METHODS: We conducted a systematic review of articles reporting the outcome of dementia among individuals with prostate cancer in those exposed to ADT versus a lesser-exposed comparison group (for example, ADT versus no-ADT; continuous versus intermittent ADT) using PubMed (1966–present), Web of Science (1945–present), Embase (1966–present) and PsycINFO (1806–present).
The search was undertaken on 4 December 2016 by two authors. We meta-analyzed studies reporting an effect estimate and controlling for confounding. Random- or fixed-effects meta-analytic models were used in the presence or absence of heterogeneity per the I2 statistic, respectively. Small study effects were evaluated using Egger and Begg’s tests.
RESULTS: Nine studies were included in the systematic review. Seven studies reported an adjusted effect estimate for dementia risk. A random-effects meta-analysis of studies reporting any dementia outcome, which included 50 541 individuals, showed an increased risk of dementia among ADT users (hazard ratio (HR), 1.47; 95% confidence interval (CI), 1.08–2.00; P = 0.02). We separately meta-analyzed studies reporting all-cause dementia (HR, 1.46; 95% CI, 1.05–2.02; Po0.001) and Alzheimer’s disease (HR, 1.25; 95% CI, 0.99–1.57; P = 0.06). There was no evidence of bias from small study effects (Egger, P = 0.19; Begg, P = 1.00).
CONCLUSION: The currently available combined evidence suggests that ADT in the treatment of prostate cancer may be associated with an increased dementia risk. The potential for neurocognitive deficits secondary to ADT should be discussed with patients and evaluated prospectively.
Androgen deprivation therapy (ADT) is a mainstay of treatment for metastatic and locoregional prostate cancer.1 Globally, there are over a million new diagnoses of prostate cancer each year2 and an estimated 50% of men with prostate cancer will ultimately utilize ADT.3 Therefore, the potential individual and health system implications of adverse effects of ADT are significant.
Recent analyses have suggested a link between the use of ADT in the treatment of prostate cancer and an increased risk of dementia.11,12 However, other studies have failed to support this association,13,14 including the largest study examining this association to date.15 Currently, a comprehensive analysis of this rapidly evolving and controversial topic has not been undertaken and the clinical implications of these findings taken collectively are unclear. In this study, our aim is to undertake a systematic review and meta-analysis of the available data on ADT in the treatment of prostate cancer and dementia risk.
MATERIALS AND METHODS
Eligible studies included all published articles that reported the outcome of a dementia diagnosis among individuals with prostate cancer, examining individuals exposed to ADT versus a lesserexposed comparison group (for example, ADT versus no-ADT; continuous versus intermittent ADT). Inclusion criteria for the quantitative meta-analysis were studies that reported an effect estimate (for example, hazard ratio) and controlled for confounding, or reported measurement of or adjustment for dementia that could be used to calculate an adjusted effect estimate.
We carried out electronic searches in PubMed (1966–present), Web of Science (1945–present), Embase (1966–present) and PsycINFO (1806–present). The search was undertaken on 4 December 2016. The search strategy used for each database query can be found in Table 1. We additionally queried the reference lists of included articles. Two investigators (KTN and SS) independently assessed the eligibility of each study by using the title and abstract for initial screening followed by review of the full text and the data extraction with consensus reached by discussion. We used a data extraction sheet developed on the basis of the Cochrane Consumers and the Communication Review Group’s data extraction template (http://cccrg.cochrane.org/author-resources). Searchers extracted the following items from each study: first-author, type of article (for example, peer reviewed or abstract), location, year of publication, dates of the data collection or enrollment, cohort type (for example, prospective or retrospective), sample size, number of individuals on ADT, outcome (for example, all-cause dementia, Alzheimer’s disease and so on), how the outcome was delineated (for example, ICD-9 codes), type of effect statistic (for example, hazard ratio), effect statistic error measures (for example, confidence interval) and effect statistic P-value. Authors were contacted for additional details as needed. We assessed the internal validity of each study included in the qualitative review based on modified Newcastle-Ottawa Scale criteria.17
We conducted a meta-analysis to calculate summary statistic hazard ratios (HRs) and 95% confidence intervals (CIs) for the risk of dementia in prostate cancer patients exposed to a course of ADT versus a lesser-exposed comparison group (for example, no ADT, shorter ADT duration, intermittent ADT). We included comparison groups exposed to a lesser amount of ADT given evidence for a dose effect of ADT on dementia risk11,18 and that this would be expected to bias the result towards the null. Only effect estimates that accounted for confounding factors (for example, multivariable adjusted or propensity score matched) were utilized in the quantitative meta-analysis. In our primary meta-analysis, we included all eligible studies reporting a dementia outcome. Where eligible studies were clearly overlapping, we included the study with the largest number of events. We additionally conducted meta-analyses separately among those studies examining the outcomes of all-cause dementia and Alzheimer’s disease. The proportion of heterogeneity due to study variation was quantified using the I2 statistic.19 Random- or fixedeffects meta-analytic models were used in the presence and absence of statistically significant heterogeneity, respectively. Heterogeneity was explored as per the Cochrane Handbook for Systematic Reviews of Interventions.20 The presence of small study effects was evaluated by visualization of a funnel plot and calculating Begg and Egger statistics. Tests were considered significant if the two-sided P-value was less than 0.05. All analyses were carried out using Stata version 12 (StataCorp, College Station, TX, USA).
Our study selection process is summarized in Figure 1. In total, we reviewed 176 studies by title and abstract with 26 studies undergoing full text review. We excluded 17 studies after full text review because they did not examine individuals with prostate cancer exposed to ADT versus a lesser-exposed comparison group (n = 10), did not examine the diagnosis of dementia as an outcome (n = 6), or were a subgroup analysis of an included study (n = 1). Nine studies meeting our inclusion criteria are summarized in Table 2, which included five studies examining all-cause dementia, two studies examining Alzheimer’s disease and one study reporting on senile dementia. Eight studies compared patients treated with or without any ADT. One study23 compared intermittent versus continuous ADT. In this study, all patients underwent 7 months of induction gonadotropin-releasing hormone agonist and an anti-androgen followed by randomization to continued continuous ADT or intermittent treatment per PSA values.
Figure 1. Flow diagram of literature search and study selection for the systematic review and meta-analysis. ADT, androgen deprivation therapy.
We assessed study quality using modified Newcastle-Ottawa Scale criteria as shown in Table 3. Two studies21,22 did not implement methodology to account for confounding, or report an effect estimate for dementia risk, and were therefore excluded from the quantitative meta-analysis. Wiechno et al.,22 found similar Mini-Mental Status Examination scores between ADT (27.59) and non-ADT (27.41) users (P = 0.66). Shahinian et al.21 reported a statistically significantly increased risk of senile dementia among ADT users (5.99 versus 4.31%, Po0.001). However, rates of prevalent senile dementia were higher at baseline and this was not accounted for in the analysis.
In our primary meta-analysis (Figure 2), we included six of the remaining eligible studies,11,13–15,23,24 as a study by Nead et al.11 overlapped with a cohort reporting a greater number of dementia events.18 A random-effects meta-analysis of these studies, which included 50 541 individuals, yielded a summary hazard ratio of 1.47 (95% CI, 1.08–2.00; P = 0.02) for the risk of dementia following ADT in the treatment of prostate cancer. A meta-analysis of studies examining all-cause dementia (n = 5) resulted in a summary hazard ratio of 1.46 (95% CI, 1.05–2.02; Po0.001). A meta-analysis of studies examining Alzheimer’s disease (n = 3) showed a summary hazard ratio of 1.25 (95% CI, 0.99–1.57; P = 0.06). A funnel plot to evaluate small study effects can be found in Supplementary Figure 1. There was no evidence of small study effects as evaluated by the Egger (P = 0.12) or Begg (P=1.00) tests.
I2 statistics to evaluate the proportion of heterogeneity due to study variation were 76% (95% CI, 47–89; Po0.001) for the primary analysis, 81% (95% CI, 55–92; Po0.001) for all-cause dementia, and 40% (95% CI, 0–82; P = 0.19) for Alzheimer’s disease. We further explored sources of heterogeneity in our primary analysis by excluding outlying studies, as described.20 We did not observe statistically significant heterogeneity when individually excluding outlying studies by Nead et al.18 (I2, 49%; 95% CI, 0–81; P = 0.10) or Khosrow-Khavar et al.15 (I2, 33%; 95% CI, 0–74; P = 0.20). Additionally, our results remained consistent with exclusion of the outlying studies by Nead et al.18 (HR, 1.15; 95% CI, 1.01–1.30; P = 0.04) or Khosrow-Khavar et al.15 (HR, 1.65; 95% CI, 1.37–1.98; Po0.001). Heterogeneity was not resolved with the individual removal of any other study.
In sensitivity analyses, we found consistent results when excluding Hershman et al.23 as the control group was exposed to intermittent ADT (HR, 1.42; 95% CI, 1.02–1.98; P = 0.04), Chung et al.13 given the possibility of remote overlap with Kao et al.14 (HR, 1.46; 95% CI, 1.05–2.02; P = 0.02), and Capitanio et al.24 as this study is published in abstract form only (HR, 1.48; 95% CI, 1.00–2.20; P = 0.05).
A causal association between ADT in the treatment of prostate cancer and an increased risk of dementia is one potential explanation for our findings and is supported by a number of factors. First, this association is biologically plausible. Neuron death, including secondary to the pathognomonic neurofibrillary tangle formation from abnormal tau production in Alzheimer’s disease, is the overarching cause of dementia in general. Androgens have been demonstrated to play a key role in neuron growth and axonal regeneration.4 Second, this association is coherent with the laboratory data showing that low testosterone levels predict future dementia5,6 and, in addition to ADT, result in elevated β-amyloid protein levels.8 Third, our findings are consistent with the data, supporting an increased risk of dementia among prostate cancer patients undergoing orchiectomy (HR, 1.67; Po0.001).25 Finally, in studies that evaluated dose response, there is evidence of a dose-response effect with those undergoing ADT for the longest duration having the greatest risk of dementia.11,18
Androgen deprivation therapy may increase dementia risk through adverse effects on vascular health. Both decreased testosterone levels and ADT have been associated with cardiometabolic diseases, including diabetes, coronary artery disease, stroke and peripheral arterial disease,26–29 which have in turn been associated with dementia risk.30–32 Autopsy series demonstrate that the sensitivity and specificity for a clinical diagnosis of Alzheimer’s disease are in the 70–90% and 40–70% range, respectively, with primary alternative pathologies, including cerebrovascular disease.33 Therefore, it is likely that studies examining Alzheimer’s disease include individuals with alternative pathologies such as vascular dementia. If the impact of ADT on vascular disease is a primary driver of the association between ADT and dementia, this misclassification could artificially strengthen the association of ADT and Alzheimer’s disease. Alternatively, the magnitude of effect of ADT on Alzheimer’s disease-specific risk factors4,8 may be smaller and thus larger samples sizes would be needed to demonstrate a statistically significant association. This may explain the finding in our study of a non-significant (P = 0.06) association between ADT and Alzheimer’s disease compared to a statistically significant association of greater magnitude with all-cause dementia.
An alternative explanation for these findings is that the largely observational and retrospective analyses included in this review are confounded. In general, individuals in the general population undergoing ADT are older and have a greater burden of medical comorbidities.15 All of the studies included in our quantitative metaanalysis, except one,23 were retrospective population-based analyses. Therefore, given concern for confounding by factors like age, we only included effect estimates that accounted for confounding factors. Specifically, four studies undertook multivariable regression,13–15,24 two studies undertook propensity score matching,11,18 and one study was a multivariable adjusted secondary analysis of the randomized data.23 While confounding alone is unlikely to completely explain the association seen between ADT and dementia, it cannot be excluded as a source of bias in our analysis.
Our primary analysis showed significant between study heterogeneity with details of individual studies summarized in Tables 2 and 3. Exclusion of the outlying studies15,18 individually resulted in resolution of heterogeneity with consistent results compared to our primary analysis. The study by Nead et al.12 differed from other included studies, in that, it used informatics-based text analysis of the electronic medical record to the capture data from clinical notes, in addition to the coded date used in the other studies. This clinical text analysis has been shown to have a high degree of accuracy in detecting health outcomes34 and may capture events that would be missed using more traditional methods. While it could represent an opportunity for outcome misclassification, it is unclear that this would be differential and therefore introduce a bias, but this cannot be excluded. The outlying study by Khosrow-Khavar et al.15 conducted an analysis utilizing a United Kingdom based primary care database. While the authors acknowledge the limitations of this data source such as incomplete data regarding secondary care and the lack of standardization of outcome definitions,35 this represents the largest study to date and utilizes robust methodology to control for bias.
This study has limitations that warrant consideration. First, as detailed above, this meta-analysis largely included retrospective studies and we therefore only included studies that conducted multivariable adjusted regression analyses. Second, studies largely relied on diagnostic codes for dementia diagnosis, which may be subject to misclassification.36 Even in controlled settings not reliant on diagnostic codes, the diagnosis of dementia remains clinical and susceptible to misclassification. While there is evidence to support relatively high accuracy in the clinical diagnosis of dementia,33,37 we cannot exclude the introduction of a misclassification bias given the reliance of included studies on claims-based analysis. Third, we were not able to evaluate the risk of dementia by prostate cancer risk groups, which would provide clinically important information for risk-benefit evaluation, particularly among intermediate risk patients.38 Finally, we were unable to conduct subgroup analyses by type of ADT, which will be important in future studies as various forms of ADT have differing effects on the hypothalamic–pituitary–gonadal axis and could differentially impact dementia risk.39
The chronic health implications of cancer therapies are of increasing importance as survival rates following cancer diagnoses continue to improve.40 Additionally, as screening recommendations and treatment practices for prostate cancer evolve,41,42 informed and thoughtful utilization of systemic therapies will be critical. In this study, we further support the association of ADT in the treatment of prostate cancer with an increased risk of dementia and add to the growing body of evidence linking ADT with neurocognitive dysfunction.9,10,43 Further prospective studies are needed to validate this finding and to explore ways of reducing this possible cognitive harm. Additionally, we should continue to evaluate the impact of ADT on short-and long-term cognitive function in general. In the meantime, the potential for neurocognitive deficits secondary to ADT, specifically dementia, should be discussed with patients.
- Bolla M, Collette L, Blank L, Warde P, Dubois JB, Mirimanoff RO et al. Long-term results with immediate androgen suppression and external irradiation in
patients with locally advanced prostate cancer (an EORTC study): a phase III randomised trial. Lancet 2002; 360: 103–106.
- Ferlay J, Soerjomataram I, Dikshit R, Eser S, Mathers C, Rebelo M et al. Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012. Int J Cancer 2015; 136: E359–E386.
- Meng MV, Grossfeld GD, Sadetsky N, Mehta SS, Lubeck DP, Carroll PR. Contemporary patterns of androgen deprivation therapy use for newly diagnosed prostate cancer. Urology 2002; 60: 7–11 discussion 11-12.
- Vest RS, Pike CJ. Gender, sex steroid hormones, and Alzheimer's disease. Hormones Behav 2013; 63: 301–307.
- Carcaillon L, Brailly-Tabard S, Ancelin ML, Tzourio C, Foubert-Samier A, Dartigues JF et al. Low testosterone and the risk of dementia in elderly men: Impact of age and education. Alzheimers Dement 2014; 10: S306–S314.
- Rosario ER, Chang L, Stanczyk FZ, Pike CJ. Age-related testosterone depletion and the development of Alzheimer disease. JAMA 2004; 292: 1431–1432.
- Cherrier MM, Matsumoto AM, Amory JK, Asthana S, Bremner W, Peskind ER et al. Testosterone improves spatial memory in men with Alzheimer disease and mild cognitive impairment. Neurology 2005; 64: 2063–2068.
- Gandy S, Almeida OP, Fonte J, Lim D, Waterrus A, Spry N et al. Chemical andropause and amyloid-beta peptide. JAMA 2001; 285: 2195–2196.
- McGinty HL, Phillips KM, Jim HS, Cessna JM, Asvat Y, Cases MG et al. Cognitive functioning in men receiving androgen deprivation therapy for prostate cancer: a systematic review and meta-analysis. Support Care Cancer 2014; 22: 2271–2280.
- Gonzalez BD, Jim HS, Booth-Jones M, Small BJ, Sutton SK, Lin HY et al. Course and predictors of cognitive function in patients with prostate cancer receiving androgen-deprivation therapy: a controlled comparison. J Clin Oncol 2015; 33: 2021–2027.
- Nead KT, Gaskin G, Chester C, Swisher-McClure S, Dudley JT, Leeper NJ et al. Androgen deprivation therapy and future alzheimer's disease risk. J Clin Oncol 2016; 34: 566–571.
- Nead KT, Gaskin G, Chester C, Swisher-McClure S, Leeper NJ, Shah NH. Association between androgen deprivation therapy and risk of dementia. JAMA Oncol 2017; 3: 49–55.
- Chung SD, Lin HC, Tsai MC, Kao LT, Huang CY, Chen KC. Androgen deprivation therapy did not increase the risk of Alzheimer's and Parkinson's disease in patients with prostate cancer. Andrology 2016; 4: 481–485.
- Kao LT, Lin HC, Chung SD, Huang CY. No increased risk of dementia in patients receiving androgen deprivation therapy for prostate cancer: a 5-year follow-up study. Asian J Androl 2016. e-pub ahead of print 27 March 2016; doi: 10.4103/1008-682X.1799528.
- Khosrow-Khavar F, Rej S, Yin H, Aprikian A, Azoulay L. Androgen deprivation therapy and the risk of dementia in patients with prostate cancer. J Clin Oncol 2016; 35: 201–207, JCO2016696203.
- Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000; 283: 2008–2012.
- Stang A. Critical evaluation of the Newcastle-Ottawa scale for the assessment of the quality of nonrandomized studies in meta-analyses. Eur J Epidemiol 2010; 25: 603–605.
- Nead KT, Gaskin G, Chester C, Swisher-McClure S, Leeper NJ, Shah NH. Association between androgen deprivation therapy and risk of dementia. JAMA Oncol 2016; 3: 49–55.
- Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003; 327: 557–560.
- Higgins JPT, Green S (eds.). Cochrane handbook for systematic reviews of interventions version 5.1.0. Cochrane Collaboration 2011. Available at www.handbook.cochrane.org.
- Shahinian VB, Kuo YF, Freeman JL, Goodwin JS. Risk of the "androgen deprivation syndrome" in men receiving androgen deprivation for prostate cancer. Arch Internal Med 2006; 166: 465–471.
- Wiechno PJ, Sadowska M, Kalinowski T, Michalski W, Demkow T. Does pharmacological castration as adjuvant therapy for prostate cancer after radiotherapy affect anxiety and depression levels, cognitive functions and quality of life? Psychooncology 2013; 22: 346–351.
- Hershman DL, Unger JM, Wright JD, Ramsey S, Till C, Tangen CM et al. Adverse health events following intermittent and continuous androgen deprivation in patients with metastatic prostate cancer. JAMA Oncol 2016; 2: 453–461.
- Capitanio U, Isbarn H, Jeldres C, Gallina A, Gagne SB, Suardi N et al. The use of luteinizing hormone releasing hormone agonists administrated to patients with prostate cancer predisposes to dementia: a population-based analysis. J Urol 2009; 181: 296.
- Isbarn H, Jeldres C, Capitanio U, Gallina A, Suardi N, Baillargeon-Gagne S et al. A systematic analysis of the detrimental effect of orchiectomy on 12 systemic morbidities. Eur Urol Suppl 2009; 8: 331.
- Keating NL, O'Malley AJ, Smith MR. Diabetes and cardiovascular disease during androgen deprivation therapy for prostate cancer. J Clin Oncol 2006; 24: 4448–4456.
- Wang C, Jackson G, Jones TH, Matsumoto AM, Nehra A, Perelman MA et al. Low testosterone associated with obesity and the metabolic syndrome contributes to sexual dysfunction and cardiovascular disease risk in men with type 2 diabetes. Diabetes Care 2011; 34: 1669–1675.
- Tsai HK, D'Amico AV, Sadetsky N, Chen MH, Carroll PR. Androgen deprivation therapy for localized prostate cancer and the risk of cardiovascular mortality. J Natl Cancer Inst 2007; 99: 1516–1524.
- Hu JC, Williams SB, O'Malley AJ, Smith MR, Nguyen PL, Keating NL. Androgendeprivation therapy for nonmetastatic prostate cancer is associated with an increased risk of peripheral arterial disease and venous thromboembolism. Eur Urol 2012; 61: 1119–1128.
- Newman AB, Fitzpatrick AL, Lopez O, Jackson S, Lyketsos C, Jagust W et al. Dementia and Alzheimer's disease incidence in relationship to cardiovascular disease in the Cardiovascular Health Study cohort. J Am Geriatr Soc 2005; 53: 1101–1107.
- Tatemichi TK, Paik M, Bagiella E, Desmond DW, Stern Y, Sano M et al. Risk of dementia after stroke in a hospitalized cohort: results of a longitudinal study. Neurology 1994; 44: 1885–1891.
- Luchsinger JA, Tang MX, Stern Y, Shea S, Mayeux R. Diabetes mellitus and risk of Alzheimer's disease and dementia with stroke in a multiethnic cohort. Am J Epidemiol 2001; 154: 635–641.
- Beach TG, Monsell SE, Phillips LE, Kukull W. Accuracy of the clinical diagnosis of Alzheimer disease at National Institute on Aging Alzheimer Disease Centers, 2005-2010. J Neuropathol Exp Neurol 2012; 71: 266–273.
- LePendu P, Iyer SV, Bauer-Mehren A, Harpaz R, Mortensen JM, Podchiyska T et al. Pharmacovigilance using clinical notes. Clin Pharmacol Ther 2013; 93: 547–555.
- Herrett E, Gallagher AM, Bhaskaran K, Forbes H, Mathur R, van Staa T et al. Data Resource Profile: Clinical Practice Research Datalink (CPRD). Int J Epidemiol 2015; 44: 827–836.
- Kukull WA, Bowen JD. Dementia epidemiology. Med Clin North Am 2002; 86: 573–590.
- Snowden JS, Thompson JC, Stopford CL, Richardson AM, Gerhard A, Neary D et al. The clinical diagnosis of early-onset dementias: diagnostic accuracy and clinicopathological relationships. Brain J Neurol 2011; 134: 2478–2492.
- Nguyen PL. Rethinking the balance of risk and benefit of androgen deprivation therapy for intermediate-risk prostate cancer. Int J Radiat Oncol Biol Phys2016; 94: 975–977.
- D'Amico AV, Braccioforte MH, Moran BJ, Chen MH. Luteinizing-hormone releasing hormone therapy and the risk of death from Alzheimer disease. Alzheimer Dis Assoc Disord 2010; 24: 85–89.
- Rowland JH, Bellizzi KM. Cancer survivorship issues: life after treatment and implications for an aging population. J Clin Oncol 2014; 32: 2662–2668.
- Hamdy FC, Donovan JL, Lane JA, Mason M, Metcalfe C, Holding P et al. 10-Year outcomes after monitoring, surgery, or radiotherapy for localized prostate cancer. N Engl J Med 2016; 375: 1415–1424.
- Jemal A, Fedewa SA, Ma J, Siegel R, Lin CC, Brawley O et al. Prostate cancer incidence and PSA testing patterns in relation to USPSTF screening recommendations. JAMA 2015; 314: 2054–2061.
- Dinh KT, Reznor G, Muralidhar V, Mahal BA, Nezolosky MD, Choueiri TK et al. Association of androgen deprivation therapy with depression in localized prostate cancer. J Clin Oncol 2016; 34: 1905–1912.
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved 1365-7852/17
Written By: KT Nead1, S Sinha2 and PL Nguyen2
1Department of Radiation Oncology, Perelman Center for Advanced Medicine, University of Pennsylvania Perelman School of Medicine, Philadelphia, PA, USA
2Department of Radiation Oncology, Dana-Farber Cancer Institute and Brigham and Women's Hospital, Harvard Medical School, Boston, MA, USA.
A letter from the desk of the associate editor: J. Kellogg Parsons