The association between Alzheimer’s disease and cancer: Systematic review – Meta-analysis 1
1
2
Tilemachos–Nektarios Papageorgakopoulos , Despina Moraitou PhD, Maria Papanikolaou PhD, 3
Magda Tsolaki MD, PhD
1. School of Psychology, Aristotle University of Thessaloniki, Greece. 2. Department of Physiology and Biophysics, University of California, Irvine, USA. 3. School of Medicine Aristotle University of Thessaloniki, Greece, Greek Association of Alzheimer’s Disease and Related Disorders
Keywords: Dementia -Inverse comorbidity -Cancer localization.
Corresponding author: Despina Moraitou PhD, Lab of Psychology, Section of Cognitive and Experimental
Psychology, School of Psychology, Aristotle University of Thessaloniki, Thessaloniki 541 24, Greece. Tel: +302310997068. Email:
[email protected]
Abstract The objective of the present study was the quantitative assessment of the previously documented inverse relationship between Alzheimer’s Disease (AD) and cancer (CA) by conducting a meta-analysis and evaluating systematic differentiations of the aforementioned relationship based on cancer localization. For the purpose of the study all available empirical data of the last decade, which met specific criteria, were included in the analysis by querying PubMed, Web of Science and Cochrane Library databases. Seven studies were included in the analysis, with a total sample of 18,887 (10,859 AD patients, 8,028 non-demented controls) participants to calculate cancer risk among AD patients, and 11 studies, with a total of 5,607,076 (1,853,318 cancer patients, 3,753,758 healthy controls) participants, were assessed to evaluate AD risk among cancer patients. The analysis revealed that AD patients appear to have a reduced risk of cancer, by 40% (RR 0.60, 95% CI 0.45 - 0.79), while cancer history was associated with a reduced risk of AD, by 15% (RR 0.85, 95% CI 0.77-0.92). Systematic differences were also identified based on site-specific cancer. Indications of heterogeneity and publication bias were present in the analysis. Our meta-analysis is only the fourth conducted on this subject, with newer evidence suggesting a mitigation of the inverse relationship. We emphasize the need for new studies to assess the inverse comorbidity hypothesis, especially in AD patients.
The association between Alzheimer’s disease and cancer Alzheimer’s disease (AD) is a neurodegenerative disease that affects primarily persons over 65 years of age and it is the most common type of dementia [1]. AD patients experience impairment of their cognitive functions which gradually leads to autonomy loss. Despite the apolipoprotein E (ApoE) gene having been identified as the most important genetic risk factor for AD [2], the underlying cellular and molecular mechanisms responsible remain largely unknown [3, 4]. The most prevailing hypotheses have been focused on metabolic factors [5, 6], immunological factors [7, 8] as well as β-amyloid and tau pathology [9]. Demetrious et al. (2014), have suggested that in some cases (sporadic AD), an inverse metabolic reprogramming of mitochondria is observed, that is the reverse of mitochondrial reprogramming observed in cancer [10].
The Inverse Warburg effect hypothesis introduced a novel approach to studying the pathogenesis of AD into modern bibliography. According to this position, attention is centered on the comparison of mechanisms causing AD, to those that lead to the development of neoplasia in cancer [11-14]. TabaresSeisdedos and Rubenstein (2013) introduced the hypothesis of an inverse comorbidity between some types of cancer and Alzheimer’s disease. Inverse comorbidity, constitutes a situation where a lower-thanexpected probability of disease is occurring in individuals who have been diagnosed with another medical condition [15]. Sánchez-Valle et al. (2017) observed an inverse relationship between AD and lung cancer [16] which is attributable to an oppositely mitochondrial metabolism regulation. Another study by White et al. (2013) supports a remarkable inverse relationship (HR 0.21, 95% CI: 0.051 – 0.87, P=0.031) between Alzheimer’s disease and non-melanoma skin cancer (White, Lipton, Hall, & Steinerman, 2013), a very common cancer type in western cultures [17, 18]. In contrast, the co-occurrence of AD with some other diseases indicates a positive relationship. For example, AD has a high comorbidity with diabetes (RR 1.54, 95% CI: 1.33–1.79), obesity (RR 1.59, 95% CI: 1.02–2.50), hypertension (RR 1.31, 95% CI: 1.01–1.70) and hypercholesterolemia (RR 1.72, 95% CI: 1.32– 2.24) [19, 20]. In each case, it is known that both cancer and AD are associated with age, as both have a higher instance in ages over 60 years and are also both associated with cellular dysfunction, whether it’s the aberrant proliferation of cells observed in cancer, or the cellular degeneration and cell death observed in dementia [21]. Lately, there has been an increased research interest on the inverse relationship between AD and CA; the further understanding of this relationship could uncover valuable knowledge and benefit both the identification of AD pathophysiological mechanisms, as well as new therapeutic approaches. Even though studies so far suggest the presence of an inverse relationship [15], these findings should be interpreted with caution, as objections have been raised concerning methodological and attribution issues [22]. The appropriate statistical analysis and the synthesis of latest empirical data with the use of meta-analysis models could facilitate a quantitative evaluation of this phenomenon, as well as its critical assessment [23]. This is exactly the aim of the present study.
Methods Search Strategies The search for publications of relevant studies was performed in the electronic databases of PubMed, Web of Science and Cochrane Library, using (in the publication title) the combination of the following search terms; “Alzheimer*”, “Dementia”, “Cancer*”, “Neoplasm*”, “Carcinoma” and “Tumor”, and limiting the publication span of the studies within the last decade (2007-2017), in an effort to only use the most recent bibliography. Moreover, the references used in the chosen publications as well as publications using the target articles in their bibliography were examined with the help of Google Scholar. The search returned among others, three meta-analyses, two of which were published in 2015 and one in 2014 [24-26]. Furthermore, one publication with relevant data was discovered [11].
www.nuclmed.gr
2
Hell J Nucl Med Suppl, Sept-Dec 2017
Selection Criteria The studies that were eventually selected had to conform to the following restrictions; (a) the study design had to be either an epidemiology and/or a case study; (b) the results had to be presented in the format of Relative Risk (Risk Ratio, RR), Odds Ratio (OR), Hazard Ratio (HR) or Standardized Incident Ratio (SIR) with a 95% Confidence Interval (CI); and (c) the diagnosis had to be stated clearly. Publications were excluded from the study for the following reasons; (a) the type of dementia was not clearly identified as AD; (b) absence of compatible statistical data and (c) studies that did not focus on human subjects. Finally, the publications selected were written primarily in English (Figure 1).
Figure 1. Diagram showing processes for study selection.
Data extraction The following data were extracted from the selected publications; First Author; Publication Year; Study Design; Sample Size; Study Span and Study Duration; Follow-up Duration, Diagnostic Criteria for Alzheimer’s Disease and Cancer and results in RR, OR, HR or SID with a 95% Confidence Interval (CI). Statistical Analysis The pooled relative risk was calculated with the use of Open Meta Analyst 12.11.14 [27-29] and R software [28, 30]. The Cochrane Review Manager 5.3.5 Calculator was used for statistical data transformations [31, 32] .
The Random-Effects Model; DerSimonian-Laird was the selected method for statistical analysis [33] with a 95% confidence interval, as the selected studies had variations in participants’ profile, sample 2
sizes, design and types of cancer studied [34]. Moreover, the heterogeneity factor (I ) was used to quantify the heterogeneity between various studies. Heterogeneity was considered statistically significant for P
