ORIGINAL ARTICLE

Prospective comparison of the prognostic utility of the MiniMental State Examination and the Montreal CognitiveAssessment in patients with brain metastases

Robert Olson & Scott Tyldesley & Hannah Carolan &

Maureen Parkinson & Taruna Chhanabhai &Michael McKenzie

Received: 22 June 2010 /Accepted: 11 October 2010 /Published online: 19 October 2010# Springer-Verlag 2010

AbstractPurpose The Mini Mental State Examination (MMSE) isthe most commonly chosen cognitive screening test (CST)in clinical practice and trials, despite its poor sensitivity,likely because of its prognostic utility. The MontrealCognitive Assessment (MoCA) is an alternative CST, ismore sensitive, and is better correlated with quality of life.Methods Sixty-five patients with brain metastases wereprospectively accrued and completed both the MMSE andMoCA. We compared the prognostic utility of both CSTs.Results The mean age of patients was 59.0 years; 42.0% hadsingle brain metastases. Median MMSE and MoCA scoreswere 28 and 22, respectively. Median overall survival (OS)was worse for individuals with below- versus above-averageMMSE scores (10.4 versus 36.3 weeks, p=0.007). Likewise,below- versus above-average MoCA scores were prognostic(6.3 versus 50.0 weeks, p<0.001). Median OS for MoCAscores <22, 22–26, and >26 were 6.3, 30.9, and 61.7 weeks,respectively (p<0.001). On multivariable analysis, below-average MMSE scores were no longer prognostic (hazardratio [HR]=1.71 [0.90–3.26]), though below-average MoCA

scores were (HR=5.44 [2.70–10.94]). Furthermore, theMoCA demonstrated superior prognostic utility when com-paring multivariable models with continuous CST scores.Conclusions Our results indicate that the MoCA is asuperior prognostic indicator than the MMSE. Furthermore,given its superior sensitivity and better correlation withquality of life, the MoCA should be preferentially chosen inclinical practice and trials.

Keywords Montreal Cognitive Assessment . Mini MentalState Examination . Prognosis . Brain metastases . Cognition

Introduction

Assessing cognitive function is important for patients withbrain tumors in both clinical and research settings.Clinically, cognitive assessments are essential in identifyingwho would benefit most from cognitive adaptive strategiesand neurocognitive rehabilitation [1–3]. Furthermore, cog-nitive impairment has been identified as an independentpredictor of survival and disease recurrence in patients withboth brain metastases and primary brain tumors [4–7].Cognitive function is likely prognostic because it asensitive indicator of brain function, but potentially alsobecause it is a product of often unmeasured prognosticpatient and tumor factors, such as socioeconomic status,tumor location, delirium, and other comorbidities [8].Because of its prognostic utility, cognitive function acts asa stratification variable in clinical trials and is essential inresearch investigating the toxicity of different treatmentregimens [6, 9]. In order to limit selection bias, cognitivefunction is often assessed using brief cognitive screeningtests (CSTs) [10].

R. Olson (*) : S. Tyldesley :H. Carolan :M. Parkinson :T. Chhanabhai :M. McKenzieVancouver Cancer Centre, BC Cancer Agency,600 West 10th Avenue,Vancouver, BC, Canada V5Z 4E6e-mail: rolson2@bccancer.bc.ca

R. Olson : S. Tyldesley :H. Carolan :M. McKenzieDivision of Radiation Oncology and DevelopmentalRadiotherapeutics, University of British Columbia,Vancouver, BC, Canada

M. ParkinsonFraser Valley Cancer Centre, BC Cancer Agency,Vancouver, BC, Canada

Support Care Cancer (2011) 19:1849–1855DOI 10.1007/s00520-010-1028-1

The most widely used CST in patients with both primaryand secondary brain tumors is the Mini Mental StateExamination (MMSE), despite limited validation in thissetting [2, 11, 12]. In addition, the MMSE continues to bechosen in large clinical trials, even after reporting of its lowsensitivity [6, 13–15]. Furthermore, the MMSE is relativelyeasy, which results in a ceiling test effect, where mostpatients have normal test scores in this population [10]. Themost likely rationale for the continued popularity of theMMSE, despite these limitations, is the demonstratedprognostic utility in patients with primary brain tumorsand brain metastases and those who have been admitted topalliative care wards [2, 7, 16].

Given the limitations of the MMSE, we explored theMontreal Cognitive Assessment (MoCA) as a potentialalternative CST, since it is more sensitive in detecting mildcognitive impairment [17]. There are several potentialbenefits to identifying a more sensitive CST. Clinically,this includes better allocation of cognitive rehabilitationservices, more accurate evaluation of positive treatmentresponse, and better prognostication. In clinical trials, moreaccurate CSTs are essential when investigating the cogni-tive impact of various treatment modalities and intensities.

We previously demonstrated that the MoCA is quick andwell tolerated by patients with brain metastases [18]. Incontrast to the MMSE, the MoCA has a wide range in testscores with little ceiling test effect [10, 17, 18]. In addition,the MoCA is more sensitive than the MMSE in patientswith brain tumors, though is less specific at the a priori cut-off scores chosen [19]. The improved sensitivity is likely aresult of the MoCA’s more extensive testing of executivefunctioning, delayed recall, and abstraction [10, 17–19].Lastly, the MoCA is better correlated with self-reportedcommunity integration and quality of life than the MMSE[19]. Since cognitive function is a well-established prog-nostic marker in brain tumor patients and since the MoCAis more sensitive than the MMSE, we sought to determineif the MoCA is a more useful prognostic test, with thepotential to identify a new role for this brief assessment.

Patients and methods

The study was approved by the University of BritishColumbia Cancer Agency Research Ethics Board. Nopotential conflicts of interest exist.

Participants

Sixty-five patients with brain metastases were prospectivelyaccrued to two studies comparing the MMSE and MoCA[10, 18, 19]. For both studies, if cognitive impairment wassevere enough to potentially limit the ability to obtain

informed consent, then families or caregivers were con-tacted to confirm consent. Formal assessments of capacitywere not deemed necessary by the local research ethicsboard, given the minimal risk associated with the studies.

In the first study, the feasibility of administering these CSTsin 40 patients with brain metastases was compared and isdescribed in detail elsewhere [10, 18]. In brief, English-speaking patients with radiographically documented brainmetastases undergoing whole brain radiotherapy at theVancouver Cancer Centre were eligible [18]. In addition tothe MoCA and MMSE, patients completed a five-point scalereporting how inconvenient the assessments were. Patientcharacteristics collected were age, diagnosis, education,medications, prior treatment, and number of metastases.

In the second study, the diagnostic accuracy of the CSTswas compared with a gold standard neuropsychologicalassessment (NPA) [19]. English-speaking patients from theVancouver Cancer Centre, Fraser Valley Cancer Centre, orthe Centre for the Southern Interior at the BC Cancer Agency,with either brain metastases or primary brain tumors, wereeligible. Patients with disabilities that prohibited questionnairecompletion or have a Karnofsky Performance Status (KPS) ofless than 70 were excluded. While both patients with brainmetastases or malignant gliomas were accrued, for thepurpose of this analysis, 25 patients with brain metastaseswere included. In addition, to the MMSE, MoCA, and NPA,self-reported measures of quality of life and communityintegration were collected [19]. As a result of the moreextensive assessments, patients in this “diagnostic” studywere younger, more educated, and achieved higher meanCST scores than patients in the “feasibility” study, asdiscussed elsewhere [10]. Patient characteristics collectedwere age, diagnosis, education, occupation, employmentstatus, medications, illicit drug use, alcohol use, priortreatment, number of metastases, and KPS.

Materials and procedure

The MMSE is a widely recognized and utilized CST andtherefore is not described in detail here. Both the MMSEand MoCA are CSTs that are administered by a health careprovider and require respondents to answer questions, readinstructions, and perform tasks with a writing instrument.Both are scored out of a possible 30 points, with lowerscores indicating more cognitive impairment. The MoCA,which is a relatively new CST, was developed in order toscreen for mild cognitive impairment, an intermediatediagnosis between normal cognitive decline from agingand frank dementia [20]. The MoCA is a one-pagedocument, freely available on the internet, and measureseight cognitive domains: visuospatial/executive, naming,memory, attention, language, abstraction, delayed recall,and orientation. [11, 17]

1850 Support Care Cancer (2011) 19:1849–1855

The MMSE and MoCA were administered to all of thestudy participants. In the feasibility study, the MMSE andMoCA were administered in random order within a 3-dayperiod, by different investigators (RO and TC) who wereblinded to the results of the other CST [18]. In the diagnosticstudy, the MMSE and MoCA were administered within a 5-day period by different investigators (RO, HC, and MP) whowere blinded to the results of the other CST and the NPA; theorder of administration alternated for each successive studyparticipant [19]. Patient characteristics (described above)were collected at time of study accrual, for both studies.

Statistical analysis

Patient characteristics and CST scores were presented usingdescriptive statistics. Correlation between the CST scoreswas assessed with Spearman’s rank correlation coefficient.CST scores were dichotomized in order to create compar-isons amenable to Kaplan–Meier curve presentation and foreasy clinical use. Since the MoCA is new to oncology andhas never been assessed as a prognostic marker, we had noa priori knowledge of potentially clinically meaningfulcutoffs; therefore, we choose the median score, because itwas free from bias, and provided two groups of similar sizefor comparison purposes.

Overall survival (OS) was measured from the date of studyaccrual until death from any cause. Patients who were alive atlast follow-up were censored. Time-to-event distributionswere computed by use of Kaplan–Meier estimates and werecompared by use of two-sided log-rank tests. Multivariableanalyses of the prognostic variables for OS were performedusing the Cox proportional hazards model. Since we foundevidence for colinearity between the MoCA and MMSE, theywere not entered into the same multivariable model. There-fore, two separate models were created and compared usingthe Akaike information criterion (AIC) from each model, witha difference greater than 2.0 considered strong evidence forsuperiority of one model over another [21, 22]. Theproportional hazards assumptions were tested for eachvariable in the final model by introducing a time-varyinginteraction term, consisting of the product of that variablewith time. A p value ≤0.05 for that interaction term wouldresult in a rejection of the null hypothesis that the hazardswere proportional. Variables were chosen if they (1) werepreviously identified as prognostic in this population [2, 23–25], (2) were previously identified as potential confounders[26], or (3) were associated with a p value <0.1 on univariateanalysis. All patients had complete information on allcovariates used, allowing comparison of multivariablemodels with the AIC. Analyses were conducted usingStatistical Package for Social Sciences, version 17.0 (SPSS,Chicago, IL, USA) and SAS for Windows, version 9.2(Cary, NC, USA).

Results

Patient characteristics

The mean age of patients was 59.0 years (range 27–85;standard deviation=10.9). At the time of analysis, 50(76.9%) of the study subjects were deceased. The mediansurvival of the study group was 26.0 weeks (95%confidence interval [CI] 12.0–40.0 weeks). Additionalpatient characteristics are shown in Table 1.

MoCA and MMSE scores

The correlation between the MoCA and the MMSE wasstrong (r=0.75; p<0.001). As displayed in Fig. 1a, theMoCA scores are roughly normally distributed, while theMMSE is non-normally distributed (skewness=−2.2; kur-tosis=5.5; Fig. 1b). In fact, 16 (24.6%) of the subjects

Table 1 Additional patient characteristics

Characteristics Samples Percentage

Age

<50 10 15.4

50–64 35 53.8

>64 20 30.8

Gender

Male 33 50.8

Female 32 49.2

Education

Secondary or less 14 21.5

Post secondary 42 64.6

Graduate 9 13.8

Diagnosis

Lung 29 44.6

Breast 9 13.8

Gastrointestinal 8 12.3

Melanoma 6 9.2

Other 13 20.0

No. of brain metastases

1 29 44.6

2 16 24.6

≥3 20 30.8

Prior treatmenta

Surgical excision 24 36.9

Radiation 9 13.8

Current medication

Anti-epileptic 3 4.6

Corticosteroid 45 69.2

Opioid 16 24.6

a For brain metastases

Support Care Cancer (2011) 19:1849–1855 1851

obtained a perfect MMSE score (Fig. 1b). Therefore, CSTscores were dichotomized using the median score.

Univariate analysis

As shown in Fig. 2, the median OS was worse forindividuals with below-average MMSE scores (10.4 versus36.3 weeks, p=0.007). Likewise, below-average MoCAscores were predictive for worse median OS (6.3 versus50.0 weeks, p<0.001; Fig. 3). Furthermore, the median OSfor MoCA scores <22, 22–26, and >26 were 6.3, 30.9, and61.7 weeks, respectively (p<0.001; Fig. 4).

In this group of patients, no other collected variableswere significantly associated with OS on univariateanalysis. Median OS for individuals ≥65 years old were10.7 weeks, and 32.4 weeks for those less than 65 (p=0.156). Median OS for lung, breast, and all other patients

was 17.7, 34.3, and 36.3 weeks, respectively (p=0.085).Subjects using corticosteroids at time of study participationhad a median OS of 21.3 weeks, compared with 44.7 weeksfor those who were not (p=0.069). The median OS for maleand female subjects was 19.6 and 30.7 weeks, respectively(p=0.365). Individuals with post-secondary education had amedian OS of 34.3 weeks, compared with 17.1 weeks forthose who did not (p=0.106). The median OS forindividuals with single compared with multiple brainmetastases was 34.3 versus 21.3 weeks (p=0.230).

Multivariable analyses

Since there was evidence of colinearity between the MoCAand MMSE, two separate multivariable models containingthe same covariates were created and subsequently com-pared. Below-average MMSE scores were no longer

Fig. 1 Histograms displaying the proportion of patients with each corresponding a MoCA score and b MMSE score

Fig. 2 Overall survival among subjects with MMSE scores <28and ≥28

Fig. 3 Overall survival among subjects with MoCA scores <22and ≥22

1852 Support Care Cancer (2011) 19:1849–1855

prognostic after controlling for age, corticosteroid use,gender, primary site, number of metastases, and educa-tion (hazard ratio [HR]=1.71; 95% CI=0.90–3.26; p=0.104). In contrast, below-average MoCA scores wereprognostic, after controlling for the same variables (HR=5.44; 95% CI=2.70–10.94; p<0.001). In order to elimi-nate the potential impact of dichotomization cut-off choiceon the outcome, we also performed the multivariableanalyses with the continuous CST scores and found thatthe MoCA (HR=0.87; 95% CI=0.82–0.92; p<0.001;Table 2) and MMSE (HR=0.82; 95% CI=0.74–0.91; p<0.001) were prognostic. The AIC of the model containingthe MoCA was 6.35 points lower than the modelcontaining the MMSE, which demonstrates strong superi-ority of the MoCA as a prognostic marker.

Discussion

The MoCA is a relatively new screening tool, which is moresensitive and better correlated with quality of life than theMMSE [19]. In this study, we demonstrated that the MoCAis the strongest predictor of overall survival of the variablescollected. Not only is it highly statistically significant (p<

0.001) but the magnitude of effect is also highly clinicallysignificant. Further research is needed in order to definewhich cut-off scores are appropriate, both for prognosticationand in screening for cognitive impairment.

Of note, patients with below-median MoCA scores had aparticularly poor prognosis, with a median OS of only6.3 weeks (Figs. 2 and 3). As previously hypothesized byBezjak et al., many patients with brain metastases may notbenefit from even short courses of palliative radiotherapy,and they therefore encouraged further research into optimi-zation of patient selection [27]. We propose the MoCA is agood candidate for future investigation as a predictivemarker; specifically, low scores may help identify individ-uals less likely to benefit from palliative whole brainradiotherapy. Conversely, high MoCA scores may helpidentify patients who are most likely to benefit from moreintensive interventions such as radiotherapy dose escala-tion, stereotactic radiosurgery, and surgical excision. There-fore, we encourage further research into this potential rolefor the MoCA in combination with other prognosticvariables.

We hypothesize that the MoCA has superior prognosticutility over the MMSE, because unlike the MMSE, it doesnot suffer from a ceiling test effect in this population andtherefore can better separate patients into different catego-ries of cognitive function (Fig. 1). This is likely secondaryto MoCA’s more extensive testing of executive functioning,delayed recall, and abstraction. In contrast, the vastmajority of patients have normal MMSE scores in boththis study (Fig. 1) and in larger clinical trials using theMMSE [2, 15, 28–30].

The MMSE ceiling effect is also seen in clinical practice,where most subjects have normal MMSE scores, and whichunfortunately is unable to accurately predict for normalcognitive function [13, 19]. In contrast, the relatively fewindividuals who have grossly abnormal MMSE scores oftenalso have clinically poor performance status, and therefore,it is unclear if the MMSE adds much to a clinician’sbedside assessment of cognitive function and prognosis. Animportant exception is in patients with a very poorprognosis, such as those admitted to palliative care wards;in this population, patients’ MMSE scores are much lower

Fig. 4 Overall survival among subjects with MoCA scores <22, 22–26, and >26

Variable Hazard ratio (95% CI) p value

Age (continuous) 1.01 (0.97–1.04) 0.81

Corticosteroid use at accrual (versus non-use) 1.27 (0.58–2.79) 0.56

Lung cancer primary (versus other) 1.50 (0.75–2.99) 0.25

Male gender (versus female) 1.07 (0.48–2.09) 0.86

MoCA (continuous increments) 0.87 (0.82–0.92) <0.001

Post Secondary Education (versus nonea) 1.00 (0.48–2.09) 0.99

Solitary brain metastases (versus multiple) 0.79 (0.41–1.54) 0.49

Table 2 Multivariable survivalanalysis of the MontrealCognitive Assessment (MoCA)

a The MoCA adjust for education(adds one point if education≤12 years)

Support Care Cancer (2011) 19:1849–1855 1853

and therefore unlikely prone to ceiling effects [16]. Thishighlights the importance of the population being studiedon interpretation of study results [10].

Although this study demonstrates a strong relationshipbetween the MoCA and survival, limitations in this dataprohibit to conclude that cognitive impairment directlyresults in decreased survival. More likely, the MoCA couldbe correlated with an unmeasured confounder or combina-tion of measured and unmeasured variables. For example,the MoCA may be a surrogate for a composite of acutedelirium, performance status, extent of disease, location oftumor, previous treatment, age, and comorbidities, all ofwhich are prognostic and potentially result in lower MoCAscores. Whether cognitive impairment is itself a cause ofworse prognosis or instead a surrogate measure of otherconfounders, our data suggest that the MoCA is a usefulprognostic index for both clinical practice and trials.

As discussed elsewhere, the introduction of a NPA to thesecond diagnostic study resulted in a statistically significantselection bias in these patients, preferentially accruingyoung, well-educated patients [10]. In contrast, the feasi-bility study had broad eligibility criteria, with excellentparticipation rate [10, 18]. Since the majority of the subjectsin this study were accrued to the feasibility study, webelieve that these results can be widely generalized to mostbrain metastases patients [10, 18, 19]. Furthermore, thepatient characteristics are typical of brain metastasespatients seen clinically and in most trials (Table 1).However, generalizing these results to extreme populationsof brain metastases, such as patients very near the end oflife, is not recommended without further validation in thesesettings.

The relatively small sample size of this study limits theinterpretation of the non-significant correlation between theother variables and survival. Likewise, other variables inthe multivariable model are likely prognostic, though thisstudy did not have the power to detect their relationshipwith survival. Despite these limitations, the associationbetween the MoCA and prognosis is highly significant,with a large clinically meaningful magnitude of effect(Table 2).

The major limitation of this study is the omission of theKPS, which is a previously identified prognostic variable,and is incorporated into the Radiation Therapy OncologyGroup (RTOG) recursive portioning analysis (RPA) prog-nostic index for brain metastases and malignant gliomas, aswell as the new graded prognostic assessments (GPA) [23,31–33]. Unfortunately, the KPS was not prospectivelycollected in our feasibility study, and KPS<70 wasexcluded from our second diagnostic study and thereforecould not be adequately assessed [18]. This raises thepossibility that the MoCA is a surrogate measure ofperformance status, as discussed above. Although the

MoCA is more time consuming to administer than collect-ing the KPS, it is also potentially more objective, with awider range of potential scores, which may be useful inclinical trials. It also provides a measure of cognitivefunction that can be used as a baseline to which post-treatment effects may be compared. Therefore, we believefurther research should assess if the MoCA adds to theRTOG’s RPA and GPA.

Conclusion

In this study, the MoCA is a superior prognostic indicatorover the MMSE. Given its ease of use, free availability,superior sensitivity, and better correlation with quality oflife, it should be preferentially chosen in both clinicalpractice and trials. However, we also encourage validationby independent investigators and in different populations ofbrain tumor patients. Furthermore, future research shouldaddress if the MoCA adds to the RTOG’s RPA and GPA.Lastly, we propose that future research investigate whetherthe MoCA can help select patients for different treatmentapproaches.

Acknowledgements Grant funding was provided by the FraserValley Interdisciplinary Research Grant and the Hershey & YvettePorte Neuro-Oncology Fund. The authors thank Dr. Brian Brooks andProf. Grant Iverson from BC Mental Health & Addictions Services fortheir input into the diagnostic study design, as well as Prof. RogerDavis from the Harvard School of Public Health for his valuable inputinto the multivariable analyses.

Conflict of interest statement No potential conflict of interestexists for any of the authors.

References

1. Klein M, Taphoorn MJB, Heimans JJ, van der Ploeg HM,Vandertop WP, Smit EF, Leenstra S, Tulleken CAF, Boogerd W,Belderbos JSA, Cleijne W, Aaronson NK (2001) Neurobehavioralstatus and health-related quality of life in newly diagnosed high-grade glioma patients. J Clin Oncol 19:4037–4047

2. Murray KJ, Scott C, Zachariah B, Michalski JM, Demas W, VoraNL, Whitton A, Movsas B (2000) Importance of the mini-mentalstatus examination in the treatment of patients with brainmetastases: a report from the radiation therapy oncology groupprotocol 91-04. Int J Radiat Oncol Biol Phys 48:59–64

3. Weitzner MA, Meyers CA (1997) Cognitive functioning andquality of life in malignant glioma patients: a review of theliterature. Psychooncology 6:169–177

4. Meyers CA, Hess KR (2003) Multifaceted end points in braintumor clinical trials: cognitive deterioration precedes MRIprogression. Neuro-oncology 5:89–95

5. Meyers CA, Smith JA, Bezjak A, Mehta MP, Liebmann J, IllidgeT, Kunkler I, Caudrelier J, Eisenberg PD, Meerwaldt J, Siemers R,Carrie C, Gaspar LE, Curran W, Phan S, Miller RA, RenschlerMF (2004) Neurocognitive function and progression in patientswith brain metastases treated with whole-brain radiation and

1854 Support Care Cancer (2011) 19:1849–1855

motexafin gadolinium: results of a randomized phase III trial. JClin Oncol 22:157–165

6. Meyers CA, Brown PD (2006) Role and relevance of neuro-cognitive assessment in clinical trials of patients with cns tumors.J Clin Oncol 24:1305–1309

7. Hegi ME, Diserens A, Gorlia T, Hamou M, de Tribolet N,Weller M, Kros JM, Hainfellner JA, Mason W, Mariani L,Bromberg JEC, Hau P, Mirimanoff RO, Cairncross JG, JanzerRC, Stupp R (2005) MGMT gene silencing and benefit fromtemozolomide in glioblastoma. N Engl J Med 352:997–1003.doi:10.1056/NEJMoa043331

8. Meyers CA, Hess KR, Yung WKA, Levin VA (2000) Cognitivefunction as a predictor of survival in patients with recurrentmalignant glioma. J Clin Oncol 18:646–650

9. Chang EJ, Wefel JS, Maor MH, Hassenbush SJ III, Mahajan A,Lang FF, Woo SY, Mathews LA, Allen PK, Shiu AS, Meyers CA(2007) A pilot study of neurocognitive function in patients withone to three new brain metastases initially treated with stereotacticradiosurgery alone. Neurosurg 60:277–284

10. Olson R, Parkinson M, McKenzie M (2010) Selection biasintroduced by neuropsychological assessments. Can J Neurol Sci37:264–268

11. Folstein MF, Folstein SE, McHugh PR (1975) “Mini-mentalstate”: a practical method for grading the cognitive state ofpatients for the clinician. J Psych Res 12:189–198

12. Brown PD, Buckner JC, O’Fallon JR, Iturria NL, Brown CA,O’Neill BP, Scheithauer BW, Dinapoli RP, Arusell RM, CurranWJ, Abrams R, Shaw EG (2003) Effects of radiotherapy oncognitive function in patients with low-grade glioma measured bythe folstein mini-mental state examination. J Clin Oncol 21:2519–2524

13. Meyers CA, Wefel JS (2003) The use of the mini-mental stateexamination to assess cognitive functioning in cancer trials: no ifs,ands, buts, or sensitivity. J Clin Oncol 21:3557–3558

14. Olson R, Ellwood A, Parkinson M, Carolan H, Ferguson K,Iverson G, McKenzie M (2009) Canadian Association ofPsychosocial Oncology Annual Conference, Vancouver, BritishColumbia. April 1-4, 2009. Abstract ID: B246

15. Shibamoto Y, Baba F, Oda K, Hayashi S, Kokubo M, Ishihara S,Itoh Y, Ogino H, Koizumi M (2008) Incidence of brain atrophyand decline in mini-mental state examination score after whole-brain radiotherapy in patients with brain metastases: a prospectivestudy. Int J Radiat Oncol Biol Phys 72:1168–1173. doi:10.1016/j.ijrobp.2008.02.054

16. Pereira J, Hanson J, Bruera E (1997) The frequency and clinicalcourse of cognitive impairment in patients with terminal cancer.Cancer 79:835–842

17. Nasreddine ZS, Phillips NA, Bedirian V, Charbonneau S,Whitehead V, Collin I, Cummings JL, Chertkow H (2005)The montreal cognitive assessment, MoCA: a brief screeningtool for mild cognitive impairment. J Am Geriat Soc 53:695–699

18. Olson RA, Chhanabhai T, McKenzie M (2008) Feasibility studyof the Montreal Cognitive Assessment (MoCA) in patients withbrain metastases. Support Care Cancer 16:1273–1278.doi:10.1007/s00520-008-0431-3

19. Olson RA, Iverson G, Parkinson M, Carolan H, Ellwood A,McKenzie M (2009) Investigation of cognitive screening meas-ures in patients with brain tumors: diagnostic accuracy andcorrelation with quality of life. J Clin Oncol (Meeting Abstracts)27:e13000

20. Gauthier S, Reisberg B, Zaudig M, Petersen RC, Ritchie K,Broich K, Belleville S, Brodaty H, Bennett D, Chertkow H,Cummings JL, de Leon M, Feldman H, Ganguli M, Hampel H,Scheltens P, Tierney MC, Whitehouse P, Winblad B (2006) Mildcognitive impairment. Lancet 367:1262–1270

21. Akaike H (1974) A new look at the statistical model identifica-tion. IEEE Trans Automat Contr 19:716–723

22. Posada D, Buckley TR (2004) Model selection and modelaveraging in phylogenetics: advantages of Akaike informationcriterion and Bayesian approaches over likelihood ratio tests. SystBiol 53:793–808. doi:10.1080/10635150490522304

23. Gaspar L, Scott C, Rotman M, Asbell S, Phillips T, Wasserman T,McKenna WG, Byhardt R (1997) Recursive partitioning analysis(RPA) of prognostic factors in three radiation therapy oncologygroup (RTOG) brain metastases trials. Int J Radiat Oncol BiolPhys 37:745–751. doi:10.1016/S0360-3016(96)00619-0

24. Sherman AM, Jaeckle K, Meyers CA (2002) Pretreatmentcognitive performance predicts survival in patients with leptome-ningeal disease. Cancer 95:1311–1316

25. Patchell RA, Tibbs PA, Regine WF, Dempsey RJ, Mohiuddin M,Kryscio RJ, Markesbery WR, Foon KA, Young B (1998)Postoperative radiotherapy in the treatment of single metastasesto the brain: a randomized trial. JAMA 280:1485–1489.doi:10.1001/jama.280.17.1485

26. Liu IY, Lacroix AZ, White LR, Kittner SJ, Wolf PA (1990)Cognitive impairment and mortality: a study of possible con-founders. Am J Epidemiol 132:136–143

27. Bezjak A, Adam J, Barton R, Panzarella T, Laperriere N,Wong CS, Mason W, Buckley C, Levin W, McLean M, Wu JS,Sia M, Kirkbride P (2002) Symptom response after palliativeradiotherapy for patients with brain metastases. Eur J Cancer38:487–496

28. Andrews DW, Scott CB, Sperduto PW, Flanders AE, Gaspar LE,Schell MC, Werner-Wasik M, Demas W, Ryu J, Bahary J,Souhami L, Rotman M, Mehta MP, Curran J, Walter J (2004)Whole brain radiation therapy with or without stereotactic radio-surgery boost for patients with one to three brain metastases:phase III results of the RTOG 9508 randomised trial. Lancet363:1665. doi:10.1016/S0140-6736(04)16250-8

29. Regine WF, Scott C, Murray K, Curran W (2001) Neurocognitiveoutcome in brain metastases patients treated with accelerated-fractionation vs. accelerated-hyperfractionated radiotherapy: ananalysis from Radiation Therapy Oncology Group Study 91-04.Int J Radiat Oncol Biol Phys 51:711–717

30. Stupp R, Mason WP, van den Bent MJ, Weller M, Fisher B,Taphoorn MJB, Belanger K, Brandes AA, Marosi C, Bogdahn U,Curschmann J, Janzer RC, Ludwin SK, Gorlia T, Allgeier A,Lacombe D, Cairncross JG, Eisenhauer E, Mirimanoff RO, theEuropean Organisation for Research and Treatment of CancerBrain Tumor and Radiotherapy Groups and the National CancerInstitute of Canada Clinical Trials Group (2005) Radiotherapyplus concomitant and adjuvant temozolomide for glioblastoma. NEngl J Med 352:987–96. doi:10.1056/NEJMoa043330

31. Curran WJ Jr, Scott CB, Horton J, Nelson JS, Weinstein AS,Fischbach AJ, Chang CH, Rotman M, Asbell SO, Krisch RE,Nelson DF (1993) Recursive partitioning analysis of prognosticfactors in three radiation therapy oncology group malignantglioma trials. J Natl Cancer Inst 85:704–710. doi:10.1093/jnci/85.9.704

32. Sperduto PW, Berkey B, Gaspar LE, Mehta M, Curran W (2008)A new prognostic index and comparison to three other indices forpatients with brain metastases: an analysis of 1, 960 patients in theRTOG database. Int J Radiat Oncol Biol Phys 70:510–514.doi:10.1016/j.ijrobp.2007.06.074

33. Sperduto PW, Chao ST, Sneed PK, Luo X, Suh J, Roberge D, BhattA, Jensen AW, Brown PD, Shih H, Kirkpatrick J, Schwer A, GasparLE, Fiveash JB, Chiang V, Knisely J, Sperduto CM,MehtaM (2010)Diagnosis-specific prognostic factors, indexes, and treatment out-comes for patients with newly diagnosed brain metastases: a multi-institutional analysis of 4, 259 patients. Int J Radiat Oncol Biol Phys77:655–661. doi:10.1016/j.ijrobp.2009.08.025

Support Care Cancer (2011) 19:1849–1855 1855