Association of Anorexia Nervosa With Risk of Cancer

IMPORTANCE Anorexia nervosa is recognized as an important cause of morbidity in young people. However, the risk of cancer in people with anorexia nervosa remains uncertain. OBJECTIVE To evaluate the association of anorexia nervosa with the risk of developing or dying of cancer. DATA SOURCES MEDLINE, Scopus, Embase, and Web of Science from database inception to January 9, 2019. STUDY SELECTION Published observational studies in humans examining the risk of cancer in people with anorexia nervosa compared with the general population or those without anorexia nervosa. Studies needed to report incidence or mortality rate ratios (RRs). DATA EXTRACTION AND SYNTHESIS Screening, data extraction, and methodological quality assessment were performed by at least 2 researchers independently. A random-effects model was used to synthesize individual studies. Heterogeneity (I2) was assessed and 95% prediction intervals (PIs) were calculated. MAIN OUTCOMES AND MEASURES All cancer incidence and cancer mortality associated with anorexia nervosa. Secondary outcomes were site-specific cancer incidence and mortality. RESULTS Seven cohort studies published in 10 articles (42 602 participants with anorexia nervosa) were included. Anorexia nervosa was not associated with risk of developing any cancer (4 studies in women; RR, 0.97; 95% CI, 0.89-1.06; P = .53; I2, 0%; 95% PI, 0.80-1.18; moderate confidence). Anorexia nervosa was associated with decreased breast cancer incidence (5 studies in women; RR, 0.60; 95% CI, 0.50-0.80; P < .001; I2, 0%; 95% PI, 0.44-0.83; high confidence). Conversely, anorexia nervosa was associated with increased risk of developing lung cancer (3 studies in women; RR, 1.50; 95% CI, 1.06-2.12; P = .001; I2, 0%; 95% PI, 0.19-16.46; low confidence) and esophageal cancer (2 studies in women; RR, 6.10; 95% CI, 2.30-16.18; P < .001; I2, 0%; low confidence). CONCLUSIONS AND RELEVANCE Among people with anorexia nervosa, risk of developing cancer did not differ compared with the general population, but a significantly reduced risk of breast cancer was observed. Understanding the mechanisms underlying these associations could have important preventive potential. JAMA Network Open. 2019;2(6):e195313. doi:10.1001/jamanetworkopen.2019.5313 Key Points Question Are people with anorexia nervosa at a higher risk of developing or dying of cancer compared with those without anorexia and the general population? Findings In this systematic review and meta-analysis of 7 cohort studies including more than 42 000 participants with anorexia nervosa, there was no association of anorexia nervosa with overall cancer incidence or mortality. Anorexia nervosa was inversely associated with breast cancer incidence but positively associated with risk of developing lung and esophageal cancer. Meaning There was no association of anorexia nervosa with risk of cancer overall and few associations of anorexia nervosa with risk of site-specific cancer. + Supplemental content Author affiliations and article information are listed at the end of this article. Open Access. This is an open access article distributed under the terms of the CC-BY License. JAMA Network Open. 2019;2(6):e195313. doi:10.1001/jamanetworkopen.2019.5313 (Reprinted) June 7, 2019 1/14 Downloaded From: https://jamanetwork.com/ by a Instituto de Salud Carlos III User on 06/14/2019 Introduction Almost 3.4 million young people throughout the world have anorexia nervosa.1 Anorexia nervosa is characterized by a severe restriction of caloric intake, extremely low body weight, fear of gaining weight or of becoming fat, and disturbance of body image. Anorexia nervosa is more commonly reported by young women and girls, but the disorder is increasingly being reported by boys, men, and women older than 40 years. Genetic factors and hormonal changes may influence risk, psychosocial and interpersonal factors can trigger onset, and changes in molecular pathways and cellular networks in the brain are associated with the illness and its comorbidities.2-4 Cancer is the second leading cause of death worldwide,5,6 with more than 9.6 million deaths in 2017.6 Multiple studies and meta-analyses suggest that excess body weight is a risk factor of several cancers.7-9 For example, a recent umbrella review8 found evidence of an association of body mass index (BMI) with cancers of digestive organs, hormone-related cancers, uterine cancer, kidney cancer, and multiple myeloma and also of adiposity with the risk of colorectal cancer, gallbladder cancer, gastric cancer, ovarian cancer, and multiple myeloma. The underlying mechanisms of the association of excess body weight with cancer are complex and are not yet fully understood. Research on how reducing body weight might lower the risk of developing cancer is limited, to our knowledge. Energy restriction (or calorie restriction) has been found to be protective against the development of cancer in experimental animal studies.10 In addition, certain calorie-limited diets or fasting could influence the risk of developing cancer. Anorexia nervosa, an excessive form of calorie restriction associated with pathological weight loss, has been proposed as a biomarker of energy restriction in humans.11-13 Several studies have evaluated whether there is a reduction in cancer development among people with anorexia nervosa.12-14 However, to our knowledge, there are no systematic reviews and meta-analyses investigating the strength of the evidence of the potential associations of anorexia nervosa with cancer. In this study, we aimed to evaluate the association of anorexia nervosa with the risk of developing or dying of cancer. Methods We followed the current methods recommendations for systematic reviews15,16 and developed a protocol (PROSPERO registration number: CRD42017067462)17 to comply with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline statement for reporting this study. We also reported this study in accordance with the Meta-analysis of Observational Studies in Epidemiology (MOOSE) reporting guideline. Our methods are briefly described here and explained in more detail in the published protocol17 and in eTables 1-7 in the Supplement. Search Strategy and Selection Criteria One of us (A.A.-A.) searched MEDLINE, Scopus, Embase, and the Web of Science to identify all relevant observational studies in humans that examined the association of anorexia nervosa with the risk of cancer published from database inception to January 9, 2019, without language restrictions. The main search strategy for MEDLINE is presented in eTable 3 in the Supplement. This search strategy was adapted to fit with other databases. To supplement these searches, references of all relevant primary studies and review articles were also screened to identify additional data sources. To be included, primary studies had to be observational (ie, cohort or case-control) studies of people with anorexia nervosa (according to standard operationalized diagnostic criteria, ie, International Classification of Diseases, Ninth Revision [ICD-9] codes 307.1 or 307.54 or ICD-10 codes F50.0-F50.1) and report the incidence or mortality rate ratios (RRs) for the risk of cancer in patients with anorexia nervosa compared with the general population or those without anorexia nervosa or have enough data (ie, number of cases and sample size, observed and expected cases) to compute these estimates. We excluded studies in which anorexia nervosa was not the exposure of interest and in which cancer was not reported as the outcome. Studies not presenting study-specific data or JAMA Network Open | Oncology Association of Anorexia Nervosa With Risk of Cancer JAMA Network Open. 2019;2(6):e195313. doi:10.1001/jamanetworkopen.2019.5313 (Reprinted) June 7, 2019 2/14 Downloaded From: https://jamanetwork.com/ by a Instituto de Salud Carlos III User on 06/14/2019 sufficient data for an outcome measure to be calculated were also excluded. We excluded case reports, case series, in vitro studies, and animal studies. Two of us (F.C.-L. and J.F.-M.) independently screened the titles and abstracts of articles retrieved from the literature search, and the full-texts of potentially eligible articles were obtained and further assessed for final inclusion. Disagreements were resolved through discussions until a consensus was reached. Outcomes The primary outcomes were all cancer incidences and cancer mortalities (all malignant neoplasms; ICD-9 codes 140-209; ICD-10 codes C00-C97). Given the varied biology of cancers, the risk of incident site-specific cancers and the risk of fatal site-specific cancers were evaluated as secondary outcomes (eTable 1 in the Supplement). Data Extraction and Quality Assessment Two of us (F.C.-L. and J.F.-M.) independently extracted article data, including first author, publication year, period of recruitment, country, study design, setting, coverage, mean or median age (or age range), proportion of female participants, race/ethnicity, parity status, profile of tobacco smoking, ascertainment of cancer diagnosis and diagnostic criteria, consideration of confounding factors, the number of participants with cancer, and maximally adjusted RR estimates with 95% CIs from the included studies. For duplicate study publications, we considered only the report with the most informative and complete data. When relevant outcome data were not available, we directly contacted the corresponding author of the study to request the information. Two pairs of us (J.F.-M., M.J.P., M.R., and/or D.M.S.-G.) independently undertook methodological quality assessment of included studies using the Newcastle-Ottawa scale18 and allocated stars for adherence to the prespecified criteria. This scale ranges from 0 stars (lowest quality) to 9 stars (highest quality) and judges each study regarding selection of study groups, comparability, and ascertainment of the outcome. We considered studies with 0 to 3 s