Y, Tarone RE, Nyren O. Long- term cancer risk among Swedish women with cosmetic breast implants: An update of a nationwide study. J Natl Cancer Inst 2006;98(8):557-60.

(29) Pukkala E, Boice Jr.JD, Hovi S-L, Hemminki E, Asko-Seljavaara S, Keskimäki I, et al. Incidence of breast cancer and other cancers among Finnish women with cosmetic breast implants, 1970-1999. J Long Term Eff Med Implants 2002;12(4):251-3.

(30) Lipworth L, Mclaughlin JK. Excess suicide risk and other external causes of death among women with cosmetic breast implants: a neglected research priority. Curr Psychiatry Rep 2010;12:234-8.

(31) Mclaughlin JK, Wise TN, Lipworth L. Increased risk of suicide among patients with breast implants: do the epidemiologic data support psychiatric consultation? Psychosomatics 2004;45:277-80.

(32) Lamberg S, Manninen M, Kulmana I, Mclaughlin JK, Liworth L, Pakkanen M, et al. Health-related quality of life issues after cosmetic breast implant surgery in Finland. Ann Plast Surg 2008;61(5):485-8.

(33) Didie ER, Phillips KA. Mortality among Canadian women with cosmetic breast implants - a letter to the editor. Am J Epidemiol 2007;165(7):846.

(34) Sarwer DB, Brown GK, Evans DL. Cosmetic breast augmentation and suicide. Am J Psychiatry 2007;164(7):1006-13.

(35) Bjornsson AS, Didie ER, Phillips KA. Body dysmorphic disorder. Dialogues Clin Neurosci 2010;12(2):221-32.

(36) Zuckerman D. Mortality in Swedish women with cosmetic breast implants: study found increased risk of suicides and cancer deaths. (Letter). BMJ 2003;326:1266.

(37) Codner MA, Mejia JD, Locke MB, Ch.B.MB, Mahoney A, Thiels C, et al. A 15- year experience with primary breast augmentation. Plast Reconstr Surg 2011;127(3):1300-10.

(38) Hvilsom GB, Hölmich LR, Henriksen TF, Lipworth L, Mclaughlin JK, Friis S. Local complications after cosmetic breast augmentation: results from the Danish Registry for Plastic Surgery of the Breast. Plast Surg Nurs 2010;30(3):172-9.

(39) Tang SS, Gui GP. A review of the oncologic and surgical management of breast cancer in the augmented breast: Diagnostic, surgical and surveillance challenges. Ann Surg Oncol 2011;Feb. 11.

(40) McCarthy CM, Klassen AF, Cano SJ, Scott A, Vanlaeken N, Lennox PA, et al. Patient satisfaction with postmastectomy breast reconstruction: a comparison of saline and silicone implants. Cancer 2010;116(24):5584-91.

Table 1. Frequency distribution for selected characteristics of women who received breast implants and women who received other cosmetic surgeries, Canadian Breast Implant Cohort Study. Breast implant O ther plastic Characteristics patients surgery patients Length of follow-up (years), N (%) <1 17 (0.1) 11 (0.1) 1 to <5 80 (0.3) 53 (0.3) 5 to <10 135 (0.6) 117 (0.7) 10 to <15 197 (0.8) 174 (1.1) 15 to <20 4,902 (20.0) 2,762 (17.4) 20 to <25 8,225 (33.5) 5,535 (34.9) ≥25 11,002 (44.8) 7,241 (45.6) Period of surgery, N (%) 1974 – 1977 4,726 (19.2) 3,011 (19.0) 1978 – 1981 5,750 (23.4) 3,766 (23.7) 1982 – 1985 6,685 (27.2) 4,706 (29.6) 1986 – 1989 7,397 (30.1) 4,410 (27.7) Age at surgery (years), N (%) 18 to <25 3,665 (14.9) 3,481 (21.9) 25 to <30 5,961 (24.3) 3,064 (19.3) 30 to <35 6,868 (28.0) 2,828 (17.8) 35 to <40 4,195 (17.1) 2,357 (14.8) 40 to <45 2,068 (8.4) 1,562 (9.8) ≥45 1,801 (7.3) 2,601 (16.4)

Mean age at surgery (SD), years 32.2 (7.8) 33.5 (10.4) Mean duration of follow up (range), years 24.4 (0.1-34.0) 24.5 (0.0-34.0) Total person-years of follow up1 599,992 389,199 Total number of deaths 1,195 885 Total number of women 24,558 15,893 1Person-years were accrued from the date of surgery until the earliest of date of death, December 31, 2006 (Ontario) or December 31, 2 007 (Quebec).

Table 2. Standardized mortality ratios (SMRs)1 for selected causes of death based on general female population mortality rates (1974- 2007) among breast implant and other cosmetic surgery women, compared with previous analysis.

Breast implant patients Other plastic surgery patients O bs. deaths SMR SMR O bs. deaths SMR SMR 1974 – 2007 1974 – 2007 1974 – 1997 1974 – 2007 1974 – 2007 1974 – 1997 Follow-up Follow-up Follow-up Follow-up Follow-up Follow-up Cause of death ICD-10 interval interval interval interval interval interval All causes A00-Z99 1179 0.71* 0.74* 874 0.60* 0.68* Infectious diseases A00-B99 16 0.65 0.96 13 0.61 0.63 Endocrine diseases E00-E90 9 0.16* 0.22* 8 0.15* 0.28* Mental disorders F00-F99 14 0.59* 0.89 8 0.27* 0.87 Circulatory diseases I00-I99 165 0.48* 0.55* 175 0.49* 0.54* Coronary heart I20-I25 84 0.48* 0.50* 87 0.46* 0.46* Cerebrovascular I60-I69 47 0.61* 0.89 47 0.62* 0.67 Respiratory diseases J00-J99 47 0.54* 0.60* 37 0.41* 0.55* Digestive diseases K00-K93 33 0.55* 0.52* 15 0.28* 0.34* Genitourinary diseases N00-N99 5 0.26* 0.16* 4 0.19* 0.19* Cancer C00-C97 (excl. C44) 626 0.80* 0.76* 457 0.73* 0.76* Breast C50 90 0.52* 0.45* 82 0.63* 0.62* Brain C71 18 0.75 0.57 28 1.55* 1.21 Genital C51-C58 53 0.69* 0.72 41 0.68* 0.59* Colorectal C18-C21 64 0.81 1.09 30 0.44* 0.66 Bronchus and lung C34.1-3, .8, .9, C39.8, C45.7 235 1.17* 1.12 151 0.95 1.12 Lymphohematopoietic C81-C96 38 0.66* 0.66* 35 0.72 0.72 Injuries V01-Y98 203 1.48* 1.37 101 1.06 1.14 Suicide X60-X84 113 2.00* 1.73* 50 1.41* 1.55* Motor vehicle V02-V04, V09.0, .2, V12-V14, accidents V19.0-V19.2, V19.4-V19.6, V20-V79, V80.3-V80.5, V81.0-V81.1, V82.0-V82.1, V83-V86, V87.0-V87.8, V88.0-V88.8, V89.0, V89.2 37 1.07 1.07 22 0.93 0.93 Other Those not listed above for injuries 53 1.15 1.15 29 0.80 0.80 Other deaths Those not included above 61 0.54* 0.54* 56 0.55* 0.55* 1The SMR is the ratio of the observed to expected cases; the expected number of cases was estimated by applying age, period and province (Ontario or Quebec) specific mortality rates to the corresponding number of person-years of follow-up observed in the cohort. *Denotes a statistically significant difference based on a two-tailed alpha of less than 5 %.

100

Table 3. Mortality rate ratios (RRs)1 for selected causes of death between breast implant and other cosmetic surgery women with comparisons with previous analysis. Internal comparison Internal comparison 1974 – 2007 1974 – 1997 Follow-up interval Follow-up interval Cause of death ICD-10 RR 95 % CI RR 95 % CI All causes A00-Z99 1.09 0.99-1.19 1.02 0.89-1.17 Infectious diseases A00-B99 0.99 0.47-2.09 1.46 0.43-4.91 Endocrine diseases E00-E90 0.98 0.37-2.57 0.72 0.19-2.73 Mental disorders F00-F99 1.98 0.82-4.78 1.09 0.27-4.44 Circulatory diseases I00-I99 0.89 0.72-1.11 0.97 0.70-1.34 Coronary heart I20-I25 0.92 0.68-1.25 1.08 0.67-1.74 Cerebrovascular I60-I69 0.86 0.57-1.30 1.21 0.67-2.17 Respiratory diseases J00-J99 1.32 0.85-2.04 1.14 0.56-2.34 Digestive diseases K00-K93 1.84 0.99-3.41 1.18 0.46-3.04 Genitourinary diseases N00-N99 1.51 0.39-5.73 1.73 0.28-10.59 Cancer C00-C97 (excl. C44) 1.05 0.93-1.19 0.96 0.79-1.16 Breast C50 0.79 0.58-1.07 0.76 0.48-1.19 Brain C71 0.47 0.26-0.86 0.42 0.15-1.15 Genital C51-C58 0.97 0.64-1.46 1.10 0.58-2.10 Colorectal C18-C21 1.67 1.08-2.59 1.42 0.73-2.77 Bronchus and lung C34.1-3, .8, .9, C39.8, C45.7 1.20 0.97-1.47 0.94 0.66-1.34 Lymphohematopoietic C81-C96 0.94 0.59-1.50 0.68 0.32-1.49 Injuries V01-Y98 1.34 1.05-1.70 1.18 0.86-1.61 Suicide X60-X84, Y87.0 1.43 1.02-1.99 1.10 0.72-1.69 Motor vehicle accidents V02-V04, V09.0, .2, V12-V14, V19.0-V19.2, V19.4-V19.6, V20-V79, V80.3-V80.5, V81.0-V81.1, V82.0-V82.1, V83-V86, V87.0-V87.8, V88.0-V88.8, V89.0, V89.2 1.15 0.68-1.96 1.35 0.69-2.64 Other Those not listed above for injuries 1.17 0.73-1.89 1.19 0.64-2.16 Other deaths Those not included above 0.80 0.48-1.34 0.90 0.22-3.71 1The RRs estimates were derived using Poisson multivariate regression model and were adjusted for attained age and province of residence.

101

Table 4. Standardized mortality ratios (SMRs)1 for suicide death based on general female population mortality rates (1974-2007) among breast implant and other cosmetic surgery women, and incidence rate ratios (RRs)2 of suicide death for breast implant vs. other cosmetic surgery women by selected surgery characteristics.

Breast implant women O ther cosmetic surgery women Internal comparison

Surgery characteristics Obs. deaths Exp. Deaths SMR O bs deaths Exp. Deaths SMR RR 95 % CI Time since surgery (years) 1 to <5 15 9.1 1.66 8 5.8 1.37 1.18 0.50-2.77 5 to <10 22 12.1 1.82* 12 7.5 1.59 1.15 0.57-2.32 10 to <15 24 12.4 1.94* 12 7.5 1.59 1.25 0.63-2.48 15 to <20 28 11.6 2.42* 11 7.1 1.54 1.59 0.79-3.22 ≥20 24 11.2 2.15* 7 7.4 0.95 2.29 0.98-5.31 Period of surgery 1974 - 1977 25 14.1 1.77* 13 9.0 1.44 1.22 0.63-2.39 1978 - 1981 29 14.9 1.95* 11 9.6 1.15 1.72 0.86-3.42 1982 - 1985 27 14.9 1.81* 20 10.0 2.00* 0.92 0.52-1.65 1986 - 1989 32 12.4 2.59* 6 6.8 0.88 2.92* 1.22-6.96 Age at surgery (years) 18 to < 30 57 21.7 2.63* 21 15.2 1.38 1.92* 1.16-3.16 30 to <40 42 26.9 1.56* 19 12.8 1.49 1.05 0.61-1.80 ≥40 14 7.6 1.84* 10 7.5 1.34 1.36 0.60-3.06 1The SMR is the ratio of the observed to expected cases; the expected number of cases was estimated by applying age, period an d province (Ontario or Quebec) specific suicide rates to the corresponding number of person-years of follow-up observed in the cohort. 2The RRs estimates were derived using Poisson multivariate regression model and were adjusted for attained age and province of residence. *Denotes a statistically significant difference based on a two-tailed alpha of less than 5 %.

102

Chapter 4: Do breast implants adversely affect prognosis among those subsequently diagnosed with breast cancer? Findings from an extended follow-up of a Canadian cohort

103

Résumé

Objectifs: Les implants mammaires pour fins esthétiques peuvent porter atteinte à la détection de cancers du sein. Les buts de cette étude étaient d’examiner si les implants mammaires ainsi que les caractéristiques des implants mammaires sont associés avec des stades plus avancés de tumeurs cancéreuses au diagnostique ainsi qu’à une survie réduite au cancer du sein. Méthodes : La population à l’étude inclut toutes les femmes diagnostiquées avec un cancer du sein dans le cadre d’une grande cohorte canadienne de femmes avec des implants mammaires. Un total de 409 femmes avec augmentation mammaire et 444 femmes avec une autre chirurgie esthétique ont été diagnostiquées avec un cancer du sein. Ces femmes ont été comparées en terme de stade au diagnostique de cancer du sein avec des modèles multivariés de régression logistique multinomiale. Des modèles de régression de Cox on été utilisés pour les analyses sur la mortalité spécifique au cancer du sein. Des comparaisons ont également été effectuées selon les caractéristiques des implants. Résultats : Comparativement aux femmes avec une autre chirurgie esthétique, les femmes avec des implants mammaires avaient des stades plus avancés au diagnostique de cancer du sein (Rapport de Cote d’avoir un stade III/IV vs. Stade I au diagnostique : 3.04, IC à 95%: 1.81 - 5.10; p < 0.001). Une augmentation statistiquement non significative dans le taux de mortalité spécifique au cancer du sein a été observée chez les femmes avec implants mammaires (Rapport de taux (RT) de mortalité = 1.32, IC à 95%: 0.94-1.83, p = 0.11). Aucune différence statistiquement significative n’a été observée dans le stade et dans la mortalité au cancer du sein selon les caractéristiques des implants. Conclusion : Au diagnostique, les cancers du sein étaient à des stades plus avancés parmi les femmes avec des implants mammaires. La survie spécifique au cancer du sein était également inférieure parmi ces femmes malgré que ces résultats fussent statistiquement non significatifs.

104

Abstract

Objectives: Cosmetic breast implants may impair the ability to detect breast cancers. The aims of this study were to examine whether implants and implant characteristics are associated with more advanced breast tumors at diagnosis and poorer survival. Methods: Study population includes all invasive breast cancer cases diagnosed during follow-up of the large Canadian Breast Implant Cohort. A total of 409 women with cosmetic breast implants and 444 women with other cosmetic surgery were diagnosed with breast cancer. These women were compared for stage at diagnosis using multinomial logistic regression models. Cox proportional hazards regression models were used for breast cancer-specific mortality analyses. Comparisons were also performed according to implant characteristics. Results: Compared to women with other cosmetic surgery, those with cosmetic breast implants had later stage at breast cancer diagnosis (Odds Ratio of having stage III/IV vs stage I at diagnosis: 3.04, 95 % confidence interval (CI): 1.81 - 5.10; p < 0.001). A non- statistically significant increase in breast cancer-specific mortality rate for women with breast implants relative to surgical controls was observed (Hazard Ratio = 1.32, 95 % CI: 0.94-1.83, p = 0.11). No statistically significant differences in stage and breast cancer mortality were observed according to implant characteristics. Conclusion: At diagnosis, breast cancers tended to be at more advanced stages among women with cosmetic breast implants. Breast cancer specific survival was also lower in these women although the reduction did not reach statistical significance.

105

Introduction

An estimated 5 to 10 million women across the world have breast implants (1). Such implants have become one of the most frequently performed cosmetic surgeries among women (2). Public health concern arose in the past decades regarding possible long term health effects of breast augmentation (1;3-5). Numerous epidemiological investigations focused on long term cancer risk, especially for breast carcinoma because of the proximity of the tissue to the implant (6-12). However, studies showed either no association or a negative association between breast implants and breast cancer development (6-12).

Nonetheless, cosmetic breast implants still raise some concerns because they may impair the ability to detect breast cancers at an early stage. Implants are radiopaque at mammography. Consequently, they can hinder visualization of breast tissue and affect the identification of breast tumors even when using specialized radiographic techniques (13- 16). Thus, breast implants could result in a later stage at diagnosis and reduced survival (17). As the female population with breast implants increases and ages, the number of breast cancers diagnosed in this population will increase and public health concerns associated with the detection and survival of these breast cancers need to be addressed.

Several epidemiological studies that focused on the detection of breast cancer have compared the stage distribution at diagnosis among women with cosmetic breast implants to stage at diagnosis of breast cancer in non-augmented women. The results of these studies were conflicting which can be attributed, at least in part, to methodological issues (18). Only three publications, including our previous report, showed a statistically significant shift towards more advanced breast tumors at diagnosis among augmented women (19-21). However, several other publications reported no statistically significant differences in breast cancer stage at diagnosis comparing augmented to non-augmented women (6;9;12;22-35). Furthermore, specific implant characteristics such as implant volume and placement might affect the detection of breast cancer (36). Specifically, implants placed directly under breast tissue (subglandular placement) are suspected to obstruct mammographic visualization of breast tissue more so than submuscular placement

106

(33;37;38). However, only our previous report was able to investigate this issue. It did not provide clear evidence of more advanced stage of breast cancer at diagnosis among women with subglandular implants compared with women whose implants are placed submuscularly (21).

In addition to the concern of possible delayed diagnosis of breast cancer among augmented women, breast cancer related survival has been subject to several investigations (21;26- 29;33). If diagnosis of breast cancer is delayed in augmented women, such delay could translate into poorer survival. To date, all published studies reported no statistically significant differences in breast cancer-specific survival when comparing augmented women with breast cancer to non-augmented women with breast cancer (21;26-29;33;39). However, the small numbers of incident breast cancer cases and insufficient follow-up time after diagnosis in these studies may have limited the statistical power necessary to detect a difference in survival. Additionally, no study has evaluated breast cancer survival according to implant characteristics.

To investigate these concerns, we provide analyses of an extended follow-up of the Canadian Breast Implant Cohort, the largest cohort assembled to date to examine long term health effects of cosmetic breast implants (21). Specifically, we evaluate whether implants are associated with more advanced breast tumors at diagnosis, and whether specific implant characteristics such as placement, implant type, implant envelope and fill volume affect the stage at diagnosis of breast cancer. We also examine whether cosmetic breast implants and specific implant characteristics are associated with poorer survival following diagnosis of breast carcinoma. With the addition of 10 more years of follow-up to our cohort, the identification of additional incident breast cancer cases and attendant mortality events can help clarify the effect, if any, of breast implants on the detection and prognosis of breast cancer.

107

Methods

Study population

Incident breast cancer cases were identified among a Canadian cohort of women (N=40,451) who previously underwent bilateral cosmetic breast augmentation (N=24,453) or other cosmetic surgery (N=15,893) in Ontario or Quebec between January 1, 1974, and December 31, 1989, and who were 18 years of age or older at the time of first cosmetic surgery. A detailed description of the study population with selection criteria is provided in our previous publications (7;21;40). Other cosmetic surgeries included the following procedures: chemical peel or dermabrasion, coronal brow lift (eyebrow and forehead lift), otoplasty (ear surgery), rhinoplasty (nose surgery), rhytidectomy (face-lift), and blepharoplasty (eyelid surgery).

Ascertainment of breast cancer cases and vital status

Incident cases of breast cancer and deaths were identified by linking personal identifying information of the cohort members to national and provincial databases. In our previous follow-up (21), cohort members were linked to the Canadian Cancer Registry (CCR) (41) and the Canadian Mortality Database (CMDB) (42) in order to identify incident breast cancer cases and vital status from the date of index cosmetic surgery until December 31, 1997. The cohort was also linked to cancer registry data before the index date of surgery, back to the year 1969, to exclude women diagnosed with cancer before their index surgeries. For the extended follow-up, incident cases of breast cancer who were diagnosed between January 1, 1998 and December 31, 2006 (Quebec) or December 31, 2007 (Ontario) were identified by linking to provincial registry databases, namely the Ontario Cancer Registry (OCR) (43) for the Ontario cohort and the Quebec Tumor Registry (QTR) for the Quebec cohort (44). The date of diagnosis was extracted from the registries for all identified incident breast cancer cases. The cohort was also linked to provincial registers to identify vital status and, if applicable, cause of death. Provincial linkage for vital status and cause of death was done for the period from January 1, 1998 to December 31, 2007 for the Quebec cohort using the mortality files of Quebec held by the Quebec Institute of Statistics

108 and the Régie d’Assurance Maladie du Québec. The linkage for vital status and cause of death for the Ontario cohort was done for the period from January 1, 1998 to December 31, 2006 with the use of the Ontario Mortality Database (OMDB), held by the Registrar General of Ontario. Where no death was found, each subject was assumed to be alive and was censored at the end of follow-up.

Ascertainment of breast cancer prognostic factors

Medical records of breast cancer cases detected through the linkages described above were reviewed. This review served both to confirm the diagnosis of breast cancer, as well as to provide clinical and pathological information about stage of breast cancer at diagnosis, tumor size, axillary lymph node involvement and tumor histology. Stage of breast cancer at diagnosis was classified according to the TNM 6th edition (45). Pathological TNM stage group was assigned to the large majority of these cases, augmented by the clinical stage group where the pathologic stage was not available. Additionally, information on body mass index (BMI) was collected among a sub-sample (N=265) of women diagnosed with breast cancer after 1998 in Quebec.

Among the 40,451 women, 409 women with breast implants and 444 non-augmented women were diagnosed with breast cancer. Women diagnosed with Ductal Carcinoma in situ (DCIS) were excluded from the analysis, so as to be consistent with our earlier paper (21) and previous investigations (9;12).

Statistical analysis

The evaluation of stage distribution of breast cancer at diagnosis was assessed using the odds ratios (ORs) and their 95% confidence intervals (CIs) which were calculated from logistic regression models and multinomial logit models adjusting for relevant confounding variables (46). First, we compared augmented women with breast cancer to other plastic surgery women with breast cancer using multinomial logistic regression to examine the effect of breast implant on the following outcomes: stage at diagnosis (TNM stage I, II, III/IV, or unknown), tumor size (<21, 21-≤50, > 50 mm, or unknown), lymph node involvement (yes, no, or unknown) and tumor histology (infiltrating duct carcinoma,

109 lobular carcinoma, or other). Restricted analyses which included only those women who received breast implants and developed breast cancer were performed to identify possible differences in stage of breast cancer at diagnosis in terms of implant characteristics at time of implantation: type of implant (silicone gel-filled implants (SGFIs), saline, or unknown), envelope (polyurethane-coated, not polyurethane-coated, or unknown), placement (subglandular, submuscular, or unknown), and fill volume (<200, ≥ 200 cc, or unknown). The influence of the following potential confounding variables was evaluated in multivariate models: age at index surgery (18 - <25, 25 - <30, 30 - <35, 35 - <40, 40 - <45, or ≥45 years), province of residence (Ontario or Quebec), calendar period of index surgery (1974-1977, 1978-1981, 1982-1985, or 1986-1989), age at diagnosis (<45, 45 - <50, 50 - <60, or ≥60 years) and period of breast cancer diagnosis. Because Ontario started their organized breast cancer screening program in 1990 and Quebec started in 1998, period of breast cancer diagnosis was grouped into three calendar periods (<1990, 1990-1997, or ≥ 1998). Evaluation of confounding in the multivariable models was done by a backward deletion approach (47). This was done by adjusting for all potential confounders and then by removing one by one in a stepwise manner the least significant confounding variables as long as the total proportional change in odds ratio compared with the fully adjusted model was less than 10 %. Covariates that were not confounders, but increased the precision of the estimates were kept in the final model.

Multivariate Cox proportional hazards model (48) was used to compare breast cancer- specific mortality (hazard) rates between the breast implant and other plastic surgery cases and within sub-groups of the former, by implant characteristics, while adjusting for relevant confounders. Mortality was assessed until December 31, 2007 for Quebec breast cancer cases and until December 31, 2006 for Ontario. Individuals whose underlying cause of death is not breast cancer were censored at the date of their death. In these analyses, no adjustment for stage of breast cancer was made in order to capture the overall effect of implants on breast cancer-specific survival. We evaluated the assumption of proportionality by inspecting plots of the log negative log mortality curves and examining the statistical significance of time-dependent covariates in proportional hazards models (48). Survival curves to investigate differential survival between groups were produced using Cox

110 proportional hazards model after adjusting for covariates. These analyses were performed with SAS, version 9.2 (49).

A sensitivity analysis was also performed to examine the potential confounding influence of BMI in the associations described above. Increased BMI is associated with lower breast cancer survival (50) and BMI is expected to be lower in women who have breast implants (51). Thus, BMI is a potential confounder in our analyses. Specifically, we used information collected for BMI at time of diagnosis among a sub sample of women (N=265; 30% of cases) to impute (52;53) values of BMI to all other breast cancer cases (54). We generated 70 simulated datasets using linear regression to impute values of BMI (55). The following variables were used to impute values of BMI: implant status, age at diagnosis, stage at diagnosis, period of diagnosis, survival time, censoring and the interaction between survival time and censoring. The Stata software was used to generate this sub analysis (56).

Ethics approval for the study was granted by the University of Toronto’s Office on Research Studies, the ethics committee of the Centre Hospitalier Affilié universitaire de Québec’s (CHA) Saint-Sacrement Hospital and the Ethics Committee for Clinical Research of Laval University.

Results Characteristics of the 409 breast implant women with incident breast cancers and the 444 breast cancer cases among the other cosmetic surgery women is presented in table 1. These numbers include an additional 227 breast implant women and 242 other plastic surgery women with invasive breast cancers compared with our previous paper (21). Mean age at surgery was higher among the other cosmetic surgery women (38.1 vs. 36.2 years). Women with other cosmetic surgery were slightly older, on average, at breast cancer diagnosis compared with implant women (53.6 vs. 52.3 years). Over 70 % of the breast cancer cases came from the province of Quebec among both the implant women and other cosmetic surgery women. More than half (55 %) of breast cancer cases in both groups were diagnosed in the period after 1998.

111

Women who received breast implants were more likely than those with other cosmetic surgery to have stage III/IV tumor (OR = 3.04, 95 % CI: 1.81-5.10, p < 0.001) (Table 2). Additionally, breast implant women had increased odds of having nodal involvement at breast cancer diagnosis compared with controls (OR = 1.60, 95 % CI: 1.17-2.19, p = 0.004). We found no statistically significant differences between the implant and other plastic surgery women for tumor size and histological type.

The investigation of the effects of implant characteristics on breast cancer stage was done by grouping the stage categories (excluding unknown) into 2 groups (Table 3). For the purpose of this analysis, stages I/II were defined as early stage cancers, while stages III/IV were defined as advanced. This analysis showed no statistically significant differences in stage of breast cancer at diagnosis according to any of the implant characteristics.

There were 76 breast cancer deaths among 400 implant women with breast cancer and 66 breast cancer deaths among 434 surgical controls with such disease. We evaluated mortality differences between implant women and other surgery women using Cox proportional hazards model adjusting for age at diagnosis, province and period of diagnosis (data not shown). We found a non-significant increase in breast cancer-specific mortality rate as shown by the elevated hazard ratio (HR) for women with breast implants relative to surgical controls (HR = 1.32, 95 % CI: 0.94-1.83, p = 0.11). When we further adjusted for stage at diagnosis the hazard ratio decreased substantially (HR = 1.05, 95 % CI: 0.75-1.47, p = 0.78). Analyses according to implant characteristics showed no differences in breast cancer-specific mortality rates within each (all p values ≥ 0.14). Figure 1 shows the breast cancer-specific survival curves adjusted for age at diagnosis, period of diagnosis and province comparing breast implant and other surgery women. Five and ten-year breast cancer specific survival rates among implant women were 87.0% and 79.0% respectively. The corresponding numbers for the other surgery group were 91.0% and 84.8 %.

Additional analyses were performed to explore the confounding effect of BMI at time of diagnosis. When further adjusting for imputed values of BMI, the strength of the relation of implant status to stage III/IV decreased slightly (OR = 2.89, 95 % CI: 1.70 – 4.90, p <

112

0.001). Additionally, results showed that the hazard ratio for breast cancer-specific mortality increased slightly to 1.39 (95 % CI: 0.95 – 2.03, p = 0.09) when further adjusting for BMI.

Discussion

Principal findings and comparison with previous studies

The extension of our previous analysis showed that with the additional follow-up time we have more than doubled the number of incident breast cancer cases and breast cancer deaths among members of our cohort. Specifically, we have confirmed the finding of our previous work that women with breast implants were more likely to have advanced breast cancers stage at diagnosis when compared to the other cosmetic surgery women (21). In addition to our previous report, two publications by Silverstein et al. reported that breast implant women presented with more advanced stage at breast cancer diagnosis when compared with non-augmented women (19;20). Although not statistically significant, several studies showed a tendency towards advanced breast tumors at diagnosis for women who received augmentation mammaplasty (6;29-35). However, several other publications found little or no evidence that implant women were diagnosed at a later stage (9;12;22- 28;57). One possible explanation is the small number of breast cancer cases among augmented women (7 to 137 cases) which may have affected study power in these studies. Our finding of later stage at diagnosis among cases occurring in women with implants are coherent with the well-established evidence that breast implants are radiopaque and obscure some portions of breast tissue from mammographic visualization (58), even in the presence of implant displacement technique mammography (33).

Questions have been raised regarding breast cancer detection according to implant characteristics, especially regarding the placement of the implant. Specifically, implants placed under the breast glands (subglandular placement), because of their proximity with breast tissue, have been shown to obscure mammograms more so than submuscular placement (37;38). Previous reports showed that subglandular placement may obscure 39% to 49% of breast tissue compared to 9–28% for submuscular implant placement (38). In this

113 study, we have found no statistically significant differences in stage of breast cancer at diagnosis according to any of the implant characteristics, including implant placement as in our previous report (21). One possible explanation is that subglandular implants, because of their proximity with breast tissue, could serve as a base against which the mass may be easier to feel at breast examination and be more palpable compared to women with submuscular implant placement (33;58). This may compensate for the potential impairment in visualization (59;60). However, this explanation needs to be further confirmed (29;30;34).

To date, although not statistically significant, four out of five studies (including our own) indicated poorer breast cancer-specific survival among augmented women (Deapen et al., HR = 2.05 (28); Handel et al., HR = 1.81 (33)) or poorer overall survival (Holmich et al., HR = 1.54 (26)). One study indicated a small improvement in breast cancer-specific survival among augmented women (Birdsell et al., HR = 0.90 (27)). For the latter studies, hazard ratios and their 95% confidence intervals were estimated from data or Kaplan-Meier survival curves in the published manuscripts (61;62). The inverse variance weighted average of the five study specific hazard ratios available to date (63) yielded a pooled hazard ratio of 1.38 (95 % CI: 1.08 – 1.76). Thus, taken together, findings to date suggest an increased risk of breast cancer-specific mortality among augmented women with breast cancer.

Studies have consistently reported that women with cosmetic breast implants are not at an increased risk of breast cancer mortality compared to women in the general population or to women with other cosmetic surgery (40;64-69). This reduction in mortality is primarily attributable to a reduction in breast cancer incidence seen consistently in implant women. The increased breast cancer mortality following the diagnosis of such disease may not be sufficient to counterbalance the significantly reduced breast cancer incidence among these women (70).

In our breast cancer survival analysis, adjustment for stage at diagnosis could explain almost entirely the observed excess in rate of breast cancer death. This observation

114 reinforces the view that the increased breast cancer mortality observed in our study may be a true effect and that the lack of statistical significance may be associated primarily to insufficient power to detect a 30% increase in hazard ratio even in our relatively large group of cases. No other study examined differences in survival according to implant characteristics which may be explained by the relatively small series of breast cancer cases in previous studies.

Strengths and limitations of the study

Several limitations of our study need to be considered. First, comparisons were made between implant women and a control group consisting of women who received other cosmetic surgery. It is well recognized that these women are a more appropriate comparison group when studying the health effects associated with cosmetic breast implants because these women tend to be similar in terms of sociodemographic and lifestyle factors to women with breast implants (51;71). Even after using this control group and adjusting for covariables in our models, residual confounding is still possible. Our sensitivity analysis of the confounding effect of BMI at time of diagnosis suggests that this factor has minimal confounding effect in both analyses of stage distribution and breast cancer-specific mortality. Additionally, information on implant characteristics was available only at time of implantation and we did not ascertain if augmented women went through a reoperation to modify or replace their implants. In fact, recent studies have reported reoperation rates ranging from approximately 20 to 30 percent after 6 years of follow-up following breast augmentation surgery (72-75). Thus reoperation may have prevented the identification of associations of some implant characteristics with stage or mortality if characteristics of the subsequent implants are different from those of the initial implants. Furthermore, no information was available in our study regarding the method used to diagnosed breast cancers which makes it difficult to conclude whether implants themselves or breast cancer screening behaviours are responsible for differences in stage of breast cancer at diagnosis. However, women undergoing cosmetic procedures are recognized to be of higher socioeconomic status and to be health conscious (76) which are both factors correlated with having regular mammographic examinations (77). Therefore,

115 using women with other cosmetic surgery as comparison group should have limited such potential residual confounding by screening behaviours.

Another limitation of this analysis but also of all other studies of the effects of breast implants on breast cancer stage at diagnosis and survival is the relative lack of statistical power, in particular, when comparing augmented women to non-augmented women. The combination, in a meta-analysis, of all studies published to date on breast cancer stage and survival may be the best approach to take full advantage of all currently available information on these two associations.

Strengths of our study include the largest sample size to date to study the effect of breast implants on stage of breast cancer at diagnosis and subsequent survival. In addition, the long follow-up period for survival analyses, the detailed information on implant characteristics and several breast cancer prognostic factors, and the use of a more comparable control group for comparisons with breast implant women are also strengths of our study.

This study provided further evidence that women with breast implants have a higher likelihood of being diagnosed with advanced breast cancers compared with non-augmented women. No differences were observed for stage at diagnosis according to implant characteristics. The more advanced stage at diagnosis among augmented women led to an association of breast implants with poorer breast cancer specific survival although this result did not reach statistical significance. No differences were found for breast cancer- specific survival according to implant characteristics. The number of women with breast implants is increasing and these women are ageing. Thus, an increasing number of women with breast implants will be diagnosed with breast cancer in coming years. Therefore, further investigations are required to clarify whether breast implants result in more advanced stage at breast cancer diagnosis and reduced survival among augmented women.

116

Acknowledgements

This study was supported by the Public Health Agency of Canada. We thank Drs. Yang Mao and Anne-Marie Ugnat for their contribution to this project. We thank the plastic surgeons in Ontario and Quebec for their participation. We also thank Sylvie Bérubé and Caty Blanchette from Quebec and Gemma Lee and Susitha Wanigaratne from Ontario, for their help in the design and conduct of this updated study.

117

References

(1) U.S.Food and Drug Administration Center for Devices and Radiological Health. FDA Update on the Safety of Silicone Gel-Filled Breast Implants. 2011.

(2) American Society of Plastic Surgeons. 2010 top five cosmetic surgical procedures. American Society of Plastic Surgeons 2011. Available from: URL: http://www.plasticsurgery.org/Documents/news-resources/statistics/2010- statisticss/Charts/2010-top-5-cosmetic-surgery-procedures-graph.pdf. 2011

(3) Health Canada. Drugs and health products: Notice of decision for silicone gel- filled breast implants. Health Canada 2009. Available from: URL: http://www.hc- sc.gc.ca/dhp-mps/prodpharma/sbd-smd/phase1-decision/md im/nd_ad_2006_lowbleed_69312-end.php. 2010

(4) Health Canada. It's your health. " Breast implants ". Health Canada 2009. Available from: URL: http://www.hc-sc.gc.ca/hl-vs/iyh-vsv/med/implants- eng.php. 2010

(5) U.S.Food and Drug Administration. Anaplastic Large Cell Lymphoma (ALCL) In Women with Breast Implants: Preliminary FDA Findings and Analyses. Center for Devices and Radiological Health; 2011.

(6) Brinton LA, Lubin J, Burich MC, Colton T, Brown SL, Hoover RN. Breast cancer following augmentation mammoplasty (United States). Cancer Causes Control 2000;11:819-27.

(7) Brisson J, Holowaty EJ, Villeneuve PJ, Xie L, Ugnat A-M, Latulippe L, et al. Cancer incidence in a cohort of Ontario and Quebec women having bilateral breast augmentation. Int J Cancer 2005;118:2854-62.

(8) Deapen DM, Hirsch EM, Brody GS. Cancer risk among Los Angeles women with cosmetic breast implants. Plast Reconstr Surg 2007;119:1987-92.

(9) Friis S, Holmich LR, Mclaughlin JK, Kjøller K, Fryzek JP, Henriksen TF, et al. Cancer risk among Danish women with cosmetic breast implants. Int J Cancer 2006;118(4):998-1003.

(10) Lipworth L, Tarone RE, Friis S, Ye W, Olsen JH, Nyren O, et al. Cancer among scandinavian women with cosmetic breast implants: A pooled long-term follow- up study. Int J Cancer 2009;124:490-3.

(11) Mclaughlin JK, Lipworth L, Fryzek JP, Weimin Y, Tarone RE, Nyren O. Long- term cancer risk among Swedish women with cosmetic breast implants: An update of a nationwide study. J Natl Cancer Inst 2006;98(8):557-60.

118

(12) Pukkala E, Boice Jr.JD, Hovi S-L, Hemminki E, Asko-Seljavaara S, Keskimäki I, et al. Incidence of breast cancer and other cancers among Finnish women with cosmetic breast implants, 1970-1999. J Long Term Eff Med Implants 2002;12(4):251-3.

(13) Fajardo LL, Harvey JA, McAleese KA, Roberts CC, Granstrom P. Breast cancer diagnosis in women with subglandular silicone gel-filled augmentation implants. Radiology 1995;194(3):859-62.

(14) Hayes HJr, Vandergrift J, Diner WC. Mammography and breast implants. Plast Reconstr Surg 1988;82(1):1-8.

(15) Eklund GW, Busby RC, Miller SH, Job JS. Improved imaging of the augmented breast. AJR Am J Roentgenol 1988;151(3):469-73.

(16) Eklund GW, Cardenosa G. The art of mammographic positionning. Radiol Clin North Am 1992;30(1):21-53.

(17) Schumann D. Health risks for women with breast implants. Nurse Pract 1994;19(7):19-30.

(18) Deapen DM. Breast implants and breast cancer: a review of incidence, detection, mortality, and survival. Plast Reconstr Surg 2007;120(Suppl. 1):70S-80S.

(19) Silverstein MJ, Gierson ED, Gamagami P, Handel N, Waisman JR. Breast cancer diagnosis and prognosis in women augmented with silicone gel-filled implants. Cancer 1990;66(1):97-101.

(20) Silverstein MJ, Handel N, Gamagami P, Waisman JR, Gierson ED, Rosser RJ, et al. Breast cancer in women after augmentation mammoplasty. Arch Surg 1988;123(6):681-5.

(21) Xie L, Brisson J, Holowaty E, Vileneuve PJ, Mao Y. The influence of cosmetic breast augmentation on the stage distribution and prognosis of women subsequently diagnosed with breast cancer. Int J Cancer 2010;126(9):2182-90.

(22) Brinton LA, Malone KE, Coates RJ, Schoenberg JB, Swanson CA, Daling JR, et al. Breast enlargement and reduction: results from a breast cancer case-control study. Plast Reconstr Surg 1996;97(2):269-75.

(23) Friis S, Mclaughlin JK, Mellemkjaer L. Breast implants and cancer risk in Denmark. Int J Cancer 1997;714:956-8.

(24) Deapen DM, Bernstein L, Brody GS. Are breast implants anticarcinogenic ? A 14-year follow-up of the Los Angeles study. Plast Reconstr Surg 1997;99(5):1346-53.

119

(25) Deapen DM, Brody GS. Augmentation mammaplasty and breast cancer: a 5-year update of the Los Angeles study. Plast Reconstr Surg 1992;89(4):660-5.

(26) Holmich LR, Mellemkjaer L, Gunnarsdottir KA, Tange UB, Krag C, Møller S, et al. Stage of breast cancer at diagnosis among women with cosmetic breast implants. Cancer 2003;88(6):832-8.

(27) Birdsell DC, Jenkins H, Berkel H. Breast cancer diagnosis and survival in women with and without breast implants. Plast Reconstr Surg 1993;92(5):795-800.

(28) Deapen DM, Hamilton A, Bernstein L, Brody GS. Breast cancer stage at diagnosis and survival among patients with prior breast implants. Plast Reconstr Surg 2000;105(2):535-40.

(29) Handel N, Silverstein MJ. Breast cancer diagnosis and prognosis in augmented women. Plast Reconstr Surg 2006;118:587-93.

(30) Skinner KA, Silberman H, Dougherty W, Gamagami P, Waisman J, Sposto R, et al. Breast cancer after augmentation mammoplasty. Ann Surg Oncol 2001;8(2):138-44.

(31) Miglioretti DL, Rutter CM, Celler BM, Cutter G, Barlow WE, Rosenberg R, et al. Effect of breast augmentation on the accuracy of mammography and cancer characteristics. JAMA 2004;291(4):442-50.

(32) Cahan AC, Ashikari R, Pressman P, Cody H, Hoffman S, Sherman JE. Breast cancer after breast augmentation with silicone implants. Ann Surg Oncol 1995;2(2):121-5.

(33) Handel N. The effect of silicone implants on the diagnosis, prognosis, and treatment of breast cancer. Plast Reconstr Surg 2007;120(Suppl. 1):81S-93S.

(34) Jakub JW, Ebert MD, Cantor A, Gardner M, Reintgen DS, Dupont EL, et al. Breast cancer in patients with prior augmentation: Presentation, stage, and lymphatic mapping. Plast Reconstr Surg 2004;114:1737-42.

(35) Tuli R, Flynn RA, Brill KL, Sabol JL, Usuki KY, Rosenberg AL. Diagnosis, treatment, and management of breast cancer in previously augmented women. Breast J 2006;12(4):343-8.

(36) Sarwer DB, Brown GK, Evans DL. Cosmetic breast augmentation and suicide. Am J Psychiatry 2007;164(7):1006-13.

(37) Silverstein MJ, Handel N, Gamagami P. The effect of silicone-gel-filled implants on mammography. Cancer 1991;68(S5):1159-63.

120

(38) Silverstein MJ, Handel N, Gamagami P, Waisman E, Gierson ED. Mammographic measurements before and after augmentation mammaplasty. Plast Reconstr Surg 1990;86(6):1126-30.

(39) Hoshaw SJ, Klein PJ, Clark BD, Cook RR, Perkins LL. Breast implants and cancer: causation, delayed detection, and survival. Plast Reconstr Surg 2001;107:1393-408.

(40) Villeneuve PJ, Holowaty EJ, Brisson J, Xie L, Ugnat A-M, Latulippe L, et al. Mortality among Canadian women with cosmetic breast implants. Am J Epidemiol 2006;164(4):334-41.

(41) Band PR, Gaudette LA, Hill GB, Holowaty EJ, Huchcroft SA, Johnston GM, et al. The making of the Canadian cancer registry: cancer incidence in Canada and its regions, 1969 to 1988. Ottawa, ON; 1993.

(42) Howe GR, Lindsay J. A generalized iterative record linkage computer system for use in medical follow-up studies. Comput Biomed Res 1981;14(4):327-40.

(43) Brenner DR, Tammemägi MC, Bull SB, Pinnaduwaje D, Andrulis IL. Using cancer registry data: agreement in cause-of-death data between the Ontario Cancer Registry and a longitudinal study of breast cancer patients. Chronic Dis Can 2009;30(1):16-9.

(44) Ministère de la Santé et des Services Sociaux du Québec. Fichier des tumeurs du Québec. Ministère de la Santé et des Services Sociaux du Québec 2010. Available from: URL: http://msssa4.msss.gouv.qc.ca/santpub/tumeurs.nsf/cat?OpenView. 2010

(45) American Joint Commission on Cancer. AJCC Cancer Staging Manual. 6th edition. Philadelphia: 2002.

(46) Kleinbaum DG, Kupper LL, Muller KE, Nizam A. Applied regression analysis and other multivariable methods. 3rd ed. Brooks/Cole Publishing Company; 1998.

(47) Rothman KJ, Greenland S, Lash TL. Modern epidemiology. 3rd ed. Philadelphia: Lippincott, Williams & Wilkins; 2008.

(48) Kleinbaum DG, Klein M. Survival analysis: a self leaning text. 2 ed. Atlanta, GA: Springer; 2005.

(49) SAS, ed 9.2 [computer program]. Cary, NC, USA: SAS Institute; 2008.

(50) Protani M, Coory M, Martin JH. Effect of obesity on survival of women with breast cancer: systematic review and meta-analysis. Breast Cancer Res Treat 2010;123(3):627-35.

121

(51) Brinton LA, Brown SL, Colton T, Burich MC, Lubin J. Characteristics of a population of women with breast implants compared with women seeking other types of plastic surgery. Plast Reconstr Surg 2000;105(3):919-29.

(52) Ali AM, Dawson SJ, Blows FM, Provenzano E, Ellis IO, Baglietto L, et al. Comparison of methods for handling missing data on immunohistochemical markers in survival analysis of breast cancer. Br J Cancer 2011;104(4):693-9.

(53) White IR, Royston P. Imputing missing covariate values for the Cox model. Stat Med 2009;28:1982-98.

(54) Little RA, Rubin DB. Statistical Analysis with Missing Data. New York: John Wiley and Sons; 2002.

(55) Bodner TE. What improves with increased missing data imputations? Struct Equation Model: Multidiscip J 2008;15:651-75.

(56) Royston P. Multiple imputation of missing values: Update of ice. Stata Journal 2005;5(4):527-36.

(57) Clark CP, Peters GN, O'Brien KM. Cancer in the augmented breast: diagnosis and prognosis. Cancer 1993;72(7):2170-4.

(58) Tang SS, Gui GP. A review of the oncologic and surgical management of breast cancer in the augmented breast: Diagnostic, surgical and surveillance challenges. Ann Surg Oncol 2011;Feb. 11.

(59) Cunningham B. Breast cancer diagnosis and prognosis in augmented women- Discussion. Plast Reconstr Surg 2006;118(3):587-93.

(60) Smalley SM. Breast implants and breast cancer screening. J Midwifery Womens Health 2003;48(5):329-35.

(61) Williamson PR, Tudur Smith C, Hutton JL, Marson AG. Aggregate data meta- analysis with time-to-event outcomes. Stat Med 2002;21:3337-51.

(62) Parmar MK, Torri V, Stewart L. Extracting summary statistics to perform meta- analyses of the published literature for survival endpoints. Stat Med 1998;17:2815-34.

(63) Borenstein M, Hedges L, Higgins J, Rothstein H. Introduction to meta-analysis. Chichester: John Wiley & Sons, Ltd; 2009.

(64) Brinton LA, Lubin JH, Burich C. Mortality among augmentation mammoplasty patients. Epidemiology 2001;12(3):321-6.

(65) Brinton LA, Lubin JH, Murray MC, Colton T. Mortality rates among augmentation mammoplasty patients: an update. Epidemiology 2006;17(2):162-9.

122

(66) Jacobsen PH, Holmich LR, Mclaughlin JK, Johansen C, Olsen JH, Kjøller K, et al. Mortality and suicide among Danish women with cosmetic breast implants. Arch Intern Med 2004;164(22):2450-5.

(67) Koot VCM, Peeters PHM, Granath F, Grobbee DE, Nyren O. Total and cause- specific mortality among Swedish women with cosmetic breast implants: prospective study. BMJ 2003;326(7388):527-8.

(68) Lipworth L, Nyren O, Ye W, Fryzek JP, Tarone RE. Excess mortality from suicide and other external causes of death among women with cosmetic breast implants. Ann Plast Surg 2007;59(2):119-23.

(69) Pukkala E, Kulmana I, Hovi S-L, Hemminki E, Keskimäki I, Pakkanen M, et al. Causes of death among Finnish women with cosmetic breast implants, 1971-2001. Ann Plast Surg 2003;51(4):339-42.

(70) Pan SY, Lavigne E, Holowaty EJ, Villeneuve PJ, Xie L, Morrison H, et al. Canadian breast implant cohort: Extended follow-up of cancer incidence. Int J Cancer 2012;Apr 19. doi:10.1002/ijc.27603. [Epub ahead of print].

(71) Kothari AR, Birch S. Individual and regional determinants of mammography uptake. Can J Public Health 2009;95(4):290-4.

(72) Codner MA, Mejia JD, Locke MB, Mahoney A, Thiels C, Nahai FR, et al. A 15- year experience with primary breast augmentation. Plast Reconstr Surg 2011;127(3):1300-10.

(73) Cunningham B. The Mentor study on contour profile gel silicone MemoryGel breast implants. Plast Reconstr Surg 2007;120:33s-9s.

(74) Cunningham B, McCue J. Safety and effectiveness of Mentor's MemoryGel implants at 6 years. Aesthetic Plast Surg 2009;33:440-4.

(75) Spear SL, Murphy DK, Slicton A, Walker PS. Inamed silicone breast implant U.S. study group. Inamed silicone breast implant core study results at 6 years. Plast Reconstr Surg 2007;120:8s-16s.

(76) Cook RR, Perkins LL. The prevalence of breast implants among women in the United States. Curr Top Microbiol Immunol 1996;210:419-25.

(77) Barrett K, Legg J. Demographic and health factors associated with mammography utilization. Am J Health Promot 2005;19(6):401-5.

123

Table 1. Frequency distribution for selected characteristics of women diagnosed with breast cancer who received breast implants and other cosmetic surgeries, Canadian Breast Implant Cohort Study. Breast implant O ther cosmetic Characteristics women with breast cancer surgery women with breast cancer (n=409) (n=444) Time since surgery, years, N (%)1 <5 29 (7.1) 38 (8.6) 5 to <10 53 (13.0) 72 (16.2) 10 to <15 96 (23.5) 94 (21.2) 15 to <20 101 (24.7) 113 (25.5) 20 to <25 80 (19.6) 88 (19.8) ≥25 50 (12.2) 39 (8.8) Mean duration of follow up (SD), years2 16.1 (7.2) 15.5 (7.1) Period of surgery, N (%) 1974 – 1977 86 (21.0) 67 (15.1) 1978 – 1981 123 (30.1) 121 (27.3) 1982 – 1985 110 (26.9) 145 (32.7) 1986 – 1989 90 (22.0) 111 (25.0) Age at surgery, years, N (%) 18 to < 35 187 (45.7) 167 (37.6) 35 to <45 150 (36.7) 162 (36.5) ≥45 72 (17.6) 115 (25.9) Mean age at surgery (SD), years 36.2 (8.4) 38.1 (10.0) Province of residence, N (%) Ontario 101 (24.7) 125 (28.2) Quebec 308 (75.3) 319 (71.9) Age at breast cancer diagnosis, years, N (%) <45 85 (20.8) 81 (18.2) 45 to <50 85 (20.8) 78 (17.6) 50 to <60 148 (36.2) 167 (37.6) ≥60 91 (22.2) 118 (26.6)

124

Mean age at breast cancer diagnosis (SD), years 52.3 (9.6) 53.6 (9.9) Length of cancer follow-up, years, N (%) <5 141 (34.5) 136 (30.6) 5 to <10 123 (30.1) 126 (28.4) 10 to <15 89 (21.8) 116 (26.1) 15 to <20 28 (6.8) 48 (10.8) ≥20 28 (6.8) 18 (4.1) Period of diagnosis, N (%) <1990 52 (12.7) 56 (12.6) 1990 to <1998 132 (32.3) 141 (31.8) ≥1998 225 (55.0) 247 (55.6) 1. Length of interval between the date of index cosmetic surgery and breast cancer diagnosis. 2. Mean duration of follow-up between the date of index cosmetic surgery and breast cancer diagnosis.

125

Table 2. Odds ratios (ORs)1 and 95% confidence intervals (CIs) for selected characteristics of incident cases of breast cancer comparing breast implant women to other cosmetic surgery women. Implant Control Characteristics Cases Cases OR 95% CI (n=409) (n=444) TNM stage at diagnosis I 127 181 1.0 - II 158 170 1.33 0.97-1.82 III/IV 55 27 3.04 1.81-5.10 Unknown 69 66 1.53 1.01-2.33 Tumour size, mm <21 209 251 1.0 - 21 to ≤50 101 115 1.08 0.78-1.50 >50 18 17 1.29 0.64-2.61 Unknown 81 61 1.63 1.11-2.39 Nodal involvement No 175 243 1.0 - Yes 154 128 1.60 1.17-2.19 Unknown 80 73 1.51 1.02-2.24 Histology Ductal 260 288 1.0 - Lobular 59 70 0.89 0.59-1.35 Non-missing others 76 74 1.15 0.80-1.67 Unknown 14 12 1.22 0.54-2.76 1Odds ratio adjusted for age at diagnosis, period of diagnosis and province.

126

Table 3. Odds ratios (ORs)1 and 95% confidence intervals (CIs) for stage distribution of breast cancer for selected breast implant characteristics.

Stage Characteristics Stage I/II cases Stage III/IV cases OR 95% CI Type of fill Saline 44 7 1.0 - Silicone 194 35 1.47 0.57-3.80 Unknown 47 13 2.22 0.71-6.90 Polyurethane coating No 180 31 1.0 - Yes 22 4 1.73 0.52-5.79 Unknown 83 20 1.81 0.90-3.66 Fill volume (cc) <200 143 28 1.0 - ≥200 141 27 0.88 0.48-1.61 Unknown 1 0 0.00 N.E. Site of implantation Submuscular 142 31 1.0 - Subglandular 108 19 0.78 0.41-1.47 Unknown 35 5 0.56 0.20-1.59 1Odds ratio adjusted for age at diagnosis, period of diagnosis and province.

127

Figure 1. Breast cancer-specific survival curves comparing breast implant with other cosmetic surgery women.

probability

Implant women

Other cosmetic surgery women Survival

Time in months after diagnosis

1Survival curves were adjusted for age at diagnosis, period of diagnosis and province using Cox proportional hazards model (p value for differences in survival curves= 0.11)

128

Chapter 5: Breast cancer detection and survival among women with cosmetic breast implants: a systematic review and meta-analysis of observational studies

129

Résumé

Objectifs : Les objectifs de cette étude étaient d’évaluer si la distribution du stade parmi les femmes diagnostiquées avec un cancer du sein diffère entre celles ayant reçu des implants mammaires pour fins esthétiques et celles sans implants, et évaluer si l’augmentation mammaire pour fins esthétiques précédent la détection d’un cancer du sein est un prédicateur de la survie au cancer du sein. Méthodes : Une revue systématique d’études observationnelles avec deux méta-analyses a été effectuée afin d’évaluer ces objectifs. Une recherche systématique de la littérature publiée avant février 2011 a été effectuée dans MEDLINE, EMBASE, Global health, CINAHL, IPAB et PsycINFO. Les publications éligibles étaient celles qui incluaient des femmes diagnostiquées avec un cancer du sein ayant reçu une augmentation mammaire pour fins esthétiques. Résultats : Notre première méta-analyse, basée sur 12 études, a démontré aucune différence statistiquement significative pour l’association entre les implants mammaires et un stade avancé au diagnostique du cancer du sein (Rapport de Cote (RC) global d’avoir un stade non localisé au diagnostique du cancer du sein : 1,26, IC à 95%: 0,99 à 1,60; I2 = 35,6 %). Toutefois, l’effet global pour les 5 études qui ont fournies des mesures d’association ajustées pour d’importants facteurs confondants a démontré une association statistiquement significative entre les implants mammaires pour fins esthétiques et un stade avancé au diagnostique du cancer du sein (RC = 1,51, IC à 95%: 1,18 à 1,92). La seconde méta- analyse, basé sur 5 études, a évalué l’association entre l’augmentation mammaire et la survie au cancer du sein. Cette méta-analyse a identifiée une réduction de la survie au cancer du sein chez les femmes avec implants mammaires pour fins esthétiques (Rapport de Taux de mortalité spécifique au cancer du sein : 1.38, IC à 95%: 1.08 à 1.75). Conclusion : La littérature publiée à ce jour suggère que les femmes avec des implants mammaires pour fins esthétiques ont un stade avancé au diagnostique du cancer du sein lorsque les estimés sont ajustés pour d’importants facteurs confondants. De plus, cette étude démontre que l’augmentation mammaire pour fins esthétiques affecte négativement la survie parmi celles préalablement diagnostiquées avec un cancer du sein. Ces résultat doivent être interprétés avec prudence alors que certaines études inclues dans la méta-

130 analyse portant sur la survie n’ont pas ajusté pour des facteurs confondants. Davantage d’investigations sont nécessaire concernant le diagnostique et le pronostic du cancer du sein parmi les femmes avec des implants mammaires.

131

Abstract

Objectives: To evaluate whether the stage distribution among women diagnosed with breast cancer differs between those who received breast implants for cosmetic purposes and those with no implants, and to evaluate whether cosmetic breast augmentation prior to the detection of breast cancer is a predictor of post-diagnosis survival. Methods: A systematic review of observational studies with two meta-analyses was undertaken to asses these objectives. A systematic search of the literature published prior to February 2011 was conducted in MEDLINE, EMBASE, Global health, CINAHL, IPAB & PsycINFO. Eligible publications were those that included women diagnosed with breast cancer and had augmentation mammaplasty for cosmetic purposes. Results: Our first meta-analysis, based on 12 studies, suggested an association between cosmetic breast implants and later stage at breast cancer diagnosis (Overall odds ratio of having non-localized breast cancers at diagnosis: 1.26, 95 % confidence interval (CI): 0.99 to 1.60; I2 = 35.6 %). However, the overall effect for the 5 studies that provided measures of association adjusted for relevant confounding factors showed a statistically significant association between cosmetic breast implants and advanced breast cancer at diagnosis (Odds ratio = 1.51, 95 % CI: 1.18 to 1.92). The second meta-analysis, based on 5 studies, evaluated the relationship between cosmetic breast implantation and survival. This meta- analysis showed reduced breast cancer survival among women who received implants compared to those who did not (Overall breast cancer-specific mortality hazard ratio: 1.38, 95 % confidence interval (CI): 1.08 to 1.75). Conclusions: The research published to date suggests that, at diagnosis, breast cancers are at more advanced stages among those with cosmetic breast implants when adjusting for relevant covariates. Additionally, there is evidence that cosmetic breast augmentation adversely affects the survival experience of women who are subsequently diagnosed with breast cancer. These findings should be taken with caution as some studies included in the meta-analysis on survival did not adjust for potential confounders. Further investigations are warranted regarding breast cancer diagnosis and prognosis among augmented women.

132

Introduction

Cosmetic breast augmentation has become increasingly popular (1). In the US, for example, cosmetic breast augmentation was the most commonly performed cosmetic surgical procedure in 2010 with 296,000 surgeries performed (2) an increase of approximately 800 % compared with the early 1990’s. Although breast augmentation is popular, controversies about the long-term health effects of breast implants remain.

The weight of evidence from epidemiological studies indicates that cosmetic breast implants are not associated with increased breast cancer risk (3-29). Concern remains, however, that implants may impair the ability to identify breast cancer at an early stage by mammography because cosmetic breast implants are radiopaque, impairing the visualization of breast tissue with mammography and making it more difficult to detect breast cancer at an early stage (30-33). Specialized radiographic techniques have been developed for women with breast implants to improve visualization which involve displacing the implant posteriorly against the chest wall and pulling breast tissue over and in front of the implant (33-36). However, one-third of the breast is still not adequately visualized despite such techniques, leading to an increase of false-negative mammograms (32). It is estimated that 1 in 8 women will be diagnosed with breast cancer some time in their lives (37). Therefore, a number of augmented women will eventually develop breast cancer which raises concerns regarding possible effects of implants on breast cancer detection.

Most studies that evaluated the detection of breast cancer among women with cosmetic breast implants compared the stage distribution of breast cancer at diagnosis between augmented and non-augmented women. The findings from these studies have been inconsistent. For instance, some studies reported that women with breast augmentation may be more likely to be diagnosed with advanced cancers (31;38-40) while others have reported no such difference (4;5;9;11;14;15;17;27;29;41-49). These conflicting results may be explained by methodological issues within studies as well as the small number of incident breast cancer cases in these studies which limit statistical power to obtain

133 significant results. In addition to the question of breast cancer detection, no study to date has been able to establish that women with breast implants, although they may be diagnosed at a more advanced stage, have a poorer survival following breast cancer diagnosis compared with non-augmented women, but these studies were also impaired by relatively low statistical power (40;42-45;48). Better understanding of the detection of breast cancer and survival patterns following breast cancer diagnosis among augmented women will aid in giving clear information on the consequences of breast augmentation surgery to these women and their physicians. The fact that implants may interfere with the early detection of breast cancer is particularly relevant and carries with it important clinical and public health implications.

Recent reviews that summarized the evidence of the long term effects of cosmetic breast implants concluded they were not associated with advanced breast cancers nor was survival affected (12;24). Although these papers were an important step forward, they were not presented as systematic reviews and were based on a qualitative rather than quantitative analysis. Through a systematic literature search, we have identified additional papers published in the past decades that were not captured by previous reviews as well as two more recent publications (40;48) providing suitable data for a quantitative meta-analysis on the diagnosis and prognosis of breast cancer among augmented women.

Specifically, our objectives were to verify the stage distribution of breast cancer and post- diagnosis survival among cosmetic breast implant women compared to non-augmented women by means of a systematic review and meta-analyses. We also sought to identify sources of heterogeneity in risk estimates in the existing literature and identify gaps in the current state of knowledge. This investigation is important in order to consolidate the existing knowledge on the long term effects of cosmetic breast implants.

134

Materials and methods

Search strategy

To identify eligible studies published before February 1st 2012, we applied a systematic literature search strategy to the following electronic databases: MEDLINE, EMBASE, Global health, CINAHL, IPAB & PsycINFO. The Cochrane Library Database of Systematic Reviews was also searched. The following keywords and subject headings were used in combination to identify relevant articles in electronic databases: breast neoplasms AND (breast implants OR breast augmentation OR mammaplasty OR mammoplasty OR breast implantation OR breast prosthesis) AND (women without implants OR non augmented women) AND (delayed diagnosis OR prognosis OR survival OR delayed detection OR staging). Reference lists from retrieved articles (4;5;9;11;14;15;17;29;31;38- 46;48;49) and published reviews (12;16;24;26) were manually examined to identify additional manuscripts. Eligible articles were original peer-reviewed published studies. Abstracts from identified articles were reviewed to assess eligibility. Additionally, as direct contact with experts has been shown to be an effective method of retrieving relevant articles, we surveyed international experts who published papers on breast cancer detection among women with cosmetic breast implants and associated survival rate patterns to request any relevant published or unpublished scientific articles (50). The search was limited to French and English articles.

Study eligibility

Eligible publications were those that included women diagnosed with breast cancer and had antecedent augmentation mammaplasty for cosmetic purposes. The comparison group consisted of women diagnosed with breast cancer who have had other common elective cosmetic surgeries or who were from the general female population.

Eligible publications for the evaluation of the association of breast implants with the stage distribution of breast cancer had to include the number of women with breast implants diagnosed with breast cancer (exposed group) and women without implants diagnosed with breast cancer (unexposed group) per stage of breast cancer at diagnosis or per status of

135 nodal involvement and/or metastases. Measures of association describing the risk or the odds of having non-localized breast tumors (nodal involvement positivity and/or metastases to distant sites) comparing the exposed breast cancer cases to the unexposed cases were used if provided in the paper. Otherwise, the crude odds ratios, their respective standard errors and 95 % confidence intervals were computed from the contingency tables.

Publications eligible for the evaluation of breast implants and survival following breast cancer diagnosis provided either hazard ratios comparing the breast cancer mortality rate after diagnosis between the exposed and unexposed group or provided Kaplan-Meier breast cancer specific survival curves graphically for both augmented women with breast cancer and non-augmented women with breast cancer. For the latter studies, hazard ratios were estimated from published statistics provided in the manuscripts or by extracting data directly from Kaplan-Meier survival curves using recommended techniques for time to event meta-analysis (51;52).

When several publications were available for the same study group, the most recent one was retained for analysis. Publications without a comparison group for the implant subjects were excluded. So were those that did not provide measures of association and did not provide crude numbers in contingency tables allowing calculation of measures of association regarding stage of breast cancer at diagnosis.

A certified librarian performed the search and two authors (E.L.; S.Y.P.) excluded studies at the first stage of eligibility evaluation. Studies identified for a more detailed assessment based on the abstract and title were discussed. Study inclusion in the meta-analysis was agreed upon by co-authors without blinding to study characteristics.

Data abstraction

Study characteristics that were extracted and reviewed included the following: source of data on implant, source of data on breast cancer diagnosis, source of data on mortality, assessment of stage of breast cancer, nodal involvement, type of comparison group, number of women with implants with breast cancer, number of women without implants with breast

136 cancer, mean length of follow-up, average age at breast cancer diagnosis, whether statistical adjustment for potential confounders was done (e.g. adjusting for age at diagnosis, period of diagnosis) and results (cell counts, odds ratios, hazard ratios and 95 % confidence intervals). Adjusted estimates were always selected over unadjusted estimates when provided in the paper. A dichotomous variable was created for the exposure variable (presence of breast implants with breast cancer vs. no implants with breast cancer) and for the outcome (stage distribution of breast cancer). Staging of breast cancer among the studies included principally the American Joint Committee on Cancer (AJCC)’s Tumor Node Metastasis (TNM) classification (53;54) without limiting the edition of the AJCC classification that was used and the U.S National Cancer Institute’s Surveillance Epidemiology and End Results (SEER) Summary Stage System (55). For the purpose of consistency in the analyses, we excluded, when possible, ductal carcinoma in situ (DCIS), lobular carcinoma in situ (LCIS) and all other non-invasive breast cancers. This was done because several publications included in our meta-analyses excluded these cases on the basis of data quality issues related to the reporting of in situ cancers. Therefore, only invasive breast cancers were considered. Considering there is a considerable variability across studies for the classification used for stage distribution of breast cancer, especially between the AJCC’s TNM and SEER’s Summary Stage System, we dichotomized the response variable as non-localized breast cancer (regional or distant) vs. localized breast cancer. This cutoff was first chosen for clinical relevancy because localized breast tumors are potentially more curable than non-localized tumors and associated with better survival rates (56). Additionally, we chose this cut off for compatibility purposes between the two cancer staging systems. Irrespective of classification system, breast tumors that did not spread to regional lymph nodes and/or distant sites were considered localized tumors. Therefore, non-localized breast tumors were considered advanced or later stage breast cancers.

Statistical analyses

The statistical analyses, forest plots, sensitivity, meta-regression and publication bias analyses were produced with Stata software, version 11 (57). Studies had to provide sufficient data to calculate an effect-size measure in order to be included in the quantitative

137 analysis. Dersimonian-Laird random effects model (58) was used to derive a pooled effect across studies for the association between cosmetic breast implants and stage distribution of breast cancer. The random effects model was used because it accounts for variations between studies in addition to sampling error within studies (59). All analyses were conducted on the natural log scale. The summary odds ratio with 95 % confidence interval was calculated from study-specific adjusted odds ratios taken directly from the study or estimated as crude odds ratios from cell counts. Study-specific confidence intervals were also taken directly from the study if reported or were calculated using the corresponding standard error. A random effects model was also used to calculate the summary hazard ratio for the association between cosmetic breast implants and survival. The pooled effect was calculated from study-specific hazard ratios that were obtained directly from the study or calculated from survival curves. In order to quantify the degree of heterogeneity across studies, we used Cochrane’s Q test (60) and the Higgins’ I-squared statistic with 95 % confidence intervals (61). The latter statistic indicates the proportion of the variance attributable to between-study variability (61). We also analyzed the influence of individual studies by omitting each study one by one in order to identify studies contributing disproportionately to the observed heterogeneity. A visual inspection was also performed using the Galbraith plot (62;63) to detect possible outlier studies that have an excessive influence on the overall estimate. To identify potential sources of heterogeneity, we examined only the association between cosmetic breast implants and stage distribution of breast cancer as there were too few studies to examine sources of heterogeneity for the association between cosmetic breast implants and survival. This was done by calculating a summary odds ratio across strata of factors selected a priori as potentially related to study quality and that were present enough across studies to perform the stratification. These factors included source of comparison group (other cosmetic surgery controls vs. population-based controls), source of exposure data (plastic surgeon records vs. medical records), breast cancer staging system (TNM staging vs. SEER staging) and statistical adjustment of the OR for potential confounders (adjusted vs. unadjusted). An evaluation of the impact of these factors on heterogeneity was also done using random-effects meta- regression models. The latter investigates how a categorical or continuous characteristic at the study level is associated with the effect estimate in the meta-analysis (64). The outcome

138 variable in the meta-regression models in this study is the odds ratio and the explanatory variables, also called potential effect modifiers, are the factors selected a priori as potentially related to study quality. The year of publication of each study as a potential source of heterogeneity was also evaluated as a continuous variable in a meta-regression model. Assessment of publication bias for the association between cosmetic breast implants and stage distribution of breast cancer was done using a funnel plot and Egger’s test(65). There were not enough studies to examine publication bias for the association between cosmetic breast implants and survival.

Results

We identified 267 unique papers after searching MEDLINE, EMBASE, Global health, CINAHL, IPAB & PsycINFO. Of these, 22 studies (n=28,924 women) met eligibility for the evaluation of stage distribution of breast cancer and breast implants and 6 studies (n=18,026 women) met eligibility for the evaluation of survival following breast cancer diagnosis and breast implants (Figure 1). One study, Xie et al. (40), included in both the meta-analysis on stage distribution of breast cancer and the one on breast cancer survival was recently up-dated and updated results (unpublished work) are used in the meta- analyses (66). No papers were identified through further investigation of previous reviews or through manual examination of references or querying of the experts. A quality assessment scale (67) also showed that publications that met eligibility are of acceptable quality to be included in these meta-analyses (supplemental file in appendix 1).

Breast implants and stage distribution of breast cancer

Twelve studies, all being cross-sectional in their design, provided sufficient data to be included in a meta-analysis to evaluate the association between cosmetic breast implants and stage distribution of breast cancer. Characteristics of the 12 publications meeting the inclusion criteria and selected for the quantitative analysis are provided in Table 1. Most of these were published after the year 2000 and were conducted in the United-States. The remainder were conducted in northern Europe or Canada. The other 10 papers were all

139 excluded because they overlapped with more recent publications with an extended follow- up of the same study group.

Results of the meta-analysis are depicted in Figure 2. The size of each box indicates the relative weight of each publication in the meta-analysis and the bars show the 95 % confidence intervals (CIs). Based on the 12 studies, the summary odds ratio with the random effects model was 1.26 (95 % CI: 0.99 to 1.60) for a non-localized stage of breast cancer at diagnosis comparing augmented women with breast cancer to non-augmented women with breast cancer. Moderate heterogeneity was observed (Q = 17.07, P = 0.11, I2 = 35.6 %).

Sensitivity analysis and publication bias

Sensitivity analyses revealed that one publication, Clark et al. (41), accounted for all the observed heterogeneity. When all 12 studies were included in the analysis I2 equalled 35.6 %. When this paper was removed, I2 equalled 0 % suggesting that it had an important influence on the overall estimate (64). Additionally, the Galbraith plot showed that Clark et al. (41) contributed disproportionately to the observed heterogeneity (data not shown). The summary odds ratio using the random effects model when that study was excluded was 1.42 (95 % CI: 1.19 to 1.68). The omission of all other studies separately resulted in random variation around the overall estimate of the remaining 11 studies. Moreover, sensitivity analyses showed that the omission of five studies (17;41;43;47;48) for which in situ breast cancers could not be excluded resulted in an overall estimate similar to the one mentioned above (OR = 1.40, 95 % CI: 1.10 to 1.78). The Egger’s test did not indicate the presence of publication bias (p = 0.16). Additionally, a visual inspection of the funnel plot did not suggest publication bias as the studies were distributed symmetrically about the combined effect size (data not shown).

140

Meta-regression analysis

Stratification by variables potentially related to study quality was evaluated and results are provided in Table 2. Overall odds ratios stratified by these variables did not produce meaningful differences, except for the studies that provided adjusted estimates for relevant covariates such as age at breast cancer diagnosis and calendar period of diagnosis (OR = 1.51, 95 % CI: 1.18 to 1.92). Statistical significance of meta-regression models is also shown in table 2. Inclusion of these variables one at a time in random-effects meta- regression model did not reach statistical significance.

Breast implants and survival

A description of the 5 publications, all with a cohort design, included in the meta-analysis for the association between cosmetic breast implants and breast cancer-specific survival following breast cancer diagnosis is shown in Table 3. The only paper that was excluded overlapped a more recent publication.

The results of the meta-analysis of cosmetic breast implants and breast cancer-specific mortality are illustrated in Figure 3. The overall hazard ratio comparing the breast cancer- specific mortality post-diagnosis between augmented women with breast cancer to non- augmented women with breast cancer was 1.38 (95 % CI: 1.08 to 1.75) with no heterogeneity observed (Q = 3.35, P = 0.50, I2 = 0.0 %). The Egger’s test did not indicate the presence of publication bias (p = 0.84). Omission of the publication that assessed overall mortality resulted in random variation around the overall estimate of the 4 remaining studies. There were not enough studies to do a meta-regression.

Discussion

Principal findings

141

This systematic review indicates that women with cosmetic breast implants have later stage at breast cancer diagnosis when controlling for relevant covariates. This finding can be explained by multiple mechanisms, the first being that both silicone and saline implants create radiopaque shadows on mammograms which impairs the visualization of breast tissue (68). The amount of parenchymal breast tissue obscured at mammography by the implant is known to be between 22 and 83 % (69). Additionally, insufficient compression of the breast to visualize the parenchyma and the production of implant-related artifacts on the film can also make it difficult to interpret mammographic exams in women with augmented breasts (32;34). It has been also shown that capsular contracture, which develops in about 15 to 20 % of augmented women, can reduce mammographic sensitivity by 30 to 50 % (68). Furthermore, it has been suggested that specific breast implant characteristics might affect the detection of breast cancer (70). Specifically, implants placed under the breast glands (subglandular placement), because of their proximity with breast tissue, are suspected to obstruct mammographic visualization of the breast more so than submuscular placement (30;71). However, to date, only one study has been able to evaluate the stage distribution of breast cancer according to implant placement (40). Results from this study were inconclusive.

Despite the fact that implant displacement techniques are widely used with mammography, studies suggest that breast tissue is still not adequately visualized (48;68). A recent report suggests that breast magnetic resonance imaging (MRI) may be a helpful diagnostic tool for women with breast implants because it allows examination of all breast tissue surrounding the implant (72) and, thus, has greater sensitivity than mammography (73). Thus, the use of MRI as a screening modality among augmented women at high risk of breast cancer should be considered.

Stratification and meta-regression models showed that no potential factors seem to unduly affect study results. When we calculated a summary effect for the three studies that used women who received other elective cosmetic surgery (chemical peel or dermabrasion, coronal brow lift, otoplasty, rhinoplasty, rhytidectomy or blepharoplasty) as the comparison group, we observed a 1.53 (95 % CI: 0.89 to 2.64) increased odds of non-localized breast

142 cancers among women with cosmetic breast implants. Women with cosmetic breast implants were found to have a 1.16 (95 % CI: 0.88 to 1.53) increased odds of non-localized breast cancers when compared with women in the general female population. Women with other cosmetic surgery are recognized in the scientific literature to be a more appropriate comparison group when studying the health effects associated with cosmetic breast implants because they tend to be more similar in terms of sociodemographic and lifestyle factors as well as health consciousness than women in the general population (74;75). For example, women seeking cosmetic surgeries could have better screening and self examination practices than women in the general female population which would translate in higher chances of being diagnosed after screening mammography if a breast tumor is present. This suggests that using other women with cosmetic surgery would be more adequate in terms of controlling potential confounders. Moreover, studies with adjustment for confounding factors such as the age of breast cancer at diagnosis yielded statistically significant effects stronger than those without adjustment (OR 1.51, 95 % CI: 1.18 to 1.92 compared to OR 1.07, 95 % CI: 0.74 to 1.55). In fact, one study showed that a lack of adjustment for the age at which breast cancer was diagnosed underestimates the measure of association (40). This outlines the importance of providing adjusted estimates. In meta-regression models we were not able to detect the modifying effect of the type of comparison group and the adjustment for cofactors, but a lack of statistical power due to the small number of studies may explain why the above differences in odds ratios are not statistically significant.

It has been hypothesized that the long term presence of cosmetic breast implants causes atrophy, thinning, and compression of the breast parenchyma which may facilitate the detection of palpable breast tumors on physical examination (17;29;45;48). The breast implants could serve as a base against which the mass may be more likely differentiated (76). This suggests that tumors of equal size may be more easily palpated in augmented patients, especially for implants placed in the subglandular position (69), and this may compensate somewhat for the potential impairment of mammography. However, very few studies have evaluated this issue providing no conclusive results (17;29;45). Furthermore, the fact that augmented women present more often with palpable tumors could also be

143 because of the smaller native breast volumes making tumors more pronounced with palpation (41).

In our second meta-analysis, the results demonstrate a higher risk of breast cancer-specific mortality among augmented women with breast cancer when compared to non-augmented women with breast cancer. Nevertheless, the overall estimate should still be interpreted with caution because this meta-analysis included a relatively small number of studies. Of concern, 3 of 5 studies had unadjusted hazard ratios (not adjusted for age at diagnosis, or period of diagnosis) and all 5 studies were unadjusted for other important factors such as cancer treatments and comorbidities which could translate into a biased estimate of the summary hazard ratio. Moreover, one study (42) included in this meta-analysis assessed overall mortality rather than breast cancer-specific mortality which could have biased our summary estimate towards the null. The small number of studies and insufficient amount of follow-up time in these studies are suspected to limit statistical power to clearly evaluate survival rate patterns among augmented women. Given the limited evidence, no conclusion regarding breast cancer-specific survival can be drawn and continued follow-up to further evaluate this issue is particularly relevant.

Limitations of study

Our study has several limitations specific to our analysis and to the different methods used across studies. For instance, certain studies used in both meta-analyses included in situ breast cancer cases because we were not able to exclude them based on the limited information available in the papers (17;41;43;47;48). This could have resulted in a non differential misclassification bias of the outcome variable. One study, Clark et al. (41), was responsible for all observed heterogeneity in the analysis of stage distribution of breast cancer at diagnosis. We believe that the results of this publication may be affected by a selection bias which is supported by the quality assessment scale (supplemental file in appendix 1). In fact, pooling results from several observational studies has the advantage of increasing statistical power, but not internal validity. Indeed, misclassification bias, selection bias and assessment of confounding affecting individual studies will also affect the meta-analyses. Misclassification biases within each study could also be a factor

144 affecting study-specific measures of association and consequently our pooled effect. For example, the identification of breast cancer deaths could be affected by data quality issues such as a misclassification of the cause of death. This could result in biased estimates of breast cancer-specific survival resulting in an underestimation of our pooled effect. Although we have evaluated the quality of the studies using an assessment scale (67), no threshold scores were available in order to distinguish between ‘good’ and ‘poor’ quality studies which can limit our results as we may have included studies of poorer quality in our analyses. Another limitation of our study could be related with the methods used to pool the hazard ratios from available data in each study which may have underestimated the variance of the estimates (52).

Conclusions and implications

Our results should be interpreted with caution considering the current gaps and limitations in the available literature. The accumulating evidence suggests that women with cosmetic breast implants that develop breast cancer have an increased risk of being diagnosed with non-localized breast tumors more frequently than non augmented women with breast cancer. Current evidence also suggests that cosmetic breast implants adversely affect breast cancer-specific survival following the diagnosis of such disease. Further investigations are warranted into the study of the long term effects of cosmetic breast implants on detection and prognosis of breast cancer adjusting for potential confounders.

145

References

(1) Love S, Lindsey K. Variations in development. Dr. Susan Love's breast book. Cambridge: Perseus Publishing; 2000.

(2) 2010 top five cosmetic surgical procedures. American Society of Plastic Surgeons 2011. Available from: URL: http://www.plasticsurgery.org/Documents/news- resources/statistics/2010-statisticss/Charts/2010-top-5-cosmetic-surgery- procedures-graph.pdf. 2011.

(3) Berkel H, Birdsell DC, Jenkins H. Breast augmentation: a risk factor for breast cancer? N Engl J Med 1992;326:1649-53.

(4) Brinton LA, Malone KE, Coates RJ, Schoenberg JB, Swanson CA, Daling JR, et al. Breast enlargement and reduction: results from a breast cancer case-control study. Plast Reconstr Surg 1996;97(2):269-75.

(5) Brinton LA, Lubin J, Burich MC, Colton T, Brown SL, Hoover RN. Breast cancer following augmentation mammoplasty (United States). Cancer Causes Control 2000;11:819-27.

(6) Brisson J, Holowaty EJ, Villeneuve PJ, Xie L, Ugnat A-M, Latulippe L, et al. Cancer incidence in a cohort of Ontario and Quebec women having bilateral breast augmentation. Int J Cancer 2005;118:2854-62.

(7) Bryant H, Brasher P. Breast implants and breast cancer - reanalysis of a linkage study. N Engl J Med 1995;332(23):1535-9.

(8) Deapen DM, Pike MC, Casagrande JT, Brody GS. The relationship between breast cancer and augmentation mammaplasty: an epidemiologic study. Plast Reconstr Surg 1986;77(3):361-8.

(9) Deapen DM, Brody GS. Augmentation mammaplasty and breast cancer: a 5-year update of the Los Angeles study. Plast Reconstr Surg 1992;89(4):660-5.

(10) Deapen DM, Brody GS. Augmentation mammaplasty and breast cancer: a five year update of the Los Angeles study. J Clin Epidemiol 1995;48(4):551-6.

(11) Deapen DM, Bernstein L, Brody GS. Are breast implants anticarcinogenic ? A 14-year follow-up of the Los Angeles study. Plast Reconstr Surg 1997;99(5):1346-53.

(12) Deapen DM. Breast implants and breast cancer: a review of incidence, detection, mortality, and survival. Plast Reconstr Surg 2007;120(Suppl. 1):70S-80S.

(13) Deapen DM, Hirsch EM, Brody GS. Cancer risk among Los Angeles women with cosmetic breast implants. Plast Reconstr Surg 2007;119:1987-92.

146

(14) Friis S, Mclaughlin JK, Mellemkjaer L. Breast implants and cancer risk in Denmark. Int J Cancer 1997;714:956-8.

(15) Friis S, Holmich LR, Mclaughlin JK, Kjøller K, Fryzek JP, Henriksen TF, et al. Cancer risk among Danish women with cosmetic breast implants. Int J Cancer 2006;118(4):998-1003.

(16) Hoshaw SJ, Klein PJ, Clark BD, Cook RR, Perkins LL. Breast implants and cancer: causation, delayed detection, and survival. Plast Reconstr Surg 2001;107:1393-408.

(17) Jakub JW, Ebert MD, Cantor A, Gardner M, Reintgen DS, Dupont EL, et al. Breast cancer in patients with prior augmentation: Presentation, stage, and lymphatic mapping. Plast Reconstr Surg 2004;114:1737-42.

(18) Kern KA, Flannery JT, Kuehn PG. Carcinogenic potential of silicone breast implants: a Connecticut statewide study. Plast Reconstr Surg 1997;100(3):737-47.

(19) Lipworth L, Tarone RE, Friis S, Ye W, Olsen JH, Nyren O, et al. Cancer among scandinavian women with cosmetic breast implants: A pooled long-term follow- up study. Int J Cancer 2008;124:490-3.

(20) Malone KE, Stanford JL, Daling JR, Voigt LF. Implants and breast cancer. Lancet 1992;339(8805):1365.

(21) Mclaughlin JK, Fraumeni JF Jr., Olsen JH, Mellemkjaer L. Breast implants, cancer, and systemic sclerosis. J Natl Cancer Inst 1994;86(18):1424.

(22) Mclaughlin JK, Nyren O, Blot WJ, Yin L, Josefsson S, Fraumeni JF Jr, et al. Cancer risk among women with cosmetic breast implants: a population based cohort study in Sweden. J Natl Cancer Inst 1998;90(2):156-8.

(23) Mclaughlin JK, Lipworth L, Fryzek JP, Weimin Y, Tarone RE, Nyren O. Long- term cancer risk among Swedish women with cosmetic breast implants: An update of a nationwide study. J Natl Cancer Inst 2006;98(8):557-60.

(24) Mclaughlin JK, Lipworth L, Murphy DK, Walker PS. The safety of silicone gel- filled breast implants. A review of the epidemiologic evidence. Ann Plast Surg 2007;59(5):569-80.

(25) Mellemkjaer L, Kjoller K, Friis S, Mclaughlin JK, Høgsted C, Winther JF, et al. Cancer occurence after cosmetic breast implantation in Denmark. Int J Cancer 2000;88:301-6.

(26) Meunier A, Tristant H, Sinna R, Delay E. Implants mammaires et cancer du sein. Annales de chirurgie plastique esthétique 2005;50:595-604.

147

(27) Pukkala E, Boice Jr.JD, Hovi S-L, Hemminki E, Asko-Seljavaara S, Keskimäki I, et al. Incidence of breast cancer and other cancers among Finnish women with cosmetic breast implants, 1970-1999. J Long Term Eff Med Implants 2002;12(4):251-3.

(28) Schumann D. Health risks for women with breast implants. Nurse Pract 1994;19(7):19-30.

(29) Skinner KA, Silberman H, Dougherty W, Gamagami P, Waisman J, Sposto R, et al. Breast cancer after augmentation mammoplasty. Ann Surg Oncol 2001;8(2):138-44.

(30) Silverstein MJ, Handel N, Gamagami P. The effect of silicone-gel-filled implants on mammography. Cancer 1991;68(S5):1159-63.

(31) Silverstein MJ, Handel N, Gamagami P, Waisman JR, Gierson ED, Rosser RJ, et al. Breast cancer in women after augmentation mammoplasty. Arch Surg 1988;123(6):681-5.

(32) Handel N, Silverstein MJ, Gamagami P, Jensen JA, Collins A. Factors Affecting Mammographic Visualization of the Breast After Augmentation Mammaplasty. JAMA 1992;268(14):1913-7.

(33) Fajardo LL, Harvey JA, McAleese KA, Roberts CC, Granstrom P. Breast cancer diagnosis in women with subglandular silicone gel-filled augmentation implants. Radiology 1995;194(3):859-62.

(34) Hayes HJr, Vandergrift J, Diner WC. Mammography and breast implants. Plast Reconstr Surg 1988;82(1):1-8.

(35) Eklund GW, Busby RC, Miller SH, Job JS. Improved imaging of the augmented breast. AJR Am J Roentgenol 1988;151(3):469-73.

(36) Eklund GW, Cardenosa G. The art of mammographic positionning. Radiol Clin North Am 1992;30(1):21-53.

(37) Altekruse SF, Kosary CL, Krapcho M, Neyman N, Aminou R, Waldron W, et al. SEER Cancer Statistics Review, 1975-2007. Bethesda, MD: National Cancer Institute; 2010.

(38) Silverstein MJ, Gierson ED, Gamagami P, Handel N, Waisman JR. Breast cancer diagnosis and prognosis in women augmented with silicone gel-filled implants. Cancer 1990;66(1):97-101.

(39) Silverstein MJ, Handel N, Gamagami P, Gierson ED, Furmanski M, Collins AR, et al. Breast cancer diagnosis and prognosis in women following augmentation with silicone gel-filled prostheses. Eur J Cancer 1992;92(28(2-3)):635-40.

148

(40) Xie L, Brisson J, Holowaty E, Vileneuve PJ, Mao Y. The influence of cosmetic breast augmentation on the stage distribution and prognosis of women subsequently diagnosed with breast cancer. Int J Cancer 2010;126(9):2182-90.

(41) Clark CP, Peters GN, O'Brien KM. Cancer in the augmented breast: diagnosis and prognosis. Cancer 1993;72(7):2170-4.

(42) Holmich LR, Mellemkjaer L, Gunnarsdottir KA, Tange UB, Krag C, Møller S, et al. Stage of breast cancer at diagnosis among women with cosmetic breast implants. Cancer 2003;88(6):832-8.

(43) Birdsell DC, Jenkins H, Berkel H. Breast cancer diagnosis and survival in women with and without breast implants. Plast Reconstr Surg 1993;92(5):795-800.

(44) Deapen DM, Hamilton A, Bernstein L, Brody GS. Breast cancer stage at diagnosis and survival among patients with prior breast implants. Plast Reconstr Surg 2000;105(2):535-40.

(45) Handel N, Silverstein MJ. Breast cancer diagnosis and prognosis in augmented women. Plast Reconstr Surg 2006;118:587-93.

(46) Miglioretti DL, Rutter CM, Celler BM, Cutter G, Barlow WE, Rosenberg R, et al. Effect of breast augmentation on the accuracy of mammography and cancer characteristics. JAMA 2004;291(4):442-50.

(47) Cahan AC, Ashikari R, Pressman P, Cody H, Hoffman S, Sherman JE. Breast cancer after breast augmentation with silicone implants. Ann Surg Oncol 1995;2(2):121-5.

(48) Handel N. The effect of silicone implants on the diagnosis, prognosis, and treatment of breast cancer. Plast Reconstr Surg 2007;120(Suppl. 1):81S-93S.

(49) Tuli R, Flynn RA, Brill KL, Sabol JL, Usuki KY, Rosenberg AL. Diagnosis, treatment, and management of breast cancer in previously augmented women. Breast J 2006;12(4):343-8.

(50) McManus RJ, Wilson S, Delaney BC, Fitzmaurice DA, Hyde CJ, Tobias RS, et al. Review of the usefulness of contacting other experts when conducting a literature search for systematic reviews. BMJ 1998;317(7172):1562-3.

(51) Williamson PR, Tudur Smith C, Hutton JL, Marson AG. Aggregate data meta- analysis with time-to-event outcomes. Stat Med 2002;21:3337-51.

(52) Parmar MK, Torri V, Stewart L. Extracting summary statistics to perform meta- analyses of the published literature for survival endpoints. Stat Med 1998;17:2815-34.

149

(53) American Joint Commission on Cancer. AJCC Cancer Staging Manual. 5th edition. Philadelphia: 1997.

(54) American Joint Commission on Cancer. AJCC Cancer Staging Manual. 6th edition. Philadelphia: 2002.

(55) Shambaugh EM, Weiss MA, Axtell LM, eds. Summary Staging Guide for the Cancer Surveillance, Epidemiology and End Results Reporting (SEER) Program. 1st Ed ed. Bethesda, US Department of Health, Education, and Welfare, Public Health Service, National Institutes of Health; 1977.

(56) Carlson RW, Allred DC, Anderson BO, Burstein HJ, Carter WB, Edge SB, et al. Breast cancer: noninvasive and special situations. J Natl Compr Canc Netw 2010;8(10):1182-207.

(57) Stata statistical software: release 11 [computer program]. College Station, TX: StataCorp LP; 2009.

(58) Dersimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7(3):177-88.

(59) Littell J, Corcoran J, Pillai V. Systematic review and meta-analysis. New York: Oxford University press; 2008.

(60) Cochran WG. The combination of estimates from different experiments. Biometrics 1954;10(1):101-29.

(61) Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21(11):1539-58.

(62) Galbraith RF. A note on graphical presentation estimated odds ratios from several clinical trials. Stat Med 1988;7:889-94.

(63) Galbraith RF. Graphical display of estimates having differing standard errors. Technometrics 1988;30(3).

(64) Borenstein M, Hedges L, Higgins J, Rothstein H. Introduction to meta-analysis. Chichester: John Wiley & Sons, Ltd; 2009.

(65) Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997;315:629-34.

(66) Lavigne E, Holowaty EJ, Pan SY, Xie L, Villeneuve PJ, Morrison H, et al. Do breast implants adversely affect prognosis among those subsequently diagnosed with breast cancer? Findings from an extended follow-up of a Canadian cohort. 2012. CMAJ. Submitted 2012.

150

(67) Wells GA, Shea B, O'Connell D, Peterson J, Welch V, Losos M, et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. Ottawa Hospital Research Institute 2011. Available from: URL: http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp. 2011.

(68) Tang SS, Gui GP. A review of the oncologic and surgical management of breast cancer in the augmented breast: Diagnostic, surgical and surveillance challenges. Ann Surg Oncol 2011;Feb. 11.

(69) Smalley SM. Breast implants and breast cancer screening. J Midwifery Womens Health 2003;48(5):329-35.

(70) Sarwer DB, Brown GK, Evans DL. Cosmetic breast augmentation and suicide. Am J Psychiatry 2007;164(7):1006-13.

(71) Silverstein MJ, Handel N, Gamagami P, Waisman E, Gierson ED. Mammographic measurements before and after augmentation mammaplasty. Plast Reconstr Surg 1990;86(6):1126-30.

(72) Uematsu T. Screening and diagnosis of breast cancer in augmented women. Breast Cancer 2008;15(2):159-64.

(73) Orel SG, Schnall MD. MR imaging of the breast for the detection, diagnosis, and staging of breast cancer. Radiology 2001;220:13-30.

(74) Brinton LA, Brown SL, Colton T, Burich MC, Lubin J. Characteristics of a population of women with breast implants compared with women seeking other types of plastic surgery. Plast Reconstr Surg 2000;105(3):919-29.

(75) Kothari AR, Birch S. Individual and regional determinants of mammography uptake. Can J Public Health 2009;95(4):290-4.

(76) Cunningham B. Breast cancer diagnosis and prognosis in augmented women- Discussion. Plast Reconstr Surg 2006;118(3):587-93.

151

Table 1. Characteristics of 12 studies selected for quantitative analysis to evaluate the association between cosmetic breast implants and stage distribution of breast cancer.

First author, Source of data on Source of Assessment Comparison group Women with Women Mean Average age Adjustment country, year implants data on of stage of breast without length of at breast breast breast cancer implants with breast follow-up cancer cancer breast cancer implants with (years)1 diagnosis breast cancer (years)2 Lavigne Plastic surgeon Canadian AJCC TNM Women with other 409 women 444 women 16.1 52.3 Adjusted for age at diagnosis, Canada records, hospital Cancer cosmetic surgery Localized: 195 Localized: 266 Range: Range: 25-85 period of diagnosis and 20123 discharge data Registry, diagnosed with Non-localized: Non-localized: 1-32 province of residence (Updated)* medical breast cancer 162 132 records Unknown: 52 Unknown: 46

Handel Medical records Medical AJCC TNM Women treated for 128 women 3795 women Unknown 46.8 None United-States records breast cancer at the Localized: 69 Localized: Range: 20074 same clinics as the Non-localized: 2467 29-71 implant patients 59 Non-localized: 1328

Tuli Medical records Medical AJCC TNM Women diagnosed 12 women 3565 women 12.6 49.2 Adjusted for age at diagnosis United-States records with breast cancer at In situ: 2 In situ: 809 Range: Range: 20064, ** the same hospital as Localized:5 Localized: 401 1-31 31-63 implant patients Non-localized: Non-localized: 4 202 Unknown: 1 Unknown: 2153

Friis Plastic surgeon Hospital SEER EOD Women with other 31 women 30 women Range: Unknown None Denmark records, Hospital records, cosmetic surgery Localized: 17 Localized: 13 0-30 20054 records, Danish diagnosed with Non-localized: Non-localized: Cancer breast 14 15 Registry cancer Unknown: 2

Miglioretti Mammography State cancer AJCC TNM Women diagnosed 137 women 685 women Unknown Unknown Matched (5:1) by age at United-States registries registries with breast cancer in In situ: 21 In situ: 122 diagnosis, ethnicity, 2004 the same Localized: 78 Localized: 388 mammography examination mammography Non-localized: Non-localized: within 2 years of diagnosis registry as implants 36 152 (yes/no), first or subsequent patients Unknown: 2 Unknown: 23 mammogram and mammography registry

152

Jakub Medical records Medical NSABP5 Women treated for 76 women 4186 women 14 49.5 None United-States records breast cancer at the Localized: 45 Localized: 20044 same hospital as Non-localized: 2758 implant patients 28 Non-localized: Unknown: 3 1352 Unknown: 76

Pukkala Medical records Finnish SEER EOD General female 7 women Estimates6 8.3 Unknown Standardized for age at Finland Cancer population with Localized: 4 Localized: 6.8 diagnosis, calendar period of 2002 Registry breast cancer Non-localized: Non-localized: diagnosis 2 4.8 Unknown: 1 Unknown: 0.9

Deapen Plastic surgeon Cancer SEER EOD General female 37 women External rate6 12.2 50.3 Standardized for age, United-States records registry population with In situ: 5 In situ: 3.8 Restricted to white women 2000 breast cancer Localized: 19 Localized: Non-localized: 18.5 13 Non-localized: 14.6

Brinton Plastic surgeon Medical SEER EOD Women with other 78 women 36 women 12.9 48 None United-States records, self- records, death cosmetic surgery In situ: 12 In situ: 10 20004 administered certificates with breast cancer Localized: 32 Localized: 19 questionnaire Non-localized: Non-localized: 27 6 Unknown: 7 Unknown: 1

Cahan Medical records Medical - Women treated for 22 women 611 women 10 - None United-States records breast cancer at the Localized: 13 Localized: 473 19954 same institution as Non-localized: Non-localized: implant patients 7 138 Unknown: 2 Clark Medical records Medical AJCC TNM Women diagnosed 33 women 1735 women - 43 None United-States records with breast cancer Localized: 25 Localized: 19934 and identified in the Non-localized: 1024 same breast cancer 6 Non-localized: registry as implant Unknown: 2 711 patients

153

Birdsell Health insurance Alberta AJCC TNM, General female 41 women 13,246 women 7.5 45.7 None Canada claims, medical Cancer 4th edition population with Localized: 25 Localized: Range: 19934 records Registry breast cancer Non- 2172 30-68 localized:13 Non-localized: Unknown: 3 1084 Unknown: 9990

1. Mean length of follow-up between the date of breast implantation and the earliest of breast or any other cancer diagnosis, date of death or the end of the follow-up period 2. Average age at breast cancer diagnosis among women with breast implants 3. Adjusted odds ratio obtained by the main author of this paper 4. Crude odds ratio was calculated using data in the paper 5. National Surgical Adjuvant Bowel Project staging 6. Expected cases were estimated based on information in the paper * Results from this publication are based on an updated version of the analysis. ** Results from this publication were obtained through a personal communication with corresponding author (Dr. Anne Rosenberg)

154

Table 2. Random effects overall odds ratio for the association between cosmetic breast implants and stage distribution of breast cancer stratified by variables potentially related to study quality.

Stratification variable No. of studies Overall effect (OR)1 95 % CI2 Meta-regression (p value) 3 Source of comparison group O ther non-augmented controls 9 1.16 (0.88 to 1.53) O ther cosmetic surgery controls 3 1.53 (0.89 to 2.64) 0.33

Breast cancer staging system4 SEER EOD5 4 1.11 (0.58 to 2.13) AJCC TNM6 6 1.21 (0.86 to 1.71) 0.87

O R adjustment for potential confounders Adjusted 5 1.51 (1.18 to 1.92) Unadjusted 7 1.07 (0.74 to 1.55) 0.27

Source of exposure data Medical records 8 1.18 (0.88 to 1.59) Plastic surgeon records 4 1.39 (0.86 to 2.23) 0.53

Year of publication 12 0.037 (0.00 to 0.07) 0.06 1. OR = Odds ratio 2. CI = Confidence interval 3. Statistical significance for meta-regression model 4. Two studies were not included in this analysis 5. Surveillance Epidemiology and End Results summary stage system 6. American Joint Committee on Cancer, Tumor Node Metastasis classification 7. Slope for the model including year of publication as a continuous variable

155

Table 3. Characteristics of 5 studies selected for quantitative analysis to evaluate the association between cosmetic breast implants and breast cancer-specific survival following breast cancer diagnosis.

First author, Source of data on Source of data on Reference group Women with breast Women without Mean follow- Average age at Adjustment country, breast cancer mortality implants and breast breast implants with up after breast cancer year cancer with survival breast cancer with diagnosis diagnosis probability survival probability (years)1 (years)2 Xie Canadian Cancer Canadian Mortality Women with other 400 women 434 women 7.7 52.3 Adjusted for age at Canada Registry, medical Database cosmetic surgery 98.0 % at 1 year 99.1 % at 1 year Range: 0.1- Range: 25-85 diagnosis, period of 20103 records diagnosed with 86.5 % at 5 years 90.7 % at 5 years 31.9 diagnosis and province (Updated)* breast cancer 77.2 % at 10 years 83.5 % at 10 years of residence

Handel Medical records Medical records Women treated for 120 women 3922 women 10.5 46.8 United-States breast cancer at the 99.0 % at 1 year 99.5 % at 1 year Range: 0.5- Range: 29-71 20074 same clinics as the 90.5 % at 5 years 92.5 % at 5 years 37.0 implant patients

Holmich Danish Cancer Danish Cancer General female 23 women 253 women 6.4 47.26 Denmark Registry, Hospital Registry population 95.5 % at 1 year 97.0 % at 1 year Range: 0.3- Range: 35-75 20034,5 records diagnosed with 86.0 % at 5 years 78.0 % at 5 years 15.7 breast cancer

Deapen Plastic surgeon Cancer registry, General female 37 women External rate 6.6 50.3 Standardized for age at United-States records, cancer National center for population 94.6 % at 1 year 97.9 % at 1 year Range: 0.2- diagnosis, stage at 20004 registry health statistics, 88.5 % at 5 years 84.1 % at 5 years 17.3 diagnosis and years of Death certificates diagnosis

Birdsell Health insurance Alberta Cancer General female 41 women 13,246 10.2 45.7 Canada claims, Alberta Registry population 94.0 % at 1 year 96.0 % at 1 year Range: 1.0- Range: 19934 Cancer Registry, 83.0 % at 5 years 74.0 % at 5 years 18.0 30-68 medical records 73.0 % at 10 years 62.0 % at 10 years

1. Length of interval between the date of breast cancer diagnosis and the earliest of date of death from breast cancer or the end of the follow-up period among implant women. 2. Average age at breast cancer diagnosis among women with breast implants. 3. Hazard ratio adjusted for age at diagnosis, period of diagnosis and province was provided in the paper. 4. Crude hazard ratio was calculated using data in the paper. 5. Assessed overall mortality. 6. Average age at breast cancer diagnosis among both the implant patients and comparison group. * Results from this publication are based on an updated version of the analysis (see appended material).

156

Source No. of publications

MEDLINE 433 EMBASE 143 CINAHL 8 Global health, IPAB & PsycInfo 4

267 Unique publications

245 Excluded based on title or abstract with minimal uncertainty

261 Excluded based on title or abstract with minimal uncertainty

22 Retrieved for review 10 Overlap with included studies

12 Eligible for Meta-analysis on stage distribution of breast cancer among women with cosmetic breast implants

6 Retrieved for review

1 Overlap with included studies

5 Eligible for Meta-analysis on survival post-breast cancer diagnosis among women with cosmetic breast implants

157

Figure 2. Forest plot with study-specific and random effects overall odds ratio (OR) for the association between cosmetic breast implants and stage distribution of breast cancer.

Source % (Reference) OR (95% CI) Weight

Birdsell et al., 1993 (43) 1.04 (0.53, 2.04) 8.55 Clark et al., 1993 (41) 0.35 (0.14, 0.85) 5.59 Cahan et al., 1995 (47) 1.86 (0.73, 4.76) 5.17 Deapen et al., 2000 (44) 0.87 (0.33, 2.30) 4.85 Brinton et al., 2000 (5) 2.67 (0.93, 7.64) 4.31 Pukkala et al., 2002 (27) 0.70 (0.09, 5.43) 1.28 Jakub et al., 2004 (17) 1.27 (0.79, 2.04) 13.07 Miglioretti et al., 2004 (46) 1.17 (0.75, 1.83) 14.02 Friis et al., 2005 (15) 0.71 (0.26, 1.99) 4.49 Tuli et al., 2006 (49) 1.52 (0.39, 5.88) 2.77 Handel et al., 2007 (48) 1.59 (1.11, 2.27) 17.08 Lavigne et al. 2012 (66) 1.65 (1.21, 2.25) 18.82 Overall (I-squared = 35.6%, p = 0.106) 1.26 (0.99, 1.60) 100.00

NOTE: Weights are from random effects analysis

.1 1 10 Odds Ratio

158

Figure 3. Forest plot with study-specific and random effects overall hazard ratio (HR) for the association between cosmetic breast implants and breast cancer-specific survival.

Source %

(Reference) HR (95% CI) Weight

Birdsell et al., 1993 (43) 0.90 (0.48, 1.68) 14.80

Deapen et al., 2000 (44) 2.05 (0.39, 10.80) 2.11

Holmich et al., 2003 (15) 1.54 (0.55, 4.33) 5.47

Handel et al., 2007 (48) 1.81 (1.12, 2.92) 25.29

Lavigne et al. 2012 (66) 1.32 (0.94, 1.83) 52.33

Overall (I-squared = 0.0%, p = 0.501) 1.38 (1.08, 1.75) 100.00

NOTE: Weights are from random effects analysis

.5 1 2.5 Hazard Ratio

159

Appendix

Table 1. Newcastle-Ottawa Quality assessment scale of studies for the evaluation of cosmetic breast implants with delayed detection of breast cancer. First author, country, year Selection Comparibility O utcome Lavigne, Canada, 2012 **** ** *** Handel, United-States, 2007 *** *** Tuli, United-States, 2006 *** ** Friis, Denmark, 2005 **** * ** Miglioretti, United-States, 2004 **** ** *** Jakub, United-States, 2004 *** *** Pukkala, Finland, 2002 **** ** ** Deapen, United-States, 2000 *** * *** Brinton, United-States, 2000 **** * *** Cahan, United-States, 1995 *** ** Clark, United-States, 1993 ** ** Birdsell, Canada, 1993 **** **

Table 2. Newcastle-Ottawa Quality assessment scale of studies for the evaluation of cosmetic breast implants with survival following breast cancer diagnosis. First author, country, year Selection Comparibility O utcome Lavigne, Canada, 2012 **** ** *** Handel, United-States, 2007 *** ** Deapen, United-States, 2000 *** *** Holmich, Denmark, 2003 **** * ** Birdsell, Canada, 1993 *** **

160

Conclusion

The results of this thesis which are based on a large retrospective cohort study provided further clarification of cancer incidence and mortality, breast cancer detection, and breast cancer-specific survival among women with cosmetic breast implants.

We observed a reduction in breast cancer incidence among women with cosmetic breast implants compared to women with other cosmetic surgery and this reduction persisted for more than 20 years after surgery. This finding is consistent with the weight of epidemiologic evidence showing reduced rates of breast cancer among augmented women (6-32).

We found, for the first time, a persistent statistically significant reduction in breast cancer incidence for women with subglandular implants relative to women with submuscular implants. These results suggest that the proximity of the breast implant to breast tissue does not pose additional risk for the woman. This finding supports one of the biological mechanisms previously mentioned regarding the fact that the weight and volume of breast implants may compress the glandular tissue resulting in a decreased blood supply that may reduce the rate of cell proliferation (10). However, we believe that further investigations of breast cancer incidence according to implant characteristics, particularly for implant placement, should be done in order to determine whether this reduction in breast cancer incidence is seen in other cohorts.

Our study also showed that women with subglandular polyurethane covered implants may have an increase in breast cancer rate for the first few years after breast augmentation surgery that decreases with increasing follow-up. This observation suggests a possible tumor promotion effect of the biodegradation products of polyurethane. This finding confirms the previous concerns raised over the carcinogenic effect of the biodegradation products of polyurethane (33;91-93) and consequently regarding the use of polyurethane envelopes for cosmetic breast augmentation (33).Given that polyurethane coated implants are still in use in some parts of the world such as Europe, we believe further assessments

161 are needed to clarify the rate of biodegradation of polyurethane among augmented women and the potential tumor promotion effect of the biodegradation products of polyurethane coated implants among these women.

Our investigation of rarer forms of cancers, including hematopoietic cancers, showed no differences in risks between augmented women and other plastic surgery women. This further evidence provides timely reassuring information, especially given the recent concern raised in the scientific literature over hematopoietic malignancies among women with cosmetic breast implants (41).

The results of this thesis combined with previous epidemiologic studies (42-49) show elevated rates of suicide among women with cosmetic breast implants. These results are relevant and consistent with the scientific literature suggesting underlying psychological conditions that need to be acknowledged among augmented women (84;96;127). As well, we were able to characterize the risk of suicide according to time since implantation and age at surgery. Specifically, our results were suggestive of a trend of increasing attributable risk of suicide with increasing time since implantation. We also found that women who had breast augmentation under age 30 had the highest increase in suicide rate compared to either other cosmetic surgery women or women in the general population. These findings provide support for monitoring the presence of any psychiatric conditions among women interested in cosmetic breast augmentation at a young age and over the long term for those who already have cosmetic breast implants. Overall, our analyses of mortality among augmented women, especially for suicide deaths, provide further evidence that underlying psychological conditions should be monitored among these women. There should be further investigations to determine whether the elevation in risk of suicide increases with time since implantation and age at which surgery was performed.

Reduced mortality rates were observed among augmented women compared to women in the general population for several causes of death which seemed to be correlated with the improved health status of women undergoing cosmetic surgery. However, when comparisons were made with the other cosmetic surgery women, little differences were

162 observed. There should be further studies focusing on the long term lifestyle risk factors of augmented women that could possibly explain our findings.

Our new results in combination with our systematic review and meta-analysis show that, based on current literature, women with cosmetic breast implants that develop breast cancer have an increased risk of being diagnosed with non-localized (advanced) breast tumors than non augmented women with breast cancer. Current evidence also suggests that cosmetic breast implants adversely affect breast cancer-specific survival following the diagnosis of such disease. These findings confirm the previous concerns (50-53) that breast implants may hinder the detection of breast cancer at an early stage which can consequently affect breast cancer survival. Given that the female population with breast implants increases and ages, the number of breast cancers diagnosed in this population will increase and public health concerns associated with the detection and survival of these breast cancers are of great importance. Therefore, our findings are timely and provide support that the detection of breast cancer and the associated survival of such disease need to be acknowledged seriously among these women. We also believe that further investigations are required in order to confirm our findings and clarify breast cancer diagnosis and prognosis according to specific implant characteristics.

In conclusion, cosmetic breast implants have been used for nearly half a century and over that period of time several concerns have been raised regarding their use. Many women around the world have cosmetic breast implants and given their increasing popularity, more women are expected to have breast augmentation for cosmetic purposes. Therefore, women considering such procedure, women who already have implants, physicians and other health professionals who treat these women as well as public health professionals should be informed of the research-based information regarding the use of cosmetic breast implants. Specifically, they should be reassured that cosmetic breast implants are not associated with an increased risk of developing breast cancer or any other rarer forms of cancers. However, they should also be informed of the increased risk of breast cancer associated with subglandular polyurethane covered implants, the elevated rates of suicide among augmented women, particularly for women who had breast augmentation at a young age

163 and over the long term for those who already have cosmetic breast implants, and the fact that cosmetic breast augmentation can adversely affect the diagnosis and prognosis of breast cancer. There should be further investigations in order to fill the remaining research gaps identified in this thesis including the clarification of breast cancer incidence according to implant characteristics, particularly for implant placement, and the diagnosis and prognosis of breast cancer.

164

References

(1) Love S, Lindsey K. Variations in development. Dr. Susan Love's breast book. Cambridge: Perseus Publishing; 2000.

(3) American Society of Plastic Surgeons. Cosmetic surgery trends, 2000, 2008, 2009. Available from: URL: http://www.plasticsurgery.org/Documents/Media/statistics/2009-ASPS-member- surgeon-cosmetic-trends-statistics.pdf. 2010.

(4) Breast of Canada. September Newsletter. Available from: URL: http://www.breastofcanada.com/newslettersep.html. 2009.

(5) Yoshida SH, Swan S, Teuber SS, Gershwin ME. Silicone breast implants: immunotoxic and epidemiologic issues. Life Sci 1995;56:1299-310.

(6) Berkel H, Birdsell DC, Jenkins H. Breast augmentation: a risk factor for breast cancer? N Engl J Med 1992;326:1649-53.

(7) Brinton LA, Malone KE, Coates RJ, Schoenberg JB, Swanson CA, Daling JR, et al. Breast enlargement and reduction: results from a breast cancer case-control study. Plast Reconstr Surg 1996;97(2):269-75.

(8) Brinton LA, Lubin J, Burich MC, Colton T, Brown SL, Hoover RN. Breast cancer following augmentation mammoplasty (United States). Cancer Causes Control 2000;11:819-27.

(9) Brisson J, Holowaty EJ, Villeneuve PJ, Xie L, Ugnat A-M, Latulippe L, et al. Cancer incidence in a cohort of Ontario and Quebec women having bilateral breast augmentation. Int J Cancer 2005;118:2854-62.

(10) Bryant H, Brasher P. Breast implants and breast cancer - reanalysis of a linkage study. N Engl J Med 1995;332(23):1535-9.

(11) Deapen DM, Pike MC, Casagrande JT, Brody GS. The relationship between breast cancer and augmentation mammaplasty: an epidemiologic study. Plast Reconstr Surg 1986;77(3):361-8.

(12) Deapen DM, Brody GS. Augmentation mammaplasty and breast cancer: a 5-year update of the Los Angeles study. Plast Reconstr Surg 1992;89(4):660-5.

165

(13) Deapen DM, Brody GS. Augmentation mammaplasty and breast cancer: a five year update of the Los Angeles study. J Clin Epidemiol 1995;48(4):551-6.

(14) Deapen DM, Bernstein L, Brody GS. Are breast implants anticarcinogenic ? A 14-year follow-up of the Los Angeles study. Plast Reconstr Surg 1997;99(5):1346-53.

(15) Deapen DM. Breast implants and breast cancer: a review of incidence, detection, mortality, and survival. Plast Reconstr Surg 2007;120(Suppl. 1):70S-80S.

(16) Deapen DM, Hirsch EM, Brody GS. Cancer risk among Los Angeles women with cosmetic breast implants. Plast Reconstr Surg 2007;119:1987-92.

(17) Friis S, Mclaughlin JK, Mellemkjaer L. Breast implants and cancer risk in Denmark. Int J Cancer 1997;714:956-8.

(18) Friis S, Holmich LR, Mclaughlin JK, Kjøller K, Fryzek JP, Henriksen TF, et al. Cancer risk among Danish women with cosmetic breast implants. Int J Cancer 2006;118(4):998-1003.

(19) Hoshaw SJ, Klein PJ, Clark BD, Cook RR, Perkins LL. Breast implants and cancer: causation, delayed detection, and survival. Plast Reconstr Surg 2001;107:1393-408.

(20) Jakub JW, Ebert MD, Cantor A, Gardner M, Reintgen DS, Dupont EL, et al. Breast cancer in patients with prior augmentation: Presentation, stage, and lymphatic mapping. Plast Reconstr Surg 2004;114:1737-42.

(21) Kern KA, Flannery JT, Kuehn PG. Carcinogenic potential of silicone breast implants: a Connecticut statewide study. Plast Reconstr Surg 1997;100(3):737-47.

(22) Lipworth L, Tarone RE, Friis S, Ye W, Olsen JH, Nyren O, et al. Cancer among scandinavian women with cosmetic breast implants: A pooled long-term follow- up study. Int J Cancer 2008;124:490-3.

(23) Malone KE, Stanford JL, Daling JR, Voigt LF. Implants and breast cancer. Lancet 1992;339(8805):1365.

(24) Mclaughlin JK, Fraumeni JF Jr., Olsen JH, Mellemkjaer L. Breast implants, cancer, and systemic sclerosis. J Natl Cancer Inst 1994;86(18):1424.

(25) Mclaughlin JK, Nyren O, Blot WJ, Yin L, Josefsson S, Fraumeni JF Jr, et al. Cancer risk among women with cosmetic breast implants: a population based cohort study in Sweden. J Natl Cancer Inst 1998;90(2):156-8.

(26) Mclaughlin JK, Lipworth L, Fryzek JP, Weimin Y, Tarone RE, Nyren O. Long- term cancer risk among Swedish women with cosmetic breast implants: An update of a nationwide study. J Natl Cancer Inst 2006;98(8):557-60.

166

(27) Mclaughlin JK, Lipworth L, Murphy DK, Walker PS. The safety of silicone gel- filled breast implants. A review of the epidemiologic evidence. Ann Plast Surg 2007;59(5):569-80.

(28) Mellemkjaer L, Kjoller K, Friis S, Mclaughlin JK, Høgsted C, Winther JF, et al. Cancer occurence after cosmetic breast implantation in Denmark. Int J Cancer 2000;88:301-6.

(29) Meunier A, Tristant H, Sinna R, Delay E. Implants mammaires et cancer du sein. Annales de chirurgie plastique esthétique 2005;50:595-604.

(30) Pukkala E, Boice Jr.JD, Hovi S-L, Hemminki E, Asko-Seljavaara S, Keskimäki I, et al. Incidence of breast cancer and other cancers among Finnish women with cosmetic breast implants, 1970-1999. J Long Term Eff Med Implants 2002;12(4):251-3.

(31) Schumann D. Health risks for women with breast implants. Nurse Pract 1994;19(7):19-30.

(32) Skinner KA, Silberman H, Dougherty W, Gamagami P, Waisman J, Sposto R, et al. Breast cancer after augmentation mammoplasty. Ann Surg Oncol 2001;8(2):138-44.

(33) Expert Panel on the Safety of Polyurethane-covered Breast Implants. Safety of polyurethane-covered breast implants. CMAJ 1991;145(9):1125-32.

(34) International Agency for Research on Cancer. IARC monographs on the evaluation of the carcinogenic risk of chemicals to humans - some chemicals used in plastics and elastomers. Lyon; 1986.

(35) Cohney BC, Cohney TB, Hearne VA. Augmentation mammaplasty - a further review of 20 years using the polyurethane-covered prosthesis. J Long Term Eff Med Implants 1992;1(3):269-79.

(36) Goodman CM, Cohen V, Thornby J, Netscher D. The life span of silicone gel breast implants and a comparison of mammography, ultrasonography, and magnetic resonance imaging in detecting implant rupture: a meta-analysis. Ann Plast Surg 1998;41(6):577-85.

(37) Brown SL, Middleton MS, Berg WA, Soo MS, Pennello G. Prevalence of rupture of silicone gel breast implants revealed on MR imaging in a population of women in Birmingham, Alabama. AJR Am JRoentgenol 2000;175:1057-64.

(38) International Agency for Research on Cancer.Monographs on the evaluation of carcinogenic risks to humans. Surgical implants and other foreign bodies. Lyon; 1999.

167

(39) Brinton LA. The relationship of silicone breast implants and cancer at other sites. Plast Reconstr Surg 2007;120(Suppl. 1):94S-102S.

(40) Jewell M, Spear SL, Largent J, Oefelein MG, Adams WP. Anaplastic large T-cell lymphoma and breast implants: a review of the literature. Plast Reconstr Surg 2011;128(3):651-61.

(41) U.S.Food and Drug Administration. Anaplastic Large Cell Lymphoma (ALCL) In Women with Breast Implants: Preliminary FDA Findings and Analyses. Center for Devices and Radiological Health; 2011.

(42) Brinton LA, Lubin JH, Murray MC, Colton T. Mortality rates among augmentation mammoplasty patients: an update. Epidemiology 2006;17(2):162-9.

(43) Villeneuve PJ, Holowaty EJ, Brisson J, Xie L, Ugnat A-M, Latulippe L, et al. Mortality among Canadian women with cosmetic breast implants. Am J Epidemiol 2006;164(4):334-41.

(44) Pukkala E, Kulmana I, Hovi S-L, Hemminki E, Keskimäki I, Pakkanen M, et al. Causes of death among Finnish women with cosmetic breast implants, 1971-2001. Ann Plast Surg 2003;51(4):339-42.

(45) Brinton LA, Lubin JH, Burich C. Mortality among augmentation mammoplasty patients. Epidemiology 2001;12(3):321-6.

(46) Lipworth L, Nyren O, Ye W, Fryzek JP, Tarone RE. Excess mortality from suicide and other external causes of death among women with cosmetic breast implants. Ann Plast Surg 2007;59(2):119-23.

(47) Jacobsen PH, Holmich LR, Mclaughlin JK, Johansen C, Olsen JH, Kjøller K, et al. Mortality and suicide among Danish women with cosmetic breast implants. Arch Intern Med 2004;164(22):2450-5.

(48) Koot VCM, Peeters PHM, Granath F, Grobbee DE, Nyren O. Total and cause- specific mortality among Swedish women with cosmetic breast implants: prospective study. BMJ 2003;326(7388):527-8.

(49) Rubin JP, Landfair AS, Shestak K, Lane D, Valoski A, Chang Y, et al. Health characteristics of postmenopausal women with breast implants. Plast Reconstr Surg 2010;125(3):799-810.

(50) Silverstein MJ, Handel N, Gamagami P. The effect of silicone-gel-filled implants on mammography. Cancer 1991;68(S5):1159-63.

(51) Silverstein MJ, Handel N, Gamagami P, Waisman JR, Gierson ED, Rosser RJ, et al. Breast cancer in women after augmentation mammoplasty. Arch Surg 1988;123(6):681-5.

168

(52) Handel N, Silverstein MJ, Gamagami P, Jensen JA, Collins A. Factors Affecting Mammographic Visualization of the Breast After Augmentation Mammaplasty. JAMA 1992;268(14):1913-7.

(53) Fajardo LL, Harvey JA, McAleese KA, Roberts CC, Granstrom P. Breast cancer diagnosis in women with subglandular silicone gel-filled augmentation implants. Radiology 1995;194(3):859-62.

(54) Silverstein MJ, Gierson ED, Gamagami P, Handel N, Waisman JR. Breast cancer diagnosis and prognosis in women augmented with silicone gel-filled implants. Cancer 1990;66(1):97-101.

(55) Silverstein MJ, Handel N, Gamagami P, Gierson ED, Furmanski M, Collins AR, et al. Breast cancer diagnosis and prognosis in women following augmentation with silicone gel-filled prostheses. Eur J Cancer 1992;92(28(2-3)):635-40.

(56) Xie L, Brisson J, Holowaty E, Vileneuve PJ, Mao Y. The influence of cosmetic breast augmentation on the stage distribution and prognosis of women subsequently diagnosed with breast cancer. Int J Cancer 2010;126(9):2182-90.

(57) Clark CP, Peters GN, O'Brien KM. Cancer in the augmented breast: diagnosis and prognosis. Cancer 1993;72(7):2170-4.

(58) Holmich LR, Mellemkjaer L, Gunnarsdottir KA, Tange UB, Krag C, Møller S, et al. Stage of breast cancer at diagnosis among women with cosmetic breast implants. Cancer 2003;88(6):832-8.

(59) Birdsell DC, Jenkins H, Berkel H. Breast cancer diagnosis and survival in women with and without breast implants. Plast Reconstr Surg 1993;92(5):795-800.

(60) Deapen DM, Hamilton A, Bernstein L, Brody GS. Breast cancer stage at diagnosis and survival among patients with prior breast implants. Plast Reconstr Surg 2000;105(2):535-40.

(61) Handel N, Silverstein MJ. Breast cancer diagnosis and prognosis in augmented women. Plast Reconstr Surg 2006;118:587-93.

(62) Miglioretti DL, Rutter CM, Celler BM, Cutter G, Barlow WE, Rosenberg R, et al. Effect of breast augmentation on the accuracy of mammography and cancer characteristics. JAMA 2004;291(4):442-50.

(63) Cahan AC, Ashikari R, Pressman P, Cody H, Hoffman S, Sherman JE. Breast cancer after breast augmentation with silicone implants. Ann Surg Oncol 1995;2(2):121-5.

(64) Handel N. The effect of silicone implants on the diagnosis, prognosis, and treatment of breast cancer. Plast Reconstr Surg 2007;120(Suppl. 1):81S-93S.

169

(65) Tuli R, Flynn RA, Brill KL, Sabol JL, Usuki KY, Rosenberg AL. Diagnosis, treatment, and management of breast cancer in previously augmented women. Breast J 2006;12(4):343-8.

(66) Baines CJ, Arseneau J, Davis P, Smith DC. Report on Silicone Gel-Filled Implants, prepared for the Department of National Health and Welfare, Canada. Canada; 1992.

(67) Cronin TD. Silicone breast implants. In: Leonard P.Rubin, editor. Biomaterials in Reconstructive Surgery.St-Louis: 1982. p. 552-63.

(68) Gladfelter J. Breast augmentation 101: Understanding cosmetic breast mammaplasty. Plastic Surgical Nursing 2007;27(3):136-45.

(69) American Society of Plastic Surgeons. Top 5 cosmetic surgery procedures. Available from: URL: http://www.plasticsurgery.org/Documents/Media/statistics/2009-top-5-cosmetic- surgery-procedures-graph.pdf. 2010.

(70) American Society for Aesthetic Plastic Surgery. 2007 ASAPS News Release. Cosmetic Plastic Surgery Research. Available from: URL: http://www.cosmeticplasticsurgerystatistics.com/statistics.html#2007- HIGHLIGHTS. 2009.

(71) Crerand CE, Infield AL, Sarwer DB. Psychological considerations in cosmetic breast augmentation. Plastic Surgical Nursing 2009;29(1):49-57.

(72) Lilla JA, Vistnes LM. Long-term study of reactions to various silicone breast implants in rabbits. Plast Reconstr Surg 1976;57(5):637-49.

(73) Independent Advisory Committee on Silicone Gel-filled Breast Implants. Summary of the report on silicone gel-filled breast implants. CMAJ 1992;147(8):1141-6.

(74) Canadaba.ca. A guide to breast augmentation and breast implants in Canada: Subglandular placement. Available from: URL: http://www.canadaba.ca/implantplace/overs.html. 2009.

(75) Canadaba.ca. A guide to breast augmentation and breast implants in Canada: Submuscular placement. Available from: URL: http://www.canadaba.ca/implantplace/unders.html. 2009.

(76) Handel N, Silverstein MJ, Jensen JA, Collins A, Zierk K. Comparative experience with smooth and polyurethane breast implants using the Kaplan-Meier method of survival analysis. Plast Reconstr Surg 1991;88(3):475-81.

(77) Herman S. The Même implant. Plast Reconstr Surg 1984;73(3):411-44.

170

(78) Hester TR Jr., Ford NF, Gale PJ, Hammett JL, Raymond R, Turnbull D, et al. Measurement of 2,4-toluenediamine in urine and serum samples from women with Même or Replicon breast implants. Plast Reconstr Surg 1997;100(5):1291-8.

(79) Melmed ED. Polyurethane implants: a 6-year review of 416 patients. Plast Reconstr Surg 1988;82(2):285-90.

(80) Dow Corning Corporation. Medtox project - Analysis of internal safety studies relevant to health care materials and products. Midland, Michigan; 1987.

(81) Health Canada. It's your health. " Breast implants ". Health Canada 2009 Available from: URL: http://www.hc-sc.gc.ca/hl-vs/iyh-vsv/med/implants- eng.php. 2009.

(82) Canadaba.ca. A guide to breast augmentation and breast implants in Canada: Silicone filled breast implants. Available from: URL: http://www.canadaba.ca/implantinfo/siliconeimplantscanada.html. 2009.

(83) Health Canada. Drugs and health products: Notice of decision for silicone gel- filled breast implants. Health Canada 2009. Available from: URL: http://www.hc- sc.gc.ca/dhp-mps/prodpharma/sbd-smd/phase1-decision/md im/nd.ad.2006.lowbleed.69312-end.php. 2009.

(84) Sarwer DB. The psychological aspects of cosmetic breast augmentation. Plast Reconstr Surg 2007;120(Suppl. 1):110S-7S.

(85) Figueroa-Haas C. Effect of breast augmentation mammoplasty on self-esteem and sexuality: a quantitative study. Plastic Surgical Nursing 2007;27(1):16-36.

(86) Bjornsson AS, Didie ER, Phillips KA. Body dysmorphic disorder. Dialogues Clin Neurosci 2010;12(2):221-32.

(87) Kothari AR, Birch S. Individual and regional determinants of mammography uptake. Can J Public Health 2009;95(4):290-4.

(88) Brinton LA, Brown SL, Colton T, Burich MC, Lubin J. Characteristics of a population of women with breast implants compared with women seeking other types of plastic surgery. Plast Reconstr Surg 2000;105(3):919-29.

(89) Schwartz AW, Erich JB. Experimental study of polyvinyl-Formal (ivalon) sponge as a substitute for tissue. Plast Reconstr Surg 1960;25:1-14.

(90) Brown JB, Fryer MP, Kollias P, Ohlwiler DA, Templeton JB. Silicone and Teflon prostheses, including full jaw substitution: laboratory and clinical studies of Etheron. Ann Surg 1963;157:932-43.

(91) Turner FC. Sarcomas at sites of subcutaneously implanted Bakelite discs in rats. J Natl Cancer Inst 1941;2(81).

171

(92) Oppenheimer BS, Oppenheimer ET, Stout AP. Sarcomas induced in rodents by embedding various plastic films. Proc Soc Exp Biol Med 1952;79:366-9.

(93) Oppenheimer ET, Willhite M, Stout AP, Danishefsky I, Fishman MM. A comparative study of the effects of imbedding cellophane and polystyrene films in rats. Cancer Res 1964;24:379-87.

(94) Berry MG, Davies DM. Breast augmentation: part 1 - A review of the silicone prosthesis. J Plast Reconstr Aesthet Surg 2009.

(95) Didie ER, Phillips KA. Mortality among Canadian women with cosmetic breast implants - a letter to the editor. Am J Epidemiol 2007;165(7):846.

(96) Sarwer DB, Brown GK, Evans DL. Cosmetic breast augmentation and suicide. Am J Psychiatry 2007;164(7):1006-13.

(97) Smalley SM. Breast implants and breast cancer screening. J Midwifery Womens Health 2003;48(5):329-35.

(98) Hayes HJr, Vandergrift J, Diner WC. Mammography and breast implants. Plast Reconstr Surg 1988;82(1):1-8.

(99) Eklund GW, Busby RC, Miller SH, Job JS. Improved imaging of the augmented breast. AJR Am J Roentgenol 1988;151(3):469-73.

(100) Eklund GW, Cardenosa G. The art of mammographic positionning. Radiol Clin North Am 1992;30(1):21-53.

(101) Toronto Cosmetic Clinic. Breast Augmentation and Mammograms. Toronto Cosmetic Clinic 2009. Available from: URL: http://www.tcclinic.com/augmentationmammoplasty/breast- augmentationmammogram.php. 2009.

(102) Sarwer DB, Brown GK, Evans DL. Cosmetic breast augmentation and suicide. Am J Psychiatry 2007;164(7):1006-13.

(103) Silverstein MJ, Handel N, Gamagami P, Waisman E, Gierson ED. Mammographic measurements before and after augmentation mammaplasty. Plast Reconstr Surg 1990;86(6):1126-30.

(104) Band PR, Gaudette LA, Hill GB, Holowaty EJ, Huchcroft SA, Johnston GM, et al. The making of the Canadian cancer registry: cancer incidence in Canada and its regions, 1969 to 1988. Ottawa, ON; 1993.

(105) Howe GR, Lindsay J. A generalized iterative record linkage computer system for use in medical follow-up studies. Comput Biomed Res 1981;14(4):327-40.

172

(106) Brenner DR, Tammemägi MC, Bull SB, Pinnaduwaje D, Andrulis IL. Using cancer registry data: agreement in cause-of-death data between the Ontario Cancer Registry and a longitudinal study of breast cancer patients. Chronic Dis Can 2009;30(1):16-9.

(107) Ministère de la Santé et des Services Sociaux du Québec. Fichier des tumeurs du Québec. Ministère de la Santé et des Services Sociaux du Québec 2010. Available from: URL: http://msssa4.msss.gouv.qc.ca/santpub/tumeurs.nsf/cat?OpenView. 2010.

(108) World Health Organization. International Classification for Diseases. 9th Revision. Geneva: 1977.

(109) Goldberg MS, Carpenter M, Thériault G, Fair M. The accuracy of ascertaining vital status in a historical cohort study of synthetic textiles workers using computerized record linkage to the Canadian Mortality Data Base. Can J Public Health 1993;84(3):201-4.

(110) Schnatter AR, Acquavella JF, Thompson FS, Donaleski D, Thériault G. An analysis of death ascertainment and follow-up through statistics Canada's Mortality Data Base system. Can J Public Health 1990;81(1):60-5.

(111) World Health Organization. International Classification for Diseases, 10th revision. Geneva: 2000.

(112) American Joint Commission on Cancer. AJCC Cancer Staging Manual. 5th edition. Philadelphia: 1997.

(113) American Joint Commission on Cancer. AJCC Cancer Staging Manual. 6th edition. Philadelphia: 2002.

(114) EPICURE: risk regression and data analysis software [computer program]. Seattle: Hirosoft International Corporation; 1996.

(115) Breslow NE, Day NE. Statistical methods in cancer research. Vol. II, The design and analysis of cohort studies. Lyon, France: International Agency for Research on Cancer; 1987.

(116) Kleinbaum DG, Kupper LL, Muller KE, Nizam A. Applied regression analysis and other multivariable methods. 3rd ed. Brooks/Cole Publishing Company; 1998.

(117) Kleinbaum DG, Klein M. Survival analysis: a self leaning text. 2 ed. Atlanta, GA: Springer; 2005.

(118) Rothman KJ, Greenland S, Lash TL. Modern epidemiology. 3rd ed. Philadelphia: Lippincott, Williams & Wilkins; 2008.

173

(119) SAS, ed 9.2 [computer program]. Cary, NC, USA: SAS Institute; 2008.

(120) Stata statistical software: release 11 [computer program]. College Station, TX: StataCorp LP; 2009.

(121) PASS 2008 [computer program]. Kaysville, Utah: NCSS LLC; 2008.

(122) Walter SD. Determination of significant relative risks and optimal sampling procedures in prospective and retrospective comparative studies of various sizes. Am J Epidemiol 1977;105(4):387-97.

(123) Signorini D. Sample size for Poisson regression. Biometrika 1991;78(2):446-50.

(124) Hsieh FY, Block DA, Larsen MD. A simple method of sample size calculation for linear and logistic regression. Stat Med 1998;17:1623-34.

(125) Hsieh FY, Lavori PW. Sample-size calculations for the Cox Proportional Hazards regression model with nonbinary covariates. Control Clin Trials 2000;21(552):560.

(126) Schoenfeld DA. Sample-size formula for the proportional hazards regression model. Biometrics 1983;39:499-503.

(127) Lipworth L, Mclaughlin JK. Excess suicide risk and other external causes of death among women with cosmetic breast implants: a neglected research priority. Curr Psychiatry Rep 2010;12:234-8.