- Email: [email protected]
Spontaneous Reversion of Mild Cognitive Impairment to Normal Cognition: A Systematic Review of Literature and Meta-Analysis
JAMDA xxx (2016) 1e6
JAMDA journal homepage: www.jamda.com
Original Study
Spontaneous Reversion of Mild Cognitive Impairment to Normal Cognition: A Systematic Review of Literature and Meta-Analysis Marco Canevelli MD a, *, Giulia Grande MD b, Eleonora Lacorte MSci c, Elisa Quarchioni MStat c, Matteo Cesari PhD d, e, Claudio Mariani MD b, Giuseppe Bruno PhD a, Nicola Vanacore PhD c a
Department of Neurology and Psychiatry, “Sapienza” University of Rome, Italy Center for Research and Treatment on Cognitive Dysfunctions, Biomedical and Clinical Sciences Department, “Luigi Sacco” Hospital, University of Milan, Italy c National Centre for Epidemiology, Surveillance, and Health Promotion, National Institute of Health, Rome, Italy d Gérontopôle, Centre Hospitalier Universitaire de Toulouse, Toulouse, France e Université de Toulouse III Paul Sabatier, Toulouse, France b
a b s t r a c t Keywords: Mild cognitive impairment dementia longitudinal studies overdiagnosis
Objectives: The issue of subjects with mild cognitive impairment (MCI) reverting to normal cognition (NC) has to date been taken in limited consideration, and no conclusive data are available on the rate of reversion. We aimed at systematically reviewing available longitudinal studies on MCI and metaanalyzing data with the purpose of estimating the proportion of subjects reverting to NC. Design: We performed a systematic bibliographic search on PubMed, the Cochrane Library, and the ISI Web of Science databases. We included in the review all longitudinal studies on MCI published from 1999 up to November 2015. Only studies with a longitudinal design, a follow-up 2 years, enrolling subjects with MCI, and reporting the number or the percentage of subjects reverting to NC were included. Data extraction was performed independently by 2 authors. The methodological quality of studies was also assessed by 2 independent authors using the QUIPS tool. Results: Twenty-five studies were included. The quality of evidence was found to be moderate. We observed an overall 18% (95% CI 14e22) reversion rate from MCI to NC. Results from the metaregression showed a significant association between effect size and study setting. In particular, estimates significantly varied according to study setting, with an 8% (95% CI 4e11) reversion rate in clinical-based studies and a 25% (95% CI 19e30) rate in population-based studies. The frequency of reversion from MCI to NC further increased to 26% when considering only studies of better quality. Only a few studies were designed to specifically investigate the reversion from MCI to NC, thus relevant information on this topic was frequently missing. Conclusion: Our data confirm that reversion to normality is a common outcome in subjects with MCI, thus leading to recommend a more balanced view when approaching the construct of MCI both in a clinical and in a research setting. Ó 2016 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Matteo Cesari received a research grant from Pfizer, and is Work-package leader of a project funded by the Innovative Medicines Initiative, including in the study consortium members of the European Federation of Pharmaceutical Industries and Associations (Sanofi, Servier, GlaxoSmithKline, Novartis, and Eli Lilly). He has also received honoraria from Pfizer and Nestlé for presentations at scientific meetings. The other authors declare no conflicts of interest. * Address correspondence to Marco Canevelli, MD, Memory Clinic, Department of Neurology and Psychiatry, “Sapienza” University of Rome, Viale dell’Università 30, 00185 Rome, Italy. E-mail address: [email protected] (M. Canevelli). http://dx.doi.org/10.1016/j.jamda.2016.06.020 1525-8610/Ó 2016 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
Mild cognitive impairment (MCI) is usually described as an intermediate phase between normal cognition and dementia.1,2 A subject with MCI is defined as having an objective deficit in cognitive abilities that does not affect his or her functional independence. The interest in investigating this condition has progressively increased over the last years. In particular, MCI is considered as being a relevant risk factor for dementia, and thus a promising target for specific pharmacologic and nonpharmacologic interventions. Subjects with MCI, in fact, show an annual rate of progression to dementia ranging from 5% to 15%, varying according to the setting and the operational definitions
2
M. Canevelli et al. / JAMDA xxx (2016) 1e6
considered.2 MCI is therefore frequently the object of investigation in the attempt of detecting the early pathophysiological modifications that may be responsible for the progression to dementia.3 The actual nature of MCI is not yet fully clarified, and several issues are still insufficiently explored. As an example, although a large amount of resources has been dedicated to studying cognitive decline and the way MCI progresses to dementia, relatively few efforts have been focused on investigating the possible reversion of subjects with MCI to normal cognition.4 Such tendency at univocally looking at MCI as a prodromal stage of dementia produces an unbalanced use of this construct.5,6 This negative connotation of MCI due to such a biased approach frequently leads to underestimating its fluctuations over time, and forgetting its potential to (spontaneously) revert to normal cognition. An increasing amount of longitudinal data show that the majority of subjects with MCI do not experience a worsening of cognition over time,7 and that a relevant proportion of them eventually reverts to normal cognition.8,9 Available estimates of reversion from MCI to normal cognition are quite heterogeneous, ranging from 2.1%10 to as far as 53%.11 Investigating the reversion from MCI to normal cognition (NC) has some relevant practical implications. A more accurate identification of subjects with a positive cognitive outcome may allow to better allocate healthcare resources among the heterogeneous population of MCI subjects.5 It may also prevent the misdiagnosis of cognitively normal subjects and its consequences (eg, discrimination, stigmatization, and overmedicalization).12 Moreover, a better understanding of the nature of MCI may improve the design and interpretation of clinical trials, particularly those focused on the prevention of dementia. For example, the enrollment of subjects with MCI whose cognitive status is unlikely to decline over time may result in reducing the effect size of potentially effective interventions. The aim of this study is to systematically review, analyze, and discuss results from available longitudinal studies with the aim of obtaining a more accurate estimate of the proportion of subjects with MCI reverting to NC. Analyses will also explore the role of some wellknown confounding factors that affect the outcome of MCI [ie, setting, age of participants, length of follow-up, operational definition of MCI and NC, concomitant depression, functional independence, and apolipoprotein E (ApoE) genotype].1 Methods The review was performed according to the methodology recommended by the Cochrane Collaboration group13 and reported according to the PRISMA statement14 and the MOOSE checklist.15 Data Sources and Searches All studies published between 1999 (year of the first MCI operationalization16) and November 2015 were retrieved through a structured search on PubMed, Cochrane Library, and the ISI Web of Science databases carried out by a researcher with experience in bibliographic searches. The following search terms were used: ((“mild” AND “cognitive” AND “impairment”) OR “MCI”) AND (reversion* OR remission* OR remitter* OR reverter* OR revertion OR *conversion* OR *converter* OR *progressi*) AND (“normal” OR “normality” OR “dementia” OR “dementias” OR alzheimer*) AND (cohort* OR longitudinal OR prospective OR prognos* OR “follow up” OR “follow-up” OR “followup”). All longitudinal studies enrolling subjects with MCI and investigating any type of main outcome (ie, MCI progression to dementia and/or MCI reversion to NC) were considered for evaluation. This conservative approach allowed us to consider also studies providing either marginally or nonexplicitly information on reversion from MCI to NC (eg, reporting this information as a secondary outcome or “between the lines”).
Study Selection Records identified through the bibliographic searches were independently reviewed by 2 authors (M.C. and G.G.). Articles relevant and pertinent to the topic of the review were thus selected, based on the analysis of titles and abstracts, and retrieved in full text. Disagreements in the selection process, when present, were solved by discussion, consensus, or involving a third reviewer. Possible sources of gray literature and the references of considered studies were also reviewed to identify further potentially relevant publications. Articles considered for inclusion were then individually applied a set of predefined inclusion and exclusion criteria. Only studies in English, with a longitudinal design and a follow-up equal to or longer than 2 years, enrolling subjects with MCI defined according to the original Mayo Clinic criteria16 or subsequent operationalizations, and reporting the number or percentage of subjects reverting to NC during the follow-up were included. Studies focusing on other cognitive deficits or similar conditions with the potential of being considered as prodromal stages of dementia or predementia (ie, “cognitive impairment no dementia” [CIND], “age-associated memory impairment” [AAMI], and “ageassociated cognitive decline” [AACD]) were excluded in order to obtain a more homogeneous body of evidence. Conference proceedings, abstracts, and letters were also excluded. Data Extraction and Quality Assessment Two authors (M.C. and G.G.) independently extracted the following data for each included study: setting, number of participants, sociodemographic characteristics (ie, age, gender, and education), criteria adopted for defining MCI and NC, cognitive performance, functional independence, concomitant depression, ApoE genotype, length of follow-up, number and percentage of MCI participants reverting to NC, and response rate. Disagreements on the extracted information, where present, were resolved by consensus. Two different authors (E.L. and N.V.) independently assessed the quality of included studies using the Quality in Prognostic Studies (QUIPS) tool.17,18 This tool has been developed for systematic appraisal in studies on prognostic factors and considers 6 domains for analysis of potential biases: (1) inclusion, (2) attrition, (3) prognostic factor measurement, (4) confounders, (5) outcome measurement, and (6) analysis and reporting. A total of 3 to 7 prompting items is provided for each domain, to help assessing the presence of risk of bias and score it as high, moderate, or low.18 The QUIPS has been successfully adopted in several systematic reviews with moderate to substantial interrater reliability.18 Data Synthesis and Analysis Preliminary descriptive statistics [n, mean, standard deviation (SD)] were conducted to describe the samples to be included in the meta-analysis. Weighted mean values of some variables (ie, age, educational level) were calculated when only data referring to subgroups of participants were provided. A meta-analysis of the reported frequencies of MCI reversion to NC (whenever available) was conducted. All analyses were carried out using Stata (version 11.0). Meta-analyses were performed adopting a specific Stata module, Metaprop, designed to perform meta-analyses of proportions in Stata. Building on the existing Stata procedure Metan (typically used to pool risk ratios, odds ratios, mean differences, and proportions), Metaprop applies procedures that are specific to binomial data.19 Overall estimates were calculated with random effects models and a test for heterogeneity was applied using chi-square and the I2 statistics. The random effects model was chosen because the
M. Canevelli et al. / JAMDA xxx (2016) 1e6
retained studies represent a random sample of effect size, so the summary effect was a mean estimate of all effects.20 A metaregression was performed to handle heterogeneity, using the Stata procedure Metareg. Metaregression is a multivariate regression model in which the characteristics either of the studies or of the participants in the studies are used as covariates while the effect size is the dependent variable. The following variables were chosen as covariates: setting, followup length, and mean age of participants at baseline. These factors were chosen because of their influence on the outcome of MCI over time1 and on the heterogeneity of the samples enrolled in the included studies. As a further method to deal with heterogeneity, subgroup analyses were carried out considering all factors that were found to be significant in the metaregression. Results A total of 2338 articles were identified through the bibliographic searches. The flow diagram literature selection and inclusion is reported in Figure A1. The selection of studies based on titles and abstracts lead to the exclusion of 2277 studies, thus leaving 61 studies to be assessed for inclusion. By applying the predefined inclusion and exclusion criteria, we further excluded 36 studies. The main reasons for exclusion were a follow-up <2 years and the study not reporting the number of subjects reverting to NC. In 7 cases (n ¼ 14 articles), 2 articles were found to include the same sample of subjects; thus, the article providing the largest amount of information for each sample was included. One study9 was included though having a mean length of follow-up of 1.9 years, because the majority of enrolled subjects reached the threshold of 2 years of observation (SD ¼ 0.1 years). A final sample of 25 studies was included in the qualitative and quantitative analyses.9,10,21e43 The main characteristics of the included studies are reported in Table A1. Study Quality Results from the qualitative assessment of included studies are reported in Table A2. The 2 authors who performed the assessment resulted having a 90% interrater agreement. Overall, the quality of the included studies was moderate. We defined as being of “better quality” the studies reporting a low risk of bias in at least 4 domains of 6. Only 6 studies9,29,33,37,40,43 complied with this definition and were considered as being of “better quality.” Five of these 6 studies were population-based, whereas only 1 had been performed in a clinical setting.29 Additional information on the quality assessment of included studies is reported in the Appendix. Settings, Subjects A total of 6914 subjects with MCI was considered. Sample sizes widely ranged between 2228 and 184327 individuals. The weighted mean age of enrolled subjects (calculated on 20 studies with available information) was 74.8 (SD ¼ 6.6) years, ranging between 53.632 and 80.7.23 Women (51.3%) were slightly more prevalent than men (data about gender was available in 23 studies). Participants had a high educational level (weighted mean of 12.2, SD ¼ 3.0, years of education; calculated on 15 studies with available data). Fifteen studies were population-based, and recruited participants through electoral rolls, door-to-door surveys, and referral from general practitioners. In contrast, 10 studies enrolled participants evaluated in clinical settings (eg, memory clinics). The included studies had a follow-up range between 1.99 and 632 years. Overall, populationbased studies had longer observation periods (mean follow-up 3.9
3
vs 3.2 years) and lower sample sizes (mean n 250.7 vs 334.3) compared with those recruiting from clinical settings. Operationalization of MCI and Definition of NC MCI was mostly defined by adopting the standard criteria developed by the International Working Group (IWG) in 200444 (used in 10 studies) or the original Mayo Clinic criteria proposed in 199916 (adopted by 8 studies). In nearly one-third of the studies, the MCI definition did not rely on standardized criteria, but was based on a consensus of experienced clinicians, on Mini Mental State Examination (MMSE) scores, or a combination of neuropsychological and functional measures. No study adopted the research criteria recently developed by the National Institute on AgingeAlzheimer’s Association.45 Studies conducted in subjects referring to clinical settings were more likely to adopt standardized criteria, whereas population-based studies more frequently elide on adapted MCI definitions (probably from largely taking advantage of data collected for different purposes). Despite this relatively univocal approach in the definition of MCI (mostly based on few and largely convergent procedures), a substantial heterogeneity concerning how these definitions were implemented was found. In particular, studies using the same MCI criteria were found to adopt different neuropsychological measures in the practical translation of the definitions. For example, among the studies adopting the IWG criteria, memory functions were variously assessed through the Wechsler Memory Scale,46 the Free-and-Cued Selective Reminding Test,47 or the Rey Auditory Verbal Learning Test,48 thus leading to a marked variability in the operationalization of the condition of interest. The definition of NC was less often reported, being specifically described only in half of the included studies. When available, the definition of NC mostly relied on standardized test scores. Participants were usually classified as “cognitively normal” if the score of the different neuropsychological tests was 1.5 SDs within the age- and (less frequently) gender-adjusted normative means. Some studies adopted more restrictive criteria, assuming 1 SD as the cut-point for NC. Among the studies not explicitly defining NC, subjects were implicitly considered as “normal” if not fulfilling the adopted criteria for MCI or dementia. Concomitant Depression, Functional Independence, ApoE Genotype Nearly 75% of the retained studies investigated the presence of a concomitant depressive disorder, defined according to standardized diagnostic criteria or through validated screening tools. Nineteen studies described the assessment of the functional independence of participants in the conduction of complex activities, a crucial aspect of the current MCI operational definitions. Most of the studies assessed functional autonomy by the use of the Instrumental Activities of Daily Living (IADL) scale.49 A greater heterogeneity was found for the weight of functional independence in the operationalization of MCI definitions. In this regard, most of studies included a minimal impairment in IADL for fulfilling the MCI criteria, whereas others required a full integrity of functional capacities. Finally, only 7 studies provided information about ApoE genotype. Rate of MCI Reversion to NC A total of 1243 (18.0%) subjects with MCI were reported to revert to NC in the samples enrolled in the included studies. The meta-analysis of the rates reported in the included studies resulted in an overall 18% (95% CI ¼ 14e22) reversion rate, with a relatively high degree of heterogeneity (I2 ¼ 96.1%, P < .001). Results from the metaregression showed a nonsignificant association between effect size and length of follow-up (P ¼ .356), whereas a
4
M. Canevelli et al. / JAMDA xxx (2016) 1e6
significant association was observed between effect size and setting (P ¼ .011), and an association close to statistical significance was observed between effect size and mean age (P ¼ .082), though both regression coefficients for setting and mean age were close to zero (0.163 and 0.008, respectively). However, I2 remained high (93.6%). Based on the results from metaregression, a first subgroup analysis was performed according to the setting in which the studies were carried out (Figure 1). Analyses resulted respectively in a 10% (95% CI ¼ 6e13) reversion rate for the 10 studies carried out in a clinical setting, and a 23% (95% CI ¼ 18e29) reversion rate in the 15 population-based studies. We found a possible outlier value in each setting. A 40% reversion rate was reported in one of the studies performed in a clinical setting,32 whereas the remaining 9 studies reported rates ranging between 2% and 14%. Reviewing the study, however, we found that nearly 60% of the enrolled subjects had reported a family history of Alzheimer disease (AD). As for the population-based setting, a 4% reversion rate was reported in 1 study,39 whereas the other 14 studies reported rates ranging between 12% and 41%. Reviewing the publication, we observed that a possible explanation for such a peculiar finding could be that the study had been carried out in a small rural village in Tanzania. A sensitivity analysis excluding the 2 outliers was thus performed. The exclusion of the 2 values did not substantially modify the overall reversion rates
(overall ¼ 18%, 95% CI ¼ 14e22; clinical setting ¼ 8%, 95% CI ¼ 4e11; population-based ¼ 25%, 95% CI ¼ 19e30). A further meta-analysis including only the studies defined as being of “better quality” was then performed. The analysis resulted in an overall 26% reversion rate. Considering that 6 of the 7 studies included in this analysis were population based, we performed a further analysis excluding the only study carried out in a clinical setting.29 The exclusion of this study increased the resulting reversion rate to 29% (95% CI ¼ 22e37). Results of all subgroup analyses are reported in the Appendix. Discussion According to our findings, reversion to NC is a quite common outcome in subjects with MCI, with an estimated 18% reversion rate. In other words, nearly 1 of 5 subjects with MCI shows a complete and spontaneous remission of cognitive symptoms and deficits over time. Considering that our analyses only included studies with a relatively long follow-up (ie, longer than 2 years), these results could be very likely underestimating the phenomenon. The reason we included only studies with a length of follow-up 2 years was predetermined and based on the attempt to exclude as much as possible the influence of temporary fluctuations due to weaknesses in the diagnostic tools, and
Fig. 1. Subgroup analysis of all studies stratified per setting. Data from single studies were meta-analyzed to obtain a global cumulative estimate, represented by the lower diamond in the graph, and specific cumulative estimates for either population-based studies or clinical-based studies, represented by the remaining 2 diamonds in the graph.
M. Canevelli et al. / JAMDA xxx (2016) 1e6
thus have more reliable, as based on long-term data, and realistic estimates. To our knowledge, as of today, only 1 systematic review and metaanalysis has recently tried to measure the rate of reversion from MCI to NC.4 The review reports even higher rates of reversion from MCI to NC both in the community setting (31%) and in the clinical setting (14%). However, the study by Malek-Ahmadi et al is based on a different set of studies, as a different set of eligibility criteria was applied. In particular, authors did not include a minimum length of follow-up among the eligibility criteria; thus, several included studies were based on a short period of observation. As said, this choice might have potentially lead to overestimate the rate of reversion. Moreover, the review included only studies adopting the Mayo Clinic criteria to define amnestic MCI, thus restricting the sample, and did not perform a quality assessment of the included evidence. Our review is thus the first systematic review performing a structured and objective quality assessment of the included studies. Reversion rates resulting from our meta-analysis were found to be scarcely influenced by either the mean age of participants or the length of follow-up. Comparable rates of reversion were, in fact, observed also among the oldest subjects and in the studies with the longest follow-up. These aspects suggest that reversion may indeed be considered as an intrinsic potentiality of MCI. However, as already reported and established in the literature,4,8 estimates of reversion were substantially affected by the setting in which the observation is carried out. The rarely studied but extremely relevant phenomenon of reversion from MCI to NC could be explained by several factors.9 A first explanation may be that it simply depends on a misclassification of the subjects, thus having either that cognitively normal subjects are initially misdiagnosed and classified as MCI or that subjects with MCI are misclassified as normal at follow-up. This may be due to weaknesses in the definitions and diagnostic tools.50 It might also be that MCI is an unstable and fluctuating condition, with subjects actually experiencing a reversion to NC over time. Therefore, MCI cannot be considered as necessarily being the first manifestation of an underlying neurodegenerative process, as it can be caused by transient and reversible conditions (eg, nutritional deficits, affective disorders, cerebrovascular events, physical frailty, sleep disorders, and social issues). Thus, in these cases, the progression to dementia cannot be considered neither obvious nor unavoidable, and the treatment of such conditions could explain the positive outcome.5,51 In this regard, it has already been shown that the identification and correction of potentially reversible medical conditions (ie, correction of visual loss, and discontinuation of anticholinergic medications) is associated with cognitive improvement/reversion in a sizeable proportion of individuals with mild cognitive disturbances.52 Our findings underline the need of extreme caution and a more balanced attitude when approaching the MCI construct both in the clinical and in the research setting. The bidirectional outcome of MCI should prompt a careful consideration of the risks of misdiagnosing cognitive impairment, in particular the mild forms, to a cognitively normal subject. A diagnosis of MCI can cause a subject a number of potentially harmful consequences, such as discrimination, stigmatization, and overmedicalization.5 Subjects diagnosed with MCI can experience an increase of anxiety and/or stress that can trigger a cascade reaction worsening their cognitive status, or induce them to undergo superfluous or unjustified examinations and/or interventions.12 The different outcomes of MCI (ie, conversion, stabilization, and reversion) should also be carefully considered when designing studies on the management of this condition. Widening the target population to include both patients with definite dementia and subjects with MCI might surely allow to capture the early phases of a neurodegenerative process, but it might also automatically lead to include a higher number of false positive cases.5 This means that
5
subjects who would never develop AD nor any type of dementia might end up being enrolled in an RCT, thus receiving useless, if not harmful, treatments, and reducing the effect size of possibly effective interventions. The adoption of biomarkers reflecting in vivo the occurrence of any neuropathologic variation has been hypothesized to facilitate the detection of cases of MCI with an underlying neurodegenerative condition, thus allowing a more accurate enrollment of subjects with a higher probability of developing dementia over time. However, this is clearly unfeasible in the routine clinical practice and should be limited to the research setting, so as to produce more robust evidence on the topic. Moreover, to our knowledge, no rate of reversion from MCI to NC has, to date, been estimated based on a definition of MCI including the assessment of biomarkers.45 This issue is extremely relevant, considering that 274 RCTs are currently recruiting subjects with MCI worldwide (source: www. clinicaltrials.gov; search updated in February 2016). This means that a wide part of research is currently being carried out on subjects who would never progress to dementia but would instead probably revert to NC. Adopting a more balanced view of MCI means that, considering the frequency of reversion to NC, the multidimensional approach to this condition should be strengthened. The assessment of MCI should thus not be limited to the neurologic aspects, but a more inclusive approach should be adopted to identify and target the reversible risk factors that may facilitate or promote a positive outcome of MCI (eg, sedentariness, healthy diet, social isolation, etc).5 Available evidence on this topic have some limitations that should be discussed, as they could potentially affect the results from this review. Only a small number of the included studies were designed to specifically investigate the reversion from MCI to NC. As a consequence, relevant information, such as incidence rates expressed in person-years, were rarely available. These types of data could have allowed a more precise estimate of the frequency, particularly when considering the length of follow-up. Only a small number of the included studies reported the rate of reversion in terms of risk estimates, with the majority of studies providing results as simple frequencies. Moreover, potential predictive factors of reversion to NC were rarely investigated, and further long-term outcomes after reversion (eg, reconversion to MCI or progression to dementia43) were mostly unaddressed. All these limitations underline the need for further high-quality and ad hoc studies on the topic. Conclusion In conclusion, reversion to NC is a common outcome in subjects with MCI. This further increases the already relevant heterogeneity of this condition. Considering its unstable and potentially bidirectional course, MCI should not be approached as neither a disease nor a stage of a disease, but more conservatively as an heterogeneous risk factor. Thus, subjects with MCI should not be considered as being in a prodromal phase of an unavoidable dementia, as they also have a considerable probability of reverting, even spontaneously, to normal cognition. This more cautious and balanced perspective will allow to avoid an overdiagnosis of cognitive disorders, thus consequently preventing the “labeling” of a normal individual as having a cognitive impairment, to facilitate the design and interpretation of RCTs enrolling subjects with cognitive disorders, and to adopt a more holistic approach to MCI aimed at a tailored and personalized management of both pharmacologic and nonpharmacologic interventions. Supplementary Data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jamda.2016.06.020.
6
M. Canevelli et al. / JAMDA xxx (2016) 1e6
References 1. Langa KM, Levine DA. The diagnosis and management of mild cognitive impairment: A clinical review. JAMA 2014;312:2551e2561. 2. Petersen RC. Clinical practice. Mild cognitive impairment. N Engl J Med 2011; 364:2227e2234. 3. Jack CR, Albert MS, Knopman DS, et al. Introduction to the recommendations from the National Institute on AgingeAlzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 2011;7: 257e262. 4. Malek-Ahmadi M. Reversion from mild cognitive impairment to normal cognition: A meta-analysis. Alzheimer Dis Assoc Disord; 2016 Feb 22 [Epub ahead of print]. 5. Canevelli M, Blasimme A, Vanacore N, et al. From evidence to action: Promoting a multidimensional approach to mild cognitive impairment. J Am Med Dir Assoc 2015;16:710e711. 6. Morley JE. Mild cognitive impairmentda treatable condition. J Am Med Dir Assoc 2014;15:1e5. 7. Mitchell AJ, Shiri-Feshki M. Rate of progression of mild cognitive impairment to dementiadMeta-analysis of 41 robust inception cohort studies. Acta Psychiatr Scand 2009;119:252e265. 8. Koepsell TD, Monsell SE. Reversion from mild cognitive impairment to normal or near-normal cognition: Risk factors and prognosis. Neurology 2012;79: 1591e1598. 9. Sachdev PS, Lipnicki DM, Crawford J, et al. Factors predicting reversion from mild cognitive impairment to normal cognitive functioning: A populationbased study. PLoS One 2013;8:e59649. 10. Morbelli S, Brugnolo A, Bossert I, et al. Visual versus semi-quantitative analysis of 18F-FDG-PET in amnestic MCI: An European Alzheimer’s Disease Consortium (EADC) project. J Alzheimers Dis 2015;44:815e826. 11. Ganguli M, Snitz BE, Saxton JA, et al. Outcomes of mild cognitive impairment by definition: A population study. Arch Neurol 2011;68:761e767. 12. Canevelli M, Blasimme A, Vanacore N, et al. Issues about the use of subjective cognitive decline in Alzheimer’s disease research. Alzheimers Dement 2014; 10:881e882. 13. The Cochrane Collaboration. Cochrane Handbook for Systematic Reviews of Interventions [Internet]. Available at: http://handbook.cochrane.org/; 2011. Accessed April 14, 2016. 14. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: Explanation and elaboration. BMJ 2009;339:b2700. 15. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in epidemiology: A proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000;283:2008e2012. 16. Petersen RC, Smith GE, Waring SC, et al. Mild cognitive impairment: Clinical characterization and outcome. Arch Neurol 1999;56:303e308. 17. Hayden JA, Côté P, Bombardier C. Evaluation of the quality of prognosis studies in systematic reviews. Ann Intern Med 2006;144:427e437. 18. Hayden JA, van der Windt DA, Cartwright JL, et al. Assessing bias in studies of prognostic factors. Ann Intern Med 2013;158:280e286. 19. Nyaga VN, Arbyn M, Aerts M. Metaprop: A Stata command to perform metaanalysis of binomial data. Arch Public Health 2014;72:39. 20. Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. A basic introduction to fixed-effect and random-effects models for meta-analysis. Res Synth Methods 2010;1:97e111. 21. Aretouli E, Tsilidis KK, Brandt J. Four-year outcome of mild cognitive impairment: The contribution of executive dysfunction. Neuropsychology 2013;27: 95e106. 22. Britt WG, Hansen AM, Bhaskerrao S, et al. Mild cognitive impairment: Prodromal Alzheimer’s disease or something else? J Alzheimers Dis 2011;27: 543e551. 23. Fisk JD, Merry HR, Rockwood K. Variations in case definition affect prevalence but not outcomes of mild cognitive impairment. Neurology 2003;61: 1179e1184. 24. Gallassi R, Oppi F, Poda R, et al. Are subjective cognitive complaints a risk factor for dementia? Neurol Sci 2010;31:327e336. 25. Ganguli M, Dodge HH, Shen C, DeKosky ST. Mild cognitive impairment, amnestic type: An epidemiologic study. Neurology 2004;63:115e121. 26. Gao S, Unverzagt FW, Hall KS, et al. Mild cognitive impairment, incidence, progression, and reversion: Findings from a community-based cohort of elderly African Americans. Am J Geriatr Psychiatry 2014;22:670e681.
27. Gifford KA, Liu D, Lu Z, et al. The source of cognitive complaints predicts diagnostic conversion differentially among nondemented older adults. Alzheimers Dement 2014;10:319e327. 28. Godinho C, Camozzato AL, Onyszko D, Chaves ML. Estimation of the risk of conversion of mild cognitive impairment of Alzheimer type to Alzheimer’s disease in a south Brazilian population-based elderly cohort: The PALA study. Int Psychogeriatr 2012;24:674e681. 29. Grande G, Cucumo V, Cova I, et al. Reversible mild cognitive impairment: The role of comorbidities at baseline evaluation. J Alzheimers Dis 2016;51:57e67. 30. Ishikawa T. Mild cognitive impairment in a population-based epidemiological study. Psychogeriatr Off J Jpn Psychogeriatr Soc 2007;7:104e108. 31. Kaduszkiewicz H, Eisele M, Wiese B, et al. Prognosis of mild cognitive impairment in general practice: Results of the German AgeCoDe study. Ann Fam Med 2014;12:158e165. 32. Koscik RL, La Rue A, Jonaitis EM, et al. Emergence of mild cognitive impairment in late middle-aged adults in the Wisconsin Registry for Alzheimer’s Prevention. Dement Geriatr Cogn Disord 2014;38:16e30. 33. Larrieu S, Letenneur L, Orgogozo JM, et al. Incidence and outcome of mild cognitive impairment in a population-based prospective cohort. Neurology 2002;59:1594e1599. 34. Loewenstein DA, Acevedo A, Small BJ, et al. Stability of different subtypes of mild cognitive impairment among the elderly over a 2- to 3-year follow-up period. Dement Geriatr Cogn Disord 2009;27:418e423. 35. Lopez OL, Kuller LH, Becker JT, et al. Incidence of dementia in mild cognitive impairment in the cardiovascular health study cognition study. Arch Neurol 2007;64:416e420. 36. Luis CA, Barker WW, Loewenstein DA, et al. Conversion to dementia among two groups with cognitive impairment. A preliminary report. Dement Geriatr Cogn Disord 2004;18:307e313. 37. Manly JJ, Tang M-X, Schupf N, et al. Frequency and course of mild cognitive impairment in a multiethnic community. Ann Neurol 2008;63:494e506. 38. Nordlund A, Rolstad S, Göthlin M, et al. Cognitive profiles of incipient dementia in the Goteborg MCI study. Dement Geriatr Cogn Disord 2010;30:403e410. 39. Paddick S-M, Kisoli A, Samuel M, et al. Mild cognitive impairment in rural Tanzania: Prevalence, profile, and outcomes at 4-year follow-up. Am J Geriatr Psychiatry 2015;23:950e959. 40. Pérès K, Chrysostome V, Fabrigoule C, et al. Restriction in complex activities of daily living in MCI: Impact on outcome. Neurology 2006;67:461e466. 41. Perneczky R, Tsolakidou A, Arnold A, et al. CSF soluble amyloid precursor proteins in the diagnosis of incipient Alzheimer disease. Neurology 2011;77: 35e38. 42. Perri R, Serra L, Carlesimo GA, Caltagirone C, Early Diagnosis Group of Italian Interdisciplinary Network on Alzheimer’s Disease. Preclinical dementia: An Italian multicentre study on amnestic mild cognitive impairment. Dement Geriatr Cogn Disord 2007;23:289e300. 43. Roberts RO, Knopman DS, Mielke MM, et al. Higher risk of progression to dementia in mild cognitive impairment cases who revert to normal. Neurology 2014;82:317e325. 44. Winblad B, Palmer K, Kivipelto M, et al. Mild cognitive impairmentdbeyond controversies, towards a consensus: Report of the International Working Group on Mild Cognitive Impairment. J Intern Med 2004;256:240e246. 45. Albert MS, DeKosky ST, Dickson D, et al. The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 2011;7:270e279. 46. Wechsler D. Wechsler Memory ScaleeRevised Edition. Manual. New York, NY: Psychological Corporation; 1987. 47. Buschke H. Cued recall in amnesia. J Clin Neuropsychol 1984;6:433e440. 48. Rey A. L’examen clinique en psychologie. Paris: Presses universitaires de France; 1958. 49. Lawton MP, Brody EM. Assessment of older people: Self-maintaining and instrumental activities of daily living. Gerontologist 1969;9:179e186. 50. Kochan NA, Slavin MJ, Brodaty H, et al. Effect of different impairment criteria on prevalence of “objective” mild cognitive impairment in a community sample. Am J Geriatr Psychiatry 2010;18:711e722. 51. Canevelli M, Kelaiditi E. The complex construct of mild cognitive impairment: Be aware of cognitive frailty. J Frailty Aging 2014;3:87e88. 52. Cruz-Oliver DM, Malmstrom TK, Roegner M, et al. Cognitive deficit reversal as shown by changes in the Veterans Affairs Saint Louis University Mental Status (SLUMS) examination scores 7.5 years later. J Am Med Dir Assoc 2014;15:687. e5e687.e10.
JAMDA journal homepage: www.jamda.com
Original Study
Spontaneous Reversion of Mild Cognitive Impairment to Normal Cognition: A Systematic Review of Literature and Meta-Analysis Marco Canevelli MD a, *, Giulia Grande MD b, Eleonora Lacorte MSci c, Elisa Quarchioni MStat c, Matteo Cesari PhD d, e, Claudio Mariani MD b, Giuseppe Bruno PhD a, Nicola Vanacore PhD c a
Department of Neurology and Psychiatry, “Sapienza” University of Rome, Italy Center for Research and Treatment on Cognitive Dysfunctions, Biomedical and Clinical Sciences Department, “Luigi Sacco” Hospital, University of Milan, Italy c National Centre for Epidemiology, Surveillance, and Health Promotion, National Institute of Health, Rome, Italy d Gérontopôle, Centre Hospitalier Universitaire de Toulouse, Toulouse, France e Université de Toulouse III Paul Sabatier, Toulouse, France b
a b s t r a c t Keywords: Mild cognitive impairment dementia longitudinal studies overdiagnosis
Objectives: The issue of subjects with mild cognitive impairment (MCI) reverting to normal cognition (NC) has to date been taken in limited consideration, and no conclusive data are available on the rate of reversion. We aimed at systematically reviewing available longitudinal studies on MCI and metaanalyzing data with the purpose of estimating the proportion of subjects reverting to NC. Design: We performed a systematic bibliographic search on PubMed, the Cochrane Library, and the ISI Web of Science databases. We included in the review all longitudinal studies on MCI published from 1999 up to November 2015. Only studies with a longitudinal design, a follow-up 2 years, enrolling subjects with MCI, and reporting the number or the percentage of subjects reverting to NC were included. Data extraction was performed independently by 2 authors. The methodological quality of studies was also assessed by 2 independent authors using the QUIPS tool. Results: Twenty-five studies were included. The quality of evidence was found to be moderate. We observed an overall 18% (95% CI 14e22) reversion rate from MCI to NC. Results from the metaregression showed a significant association between effect size and study setting. In particular, estimates significantly varied according to study setting, with an 8% (95% CI 4e11) reversion rate in clinical-based studies and a 25% (95% CI 19e30) rate in population-based studies. The frequency of reversion from MCI to NC further increased to 26% when considering only studies of better quality. Only a few studies were designed to specifically investigate the reversion from MCI to NC, thus relevant information on this topic was frequently missing. Conclusion: Our data confirm that reversion to normality is a common outcome in subjects with MCI, thus leading to recommend a more balanced view when approaching the construct of MCI both in a clinical and in a research setting. Ó 2016 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Matteo Cesari received a research grant from Pfizer, and is Work-package leader of a project funded by the Innovative Medicines Initiative, including in the study consortium members of the European Federation of Pharmaceutical Industries and Associations (Sanofi, Servier, GlaxoSmithKline, Novartis, and Eli Lilly). He has also received honoraria from Pfizer and Nestlé for presentations at scientific meetings. The other authors declare no conflicts of interest. * Address correspondence to Marco Canevelli, MD, Memory Clinic, Department of Neurology and Psychiatry, “Sapienza” University of Rome, Viale dell’Università 30, 00185 Rome, Italy. E-mail address: [email protected] (M. Canevelli). http://dx.doi.org/10.1016/j.jamda.2016.06.020 1525-8610/Ó 2016 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
Mild cognitive impairment (MCI) is usually described as an intermediate phase between normal cognition and dementia.1,2 A subject with MCI is defined as having an objective deficit in cognitive abilities that does not affect his or her functional independence. The interest in investigating this condition has progressively increased over the last years. In particular, MCI is considered as being a relevant risk factor for dementia, and thus a promising target for specific pharmacologic and nonpharmacologic interventions. Subjects with MCI, in fact, show an annual rate of progression to dementia ranging from 5% to 15%, varying according to the setting and the operational definitions
2
M. Canevelli et al. / JAMDA xxx (2016) 1e6
considered.2 MCI is therefore frequently the object of investigation in the attempt of detecting the early pathophysiological modifications that may be responsible for the progression to dementia.3 The actual nature of MCI is not yet fully clarified, and several issues are still insufficiently explored. As an example, although a large amount of resources has been dedicated to studying cognitive decline and the way MCI progresses to dementia, relatively few efforts have been focused on investigating the possible reversion of subjects with MCI to normal cognition.4 Such tendency at univocally looking at MCI as a prodromal stage of dementia produces an unbalanced use of this construct.5,6 This negative connotation of MCI due to such a biased approach frequently leads to underestimating its fluctuations over time, and forgetting its potential to (spontaneously) revert to normal cognition. An increasing amount of longitudinal data show that the majority of subjects with MCI do not experience a worsening of cognition over time,7 and that a relevant proportion of them eventually reverts to normal cognition.8,9 Available estimates of reversion from MCI to normal cognition are quite heterogeneous, ranging from 2.1%10 to as far as 53%.11 Investigating the reversion from MCI to normal cognition (NC) has some relevant practical implications. A more accurate identification of subjects with a positive cognitive outcome may allow to better allocate healthcare resources among the heterogeneous population of MCI subjects.5 It may also prevent the misdiagnosis of cognitively normal subjects and its consequences (eg, discrimination, stigmatization, and overmedicalization).12 Moreover, a better understanding of the nature of MCI may improve the design and interpretation of clinical trials, particularly those focused on the prevention of dementia. For example, the enrollment of subjects with MCI whose cognitive status is unlikely to decline over time may result in reducing the effect size of potentially effective interventions. The aim of this study is to systematically review, analyze, and discuss results from available longitudinal studies with the aim of obtaining a more accurate estimate of the proportion of subjects with MCI reverting to NC. Analyses will also explore the role of some wellknown confounding factors that affect the outcome of MCI [ie, setting, age of participants, length of follow-up, operational definition of MCI and NC, concomitant depression, functional independence, and apolipoprotein E (ApoE) genotype].1 Methods The review was performed according to the methodology recommended by the Cochrane Collaboration group13 and reported according to the PRISMA statement14 and the MOOSE checklist.15 Data Sources and Searches All studies published between 1999 (year of the first MCI operationalization16) and November 2015 were retrieved through a structured search on PubMed, Cochrane Library, and the ISI Web of Science databases carried out by a researcher with experience in bibliographic searches. The following search terms were used: ((“mild” AND “cognitive” AND “impairment”) OR “MCI”) AND (reversion* OR remission* OR remitter* OR reverter* OR revertion OR *conversion* OR *converter* OR *progressi*) AND (“normal” OR “normality” OR “dementia” OR “dementias” OR alzheimer*) AND (cohort* OR longitudinal OR prospective OR prognos* OR “follow up” OR “follow-up” OR “followup”). All longitudinal studies enrolling subjects with MCI and investigating any type of main outcome (ie, MCI progression to dementia and/or MCI reversion to NC) were considered for evaluation. This conservative approach allowed us to consider also studies providing either marginally or nonexplicitly information on reversion from MCI to NC (eg, reporting this information as a secondary outcome or “between the lines”).
Study Selection Records identified through the bibliographic searches were independently reviewed by 2 authors (M.C. and G.G.). Articles relevant and pertinent to the topic of the review were thus selected, based on the analysis of titles and abstracts, and retrieved in full text. Disagreements in the selection process, when present, were solved by discussion, consensus, or involving a third reviewer. Possible sources of gray literature and the references of considered studies were also reviewed to identify further potentially relevant publications. Articles considered for inclusion were then individually applied a set of predefined inclusion and exclusion criteria. Only studies in English, with a longitudinal design and a follow-up equal to or longer than 2 years, enrolling subjects with MCI defined according to the original Mayo Clinic criteria16 or subsequent operationalizations, and reporting the number or percentage of subjects reverting to NC during the follow-up were included. Studies focusing on other cognitive deficits or similar conditions with the potential of being considered as prodromal stages of dementia or predementia (ie, “cognitive impairment no dementia” [CIND], “age-associated memory impairment” [AAMI], and “ageassociated cognitive decline” [AACD]) were excluded in order to obtain a more homogeneous body of evidence. Conference proceedings, abstracts, and letters were also excluded. Data Extraction and Quality Assessment Two authors (M.C. and G.G.) independently extracted the following data for each included study: setting, number of participants, sociodemographic characteristics (ie, age, gender, and education), criteria adopted for defining MCI and NC, cognitive performance, functional independence, concomitant depression, ApoE genotype, length of follow-up, number and percentage of MCI participants reverting to NC, and response rate. Disagreements on the extracted information, where present, were resolved by consensus. Two different authors (E.L. and N.V.) independently assessed the quality of included studies using the Quality in Prognostic Studies (QUIPS) tool.17,18 This tool has been developed for systematic appraisal in studies on prognostic factors and considers 6 domains for analysis of potential biases: (1) inclusion, (2) attrition, (3) prognostic factor measurement, (4) confounders, (5) outcome measurement, and (6) analysis and reporting. A total of 3 to 7 prompting items is provided for each domain, to help assessing the presence of risk of bias and score it as high, moderate, or low.18 The QUIPS has been successfully adopted in several systematic reviews with moderate to substantial interrater reliability.18 Data Synthesis and Analysis Preliminary descriptive statistics [n, mean, standard deviation (SD)] were conducted to describe the samples to be included in the meta-analysis. Weighted mean values of some variables (ie, age, educational level) were calculated when only data referring to subgroups of participants were provided. A meta-analysis of the reported frequencies of MCI reversion to NC (whenever available) was conducted. All analyses were carried out using Stata (version 11.0). Meta-analyses were performed adopting a specific Stata module, Metaprop, designed to perform meta-analyses of proportions in Stata. Building on the existing Stata procedure Metan (typically used to pool risk ratios, odds ratios, mean differences, and proportions), Metaprop applies procedures that are specific to binomial data.19 Overall estimates were calculated with random effects models and a test for heterogeneity was applied using chi-square and the I2 statistics. The random effects model was chosen because the
M. Canevelli et al. / JAMDA xxx (2016) 1e6
retained studies represent a random sample of effect size, so the summary effect was a mean estimate of all effects.20 A metaregression was performed to handle heterogeneity, using the Stata procedure Metareg. Metaregression is a multivariate regression model in which the characteristics either of the studies or of the participants in the studies are used as covariates while the effect size is the dependent variable. The following variables were chosen as covariates: setting, followup length, and mean age of participants at baseline. These factors were chosen because of their influence on the outcome of MCI over time1 and on the heterogeneity of the samples enrolled in the included studies. As a further method to deal with heterogeneity, subgroup analyses were carried out considering all factors that were found to be significant in the metaregression. Results A total of 2338 articles were identified through the bibliographic searches. The flow diagram literature selection and inclusion is reported in Figure A1. The selection of studies based on titles and abstracts lead to the exclusion of 2277 studies, thus leaving 61 studies to be assessed for inclusion. By applying the predefined inclusion and exclusion criteria, we further excluded 36 studies. The main reasons for exclusion were a follow-up <2 years and the study not reporting the number of subjects reverting to NC. In 7 cases (n ¼ 14 articles), 2 articles were found to include the same sample of subjects; thus, the article providing the largest amount of information for each sample was included. One study9 was included though having a mean length of follow-up of 1.9 years, because the majority of enrolled subjects reached the threshold of 2 years of observation (SD ¼ 0.1 years). A final sample of 25 studies was included in the qualitative and quantitative analyses.9,10,21e43 The main characteristics of the included studies are reported in Table A1. Study Quality Results from the qualitative assessment of included studies are reported in Table A2. The 2 authors who performed the assessment resulted having a 90% interrater agreement. Overall, the quality of the included studies was moderate. We defined as being of “better quality” the studies reporting a low risk of bias in at least 4 domains of 6. Only 6 studies9,29,33,37,40,43 complied with this definition and were considered as being of “better quality.” Five of these 6 studies were population-based, whereas only 1 had been performed in a clinical setting.29 Additional information on the quality assessment of included studies is reported in the Appendix. Settings, Subjects A total of 6914 subjects with MCI was considered. Sample sizes widely ranged between 2228 and 184327 individuals. The weighted mean age of enrolled subjects (calculated on 20 studies with available information) was 74.8 (SD ¼ 6.6) years, ranging between 53.632 and 80.7.23 Women (51.3%) were slightly more prevalent than men (data about gender was available in 23 studies). Participants had a high educational level (weighted mean of 12.2, SD ¼ 3.0, years of education; calculated on 15 studies with available data). Fifteen studies were population-based, and recruited participants through electoral rolls, door-to-door surveys, and referral from general practitioners. In contrast, 10 studies enrolled participants evaluated in clinical settings (eg, memory clinics). The included studies had a follow-up range between 1.99 and 632 years. Overall, populationbased studies had longer observation periods (mean follow-up 3.9
3
vs 3.2 years) and lower sample sizes (mean n 250.7 vs 334.3) compared with those recruiting from clinical settings. Operationalization of MCI and Definition of NC MCI was mostly defined by adopting the standard criteria developed by the International Working Group (IWG) in 200444 (used in 10 studies) or the original Mayo Clinic criteria proposed in 199916 (adopted by 8 studies). In nearly one-third of the studies, the MCI definition did not rely on standardized criteria, but was based on a consensus of experienced clinicians, on Mini Mental State Examination (MMSE) scores, or a combination of neuropsychological and functional measures. No study adopted the research criteria recently developed by the National Institute on AgingeAlzheimer’s Association.45 Studies conducted in subjects referring to clinical settings were more likely to adopt standardized criteria, whereas population-based studies more frequently elide on adapted MCI definitions (probably from largely taking advantage of data collected for different purposes). Despite this relatively univocal approach in the definition of MCI (mostly based on few and largely convergent procedures), a substantial heterogeneity concerning how these definitions were implemented was found. In particular, studies using the same MCI criteria were found to adopt different neuropsychological measures in the practical translation of the definitions. For example, among the studies adopting the IWG criteria, memory functions were variously assessed through the Wechsler Memory Scale,46 the Free-and-Cued Selective Reminding Test,47 or the Rey Auditory Verbal Learning Test,48 thus leading to a marked variability in the operationalization of the condition of interest. The definition of NC was less often reported, being specifically described only in half of the included studies. When available, the definition of NC mostly relied on standardized test scores. Participants were usually classified as “cognitively normal” if the score of the different neuropsychological tests was 1.5 SDs within the age- and (less frequently) gender-adjusted normative means. Some studies adopted more restrictive criteria, assuming 1 SD as the cut-point for NC. Among the studies not explicitly defining NC, subjects were implicitly considered as “normal” if not fulfilling the adopted criteria for MCI or dementia. Concomitant Depression, Functional Independence, ApoE Genotype Nearly 75% of the retained studies investigated the presence of a concomitant depressive disorder, defined according to standardized diagnostic criteria or through validated screening tools. Nineteen studies described the assessment of the functional independence of participants in the conduction of complex activities, a crucial aspect of the current MCI operational definitions. Most of the studies assessed functional autonomy by the use of the Instrumental Activities of Daily Living (IADL) scale.49 A greater heterogeneity was found for the weight of functional independence in the operationalization of MCI definitions. In this regard, most of studies included a minimal impairment in IADL for fulfilling the MCI criteria, whereas others required a full integrity of functional capacities. Finally, only 7 studies provided information about ApoE genotype. Rate of MCI Reversion to NC A total of 1243 (18.0%) subjects with MCI were reported to revert to NC in the samples enrolled in the included studies. The meta-analysis of the rates reported in the included studies resulted in an overall 18% (95% CI ¼ 14e22) reversion rate, with a relatively high degree of heterogeneity (I2 ¼ 96.1%, P < .001). Results from the metaregression showed a nonsignificant association between effect size and length of follow-up (P ¼ .356), whereas a
4
M. Canevelli et al. / JAMDA xxx (2016) 1e6
significant association was observed between effect size and setting (P ¼ .011), and an association close to statistical significance was observed between effect size and mean age (P ¼ .082), though both regression coefficients for setting and mean age were close to zero (0.163 and 0.008, respectively). However, I2 remained high (93.6%). Based on the results from metaregression, a first subgroup analysis was performed according to the setting in which the studies were carried out (Figure 1). Analyses resulted respectively in a 10% (95% CI ¼ 6e13) reversion rate for the 10 studies carried out in a clinical setting, and a 23% (95% CI ¼ 18e29) reversion rate in the 15 population-based studies. We found a possible outlier value in each setting. A 40% reversion rate was reported in one of the studies performed in a clinical setting,32 whereas the remaining 9 studies reported rates ranging between 2% and 14%. Reviewing the study, however, we found that nearly 60% of the enrolled subjects had reported a family history of Alzheimer disease (AD). As for the population-based setting, a 4% reversion rate was reported in 1 study,39 whereas the other 14 studies reported rates ranging between 12% and 41%. Reviewing the publication, we observed that a possible explanation for such a peculiar finding could be that the study had been carried out in a small rural village in Tanzania. A sensitivity analysis excluding the 2 outliers was thus performed. The exclusion of the 2 values did not substantially modify the overall reversion rates
(overall ¼ 18%, 95% CI ¼ 14e22; clinical setting ¼ 8%, 95% CI ¼ 4e11; population-based ¼ 25%, 95% CI ¼ 19e30). A further meta-analysis including only the studies defined as being of “better quality” was then performed. The analysis resulted in an overall 26% reversion rate. Considering that 6 of the 7 studies included in this analysis were population based, we performed a further analysis excluding the only study carried out in a clinical setting.29 The exclusion of this study increased the resulting reversion rate to 29% (95% CI ¼ 22e37). Results of all subgroup analyses are reported in the Appendix. Discussion According to our findings, reversion to NC is a quite common outcome in subjects with MCI, with an estimated 18% reversion rate. In other words, nearly 1 of 5 subjects with MCI shows a complete and spontaneous remission of cognitive symptoms and deficits over time. Considering that our analyses only included studies with a relatively long follow-up (ie, longer than 2 years), these results could be very likely underestimating the phenomenon. The reason we included only studies with a length of follow-up 2 years was predetermined and based on the attempt to exclude as much as possible the influence of temporary fluctuations due to weaknesses in the diagnostic tools, and
Fig. 1. Subgroup analysis of all studies stratified per setting. Data from single studies were meta-analyzed to obtain a global cumulative estimate, represented by the lower diamond in the graph, and specific cumulative estimates for either population-based studies or clinical-based studies, represented by the remaining 2 diamonds in the graph.
M. Canevelli et al. / JAMDA xxx (2016) 1e6
thus have more reliable, as based on long-term data, and realistic estimates. To our knowledge, as of today, only 1 systematic review and metaanalysis has recently tried to measure the rate of reversion from MCI to NC.4 The review reports even higher rates of reversion from MCI to NC both in the community setting (31%) and in the clinical setting (14%). However, the study by Malek-Ahmadi et al is based on a different set of studies, as a different set of eligibility criteria was applied. In particular, authors did not include a minimum length of follow-up among the eligibility criteria; thus, several included studies were based on a short period of observation. As said, this choice might have potentially lead to overestimate the rate of reversion. Moreover, the review included only studies adopting the Mayo Clinic criteria to define amnestic MCI, thus restricting the sample, and did not perform a quality assessment of the included evidence. Our review is thus the first systematic review performing a structured and objective quality assessment of the included studies. Reversion rates resulting from our meta-analysis were found to be scarcely influenced by either the mean age of participants or the length of follow-up. Comparable rates of reversion were, in fact, observed also among the oldest subjects and in the studies with the longest follow-up. These aspects suggest that reversion may indeed be considered as an intrinsic potentiality of MCI. However, as already reported and established in the literature,4,8 estimates of reversion were substantially affected by the setting in which the observation is carried out. The rarely studied but extremely relevant phenomenon of reversion from MCI to NC could be explained by several factors.9 A first explanation may be that it simply depends on a misclassification of the subjects, thus having either that cognitively normal subjects are initially misdiagnosed and classified as MCI or that subjects with MCI are misclassified as normal at follow-up. This may be due to weaknesses in the definitions and diagnostic tools.50 It might also be that MCI is an unstable and fluctuating condition, with subjects actually experiencing a reversion to NC over time. Therefore, MCI cannot be considered as necessarily being the first manifestation of an underlying neurodegenerative process, as it can be caused by transient and reversible conditions (eg, nutritional deficits, affective disorders, cerebrovascular events, physical frailty, sleep disorders, and social issues). Thus, in these cases, the progression to dementia cannot be considered neither obvious nor unavoidable, and the treatment of such conditions could explain the positive outcome.5,51 In this regard, it has already been shown that the identification and correction of potentially reversible medical conditions (ie, correction of visual loss, and discontinuation of anticholinergic medications) is associated with cognitive improvement/reversion in a sizeable proportion of individuals with mild cognitive disturbances.52 Our findings underline the need of extreme caution and a more balanced attitude when approaching the MCI construct both in the clinical and in the research setting. The bidirectional outcome of MCI should prompt a careful consideration of the risks of misdiagnosing cognitive impairment, in particular the mild forms, to a cognitively normal subject. A diagnosis of MCI can cause a subject a number of potentially harmful consequences, such as discrimination, stigmatization, and overmedicalization.5 Subjects diagnosed with MCI can experience an increase of anxiety and/or stress that can trigger a cascade reaction worsening their cognitive status, or induce them to undergo superfluous or unjustified examinations and/or interventions.12 The different outcomes of MCI (ie, conversion, stabilization, and reversion) should also be carefully considered when designing studies on the management of this condition. Widening the target population to include both patients with definite dementia and subjects with MCI might surely allow to capture the early phases of a neurodegenerative process, but it might also automatically lead to include a higher number of false positive cases.5 This means that
5
subjects who would never develop AD nor any type of dementia might end up being enrolled in an RCT, thus receiving useless, if not harmful, treatments, and reducing the effect size of possibly effective interventions. The adoption of biomarkers reflecting in vivo the occurrence of any neuropathologic variation has been hypothesized to facilitate the detection of cases of MCI with an underlying neurodegenerative condition, thus allowing a more accurate enrollment of subjects with a higher probability of developing dementia over time. However, this is clearly unfeasible in the routine clinical practice and should be limited to the research setting, so as to produce more robust evidence on the topic. Moreover, to our knowledge, no rate of reversion from MCI to NC has, to date, been estimated based on a definition of MCI including the assessment of biomarkers.45 This issue is extremely relevant, considering that 274 RCTs are currently recruiting subjects with MCI worldwide (source: www. clinicaltrials.gov; search updated in February 2016). This means that a wide part of research is currently being carried out on subjects who would never progress to dementia but would instead probably revert to NC. Adopting a more balanced view of MCI means that, considering the frequency of reversion to NC, the multidimensional approach to this condition should be strengthened. The assessment of MCI should thus not be limited to the neurologic aspects, but a more inclusive approach should be adopted to identify and target the reversible risk factors that may facilitate or promote a positive outcome of MCI (eg, sedentariness, healthy diet, social isolation, etc).5 Available evidence on this topic have some limitations that should be discussed, as they could potentially affect the results from this review. Only a small number of the included studies were designed to specifically investigate the reversion from MCI to NC. As a consequence, relevant information, such as incidence rates expressed in person-years, were rarely available. These types of data could have allowed a more precise estimate of the frequency, particularly when considering the length of follow-up. Only a small number of the included studies reported the rate of reversion in terms of risk estimates, with the majority of studies providing results as simple frequencies. Moreover, potential predictive factors of reversion to NC were rarely investigated, and further long-term outcomes after reversion (eg, reconversion to MCI or progression to dementia43) were mostly unaddressed. All these limitations underline the need for further high-quality and ad hoc studies on the topic. Conclusion In conclusion, reversion to NC is a common outcome in subjects with MCI. This further increases the already relevant heterogeneity of this condition. Considering its unstable and potentially bidirectional course, MCI should not be approached as neither a disease nor a stage of a disease, but more conservatively as an heterogeneous risk factor. Thus, subjects with MCI should not be considered as being in a prodromal phase of an unavoidable dementia, as they also have a considerable probability of reverting, even spontaneously, to normal cognition. This more cautious and balanced perspective will allow to avoid an overdiagnosis of cognitive disorders, thus consequently preventing the “labeling” of a normal individual as having a cognitive impairment, to facilitate the design and interpretation of RCTs enrolling subjects with cognitive disorders, and to adopt a more holistic approach to MCI aimed at a tailored and personalized management of both pharmacologic and nonpharmacologic interventions. Supplementary Data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jamda.2016.06.020.
6
M. Canevelli et al. / JAMDA xxx (2016) 1e6
References 1. Langa KM, Levine DA. The diagnosis and management of mild cognitive impairment: A clinical review. JAMA 2014;312:2551e2561. 2. Petersen RC. Clinical practice. Mild cognitive impairment. N Engl J Med 2011; 364:2227e2234. 3. Jack CR, Albert MS, Knopman DS, et al. Introduction to the recommendations from the National Institute on AgingeAlzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 2011;7: 257e262. 4. Malek-Ahmadi M. Reversion from mild cognitive impairment to normal cognition: A meta-analysis. Alzheimer Dis Assoc Disord; 2016 Feb 22 [Epub ahead of print]. 5. Canevelli M, Blasimme A, Vanacore N, et al. From evidence to action: Promoting a multidimensional approach to mild cognitive impairment. J Am Med Dir Assoc 2015;16:710e711. 6. Morley JE. Mild cognitive impairmentda treatable condition. J Am Med Dir Assoc 2014;15:1e5. 7. Mitchell AJ, Shiri-Feshki M. Rate of progression of mild cognitive impairment to dementiadMeta-analysis of 41 robust inception cohort studies. Acta Psychiatr Scand 2009;119:252e265. 8. Koepsell TD, Monsell SE. Reversion from mild cognitive impairment to normal or near-normal cognition: Risk factors and prognosis. Neurology 2012;79: 1591e1598. 9. Sachdev PS, Lipnicki DM, Crawford J, et al. Factors predicting reversion from mild cognitive impairment to normal cognitive functioning: A populationbased study. PLoS One 2013;8:e59649. 10. Morbelli S, Brugnolo A, Bossert I, et al. Visual versus semi-quantitative analysis of 18F-FDG-PET in amnestic MCI: An European Alzheimer’s Disease Consortium (EADC) project. J Alzheimers Dis 2015;44:815e826. 11. Ganguli M, Snitz BE, Saxton JA, et al. Outcomes of mild cognitive impairment by definition: A population study. Arch Neurol 2011;68:761e767. 12. Canevelli M, Blasimme A, Vanacore N, et al. Issues about the use of subjective cognitive decline in Alzheimer’s disease research. Alzheimers Dement 2014; 10:881e882. 13. The Cochrane Collaboration. Cochrane Handbook for Systematic Reviews of Interventions [Internet]. Available at: http://handbook.cochrane.org/; 2011. Accessed April 14, 2016. 14. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: Explanation and elaboration. BMJ 2009;339:b2700. 15. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in epidemiology: A proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000;283:2008e2012. 16. Petersen RC, Smith GE, Waring SC, et al. Mild cognitive impairment: Clinical characterization and outcome. Arch Neurol 1999;56:303e308. 17. Hayden JA, Côté P, Bombardier C. Evaluation of the quality of prognosis studies in systematic reviews. Ann Intern Med 2006;144:427e437. 18. Hayden JA, van der Windt DA, Cartwright JL, et al. Assessing bias in studies of prognostic factors. Ann Intern Med 2013;158:280e286. 19. Nyaga VN, Arbyn M, Aerts M. Metaprop: A Stata command to perform metaanalysis of binomial data. Arch Public Health 2014;72:39. 20. Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. A basic introduction to fixed-effect and random-effects models for meta-analysis. Res Synth Methods 2010;1:97e111. 21. Aretouli E, Tsilidis KK, Brandt J. Four-year outcome of mild cognitive impairment: The contribution of executive dysfunction. Neuropsychology 2013;27: 95e106. 22. Britt WG, Hansen AM, Bhaskerrao S, et al. Mild cognitive impairment: Prodromal Alzheimer’s disease or something else? J Alzheimers Dis 2011;27: 543e551. 23. Fisk JD, Merry HR, Rockwood K. Variations in case definition affect prevalence but not outcomes of mild cognitive impairment. Neurology 2003;61: 1179e1184. 24. Gallassi R, Oppi F, Poda R, et al. Are subjective cognitive complaints a risk factor for dementia? Neurol Sci 2010;31:327e336. 25. Ganguli M, Dodge HH, Shen C, DeKosky ST. Mild cognitive impairment, amnestic type: An epidemiologic study. Neurology 2004;63:115e121. 26. Gao S, Unverzagt FW, Hall KS, et al. Mild cognitive impairment, incidence, progression, and reversion: Findings from a community-based cohort of elderly African Americans. Am J Geriatr Psychiatry 2014;22:670e681.
27. Gifford KA, Liu D, Lu Z, et al. The source of cognitive complaints predicts diagnostic conversion differentially among nondemented older adults. Alzheimers Dement 2014;10:319e327. 28. Godinho C, Camozzato AL, Onyszko D, Chaves ML. Estimation of the risk of conversion of mild cognitive impairment of Alzheimer type to Alzheimer’s disease in a south Brazilian population-based elderly cohort: The PALA study. Int Psychogeriatr 2012;24:674e681. 29. Grande G, Cucumo V, Cova I, et al. Reversible mild cognitive impairment: The role of comorbidities at baseline evaluation. J Alzheimers Dis 2016;51:57e67. 30. Ishikawa T. Mild cognitive impairment in a population-based epidemiological study. Psychogeriatr Off J Jpn Psychogeriatr Soc 2007;7:104e108. 31. Kaduszkiewicz H, Eisele M, Wiese B, et al. Prognosis of mild cognitive impairment in general practice: Results of the German AgeCoDe study. Ann Fam Med 2014;12:158e165. 32. Koscik RL, La Rue A, Jonaitis EM, et al. Emergence of mild cognitive impairment in late middle-aged adults in the Wisconsin Registry for Alzheimer’s Prevention. Dement Geriatr Cogn Disord 2014;38:16e30. 33. Larrieu S, Letenneur L, Orgogozo JM, et al. Incidence and outcome of mild cognitive impairment in a population-based prospective cohort. Neurology 2002;59:1594e1599. 34. Loewenstein DA, Acevedo A, Small BJ, et al. Stability of different subtypes of mild cognitive impairment among the elderly over a 2- to 3-year follow-up period. Dement Geriatr Cogn Disord 2009;27:418e423. 35. Lopez OL, Kuller LH, Becker JT, et al. Incidence of dementia in mild cognitive impairment in the cardiovascular health study cognition study. Arch Neurol 2007;64:416e420. 36. Luis CA, Barker WW, Loewenstein DA, et al. Conversion to dementia among two groups with cognitive impairment. A preliminary report. Dement Geriatr Cogn Disord 2004;18:307e313. 37. Manly JJ, Tang M-X, Schupf N, et al. Frequency and course of mild cognitive impairment in a multiethnic community. Ann Neurol 2008;63:494e506. 38. Nordlund A, Rolstad S, Göthlin M, et al. Cognitive profiles of incipient dementia in the Goteborg MCI study. Dement Geriatr Cogn Disord 2010;30:403e410. 39. Paddick S-M, Kisoli A, Samuel M, et al. Mild cognitive impairment in rural Tanzania: Prevalence, profile, and outcomes at 4-year follow-up. Am J Geriatr Psychiatry 2015;23:950e959. 40. Pérès K, Chrysostome V, Fabrigoule C, et al. Restriction in complex activities of daily living in MCI: Impact on outcome. Neurology 2006;67:461e466. 41. Perneczky R, Tsolakidou A, Arnold A, et al. CSF soluble amyloid precursor proteins in the diagnosis of incipient Alzheimer disease. Neurology 2011;77: 35e38. 42. Perri R, Serra L, Carlesimo GA, Caltagirone C, Early Diagnosis Group of Italian Interdisciplinary Network on Alzheimer’s Disease. Preclinical dementia: An Italian multicentre study on amnestic mild cognitive impairment. Dement Geriatr Cogn Disord 2007;23:289e300. 43. Roberts RO, Knopman DS, Mielke MM, et al. Higher risk of progression to dementia in mild cognitive impairment cases who revert to normal. Neurology 2014;82:317e325. 44. Winblad B, Palmer K, Kivipelto M, et al. Mild cognitive impairmentdbeyond controversies, towards a consensus: Report of the International Working Group on Mild Cognitive Impairment. J Intern Med 2004;256:240e246. 45. Albert MS, DeKosky ST, Dickson D, et al. The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 2011;7:270e279. 46. Wechsler D. Wechsler Memory ScaleeRevised Edition. Manual. New York, NY: Psychological Corporation; 1987. 47. Buschke H. Cued recall in amnesia. J Clin Neuropsychol 1984;6:433e440. 48. Rey A. L’examen clinique en psychologie. Paris: Presses universitaires de France; 1958. 49. Lawton MP, Brody EM. Assessment of older people: Self-maintaining and instrumental activities of daily living. Gerontologist 1969;9:179e186. 50. Kochan NA, Slavin MJ, Brodaty H, et al. Effect of different impairment criteria on prevalence of “objective” mild cognitive impairment in a community sample. Am J Geriatr Psychiatry 2010;18:711e722. 51. Canevelli M, Kelaiditi E. The complex construct of mild cognitive impairment: Be aware of cognitive frailty. J Frailty Aging 2014;3:87e88. 52. Cruz-Oliver DM, Malmstrom TK, Roegner M, et al. Cognitive deficit reversal as shown by changes in the Veterans Affairs Saint Louis University Mental Status (SLUMS) examination scores 7.5 years later. J Am Med Dir Assoc 2014;15:687. e5e687.e10.