Screening for Cognitive Impairment in Older Adults with Hematological Malignancies Using the Montreal Cognitive Assessment and Neuropsychological Testing

Abstract

OBJECTIVES

The primary objective of the current study is to describe the prevalence and profile of cognitive domains affected in older adults with hematological malignancies evaluated for hematopoietic cell transplantation (HCT) using the Montreal Cognitive Assessment (MoCA) and neuropsychological tests. The secondary objective is to determine if a specific MoCA cut-off score would correlate with the identification of cognitive impairment detected by neuropsychological tests. This would facilitate interpretation of cognitive screening and referral of patients who would likely need further neuropsychological testing.

MATERIALS AND METHODS

Fifty-one patients 60 years and older who were evaluated for HCT were assessed using a battery of standardized neuropsychological tests and MoCA. We analyzed Receiver Operating Characteristics (ROC) comparing MoCA scores and four different neuropsychological test criteria for cognitive impairment.

RESULTS

The prevalence of cognitive impairment detected by neuropsychological tests was 53 to 70.6% using the criteria for patients with cancer by the International Cancer Cognition Task Force (ICCTF). The following cognitive domains were most affected: language, learning and memory, visuospatial skills, and executive function. MoCA is an appropriate screening test for cognitive impairment. Using the ICCTF criteria, 86 to 100% of patients are correctly classified as having significant cognitive impairment on neuropsychological tests using a cut-off score of 20 or less.

CONCLUSION

There is a high prevalence of cognitive impairment identified by neuropsychological tests in older patients with hematological malignancies evaluated for HCT. Identification of an appropriate MoCA cut-off score in this population is important to identify patients who would benefit from further assessment.

INTRODUCTION

The majority of hematologic malignancies are diagnosed at a median age of 65 years.¹ Hematopoietic cell transplantation (HCT) is a life-prolonging treatment for patients with hematologic malignancies. Several advances have led to more older patients receiving HCT, including better risk stratification, reduced intensity conditioning regimens, and improved supportive care. Cognitive adverse effects are risks of HCT even among younger patients, can occur early or late after HCT and are a major cause of morbidity.²

Screening for cognitive impairment is not routine in oncology and cognitive deficits are likely under recognized. Literature on cognitive function in older adults with hematological malignancy suggests that impaired working memory prior to treatment predicts survival in patients receiving intensive chemotherapy.³ Recent studies in younger HCT recipients suggest that increasing age is associated with worse cognitive function and a slower recovery.^4,5 Cognitive impairment before HCT increases risk for delirium and older age predicts the severity of delirium.⁶ This is significant because delirium is a risk factor for slower cognitive recovery, persistent cognitive decline, functional decline, increased number of hospitalizations and time spent in the hospital in HCT recipients⁷ and hospitalized older adults.^8–10

Administration of a complete battery of neuropsychological tests for every patient is not feasible due to time and staff training in a busy clinic and the limited availability of neuropsychologists. Hence, the use of a screening test for cognitive impairment to identify patients who may benefit from a more in-depth evaluation of their cognitive function would be of significant clinical relevance. Available screening tools used in geriatrics and geriatric oncology include the Mini-Mental State Examination (MMSE), the Short-Blessed Test, and the Montreal Cognitive Assessment (MoCA). The MoCA stands out due to its high sensitivity and specificity for identifying mild cognitive impairment (MCI),¹¹ published normative data and cut-off scores, alternate forms, and global assessment of major cognitive domains. A cut-off score of less than 26 has superior sensitivity and less ceiling effect compared to MMSE for detecting early dementia and MCI in the initial validation studies.^11–13 However, further studies suggest that caution should be used in application of the established cut-off score in different populations.¹⁴ The relevance of the universal MoCA cut-off of <26 for screening of cognitive deficits in older adults with hematological malignancies has not been reported.

The primary objective of the current study is to describe the prevalence and profile of cognitive domains affected in older adults with hematological malignancies being evaluated for HCT using neuropsychological tests and the MoCA. The secondary objective is to identify a MoCA cut-off score for cognitive impairment identified by neuropsychological tests. Finally, to demonstrate the relationship between MoCA cut-off score for cognitive impairment determined by four different definitions including definitions for patients with cancer recommended by the International Cognition and Cancer Task Force (ICCTF).¹⁵ Previous literature in HCT have used varying definitions of cognitive impairment so inclusion of multiple definitions in our work may help provide meaningful comparison. We hypothesize that older adults with hematological malignancies will have a high prevalence of cognitive impairment because of previous chemotherapy and increasing prevalence of cognitive impairment unrelated to cancer and cancer treatment.

MATERIALS AND METHODS

Study Population

The study was conducted in the context of pretransplant evaluation for patients 60 years and older undergoing evaluation for HCT. Patients were referred to our program by their oncologist to optimize health and function. There is a total of fifty-one patients included in the analysis of data. These patients have completed the MoCA, a battery of neuropsychological tests and geriatric assessment from August 2016 to December 2018. Twenty-four patients completed the neuropsychological tests as a part of our program’s routine clinic assessment. Due to logistic reasons (scheduling, limited visit time), the neuropsychological tests were later removed from the clinic assessment. The remaining twenty-seven patients completed the neuropsychological tests after this change as part of a research study examining changes in cognitive function in patients 60 years and older before and after HCT.

Cognitive Testing

MoCA consists of thirteen tasks measuring the following cognitive domains: visuospatial, executive functions, memory, attention, language, abstraction, and orientation. A total score is calculated by adding scores of the thirteen tasks. The maximum possible score is 30 points. The total score is corrected for educational level. The secondary objective of the study is to define an optimal MoCA cut off score to detect impairment on a brief battery of neuropsychological tests, which is the “gold standard” for this study. The neuropsychological tests used were divided into four categories: language (Phonemic Verbal Fluency, Boston Naming Test), learning and memory (Word List Learning, Delayed Recall and Recognition), visuospatial skills (Praxis 2D and 3D), and executive functioning (Trail Making Test Part A and B). A description of all of the instruments discussed in this section can be found in greater detail in Appendix A, Table 1. Cognitive domains were chosen based on previous studies of domains affected in HCT patients¹⁶ and relevance to patients with cancer according to the International Cognition and Cancer Task Force (ICCTF)¹⁵ and older patients.¹⁷ The neuropsychological tests were administered by a trained research assistant and took on average 40 minutes to complete.

Geriatric Assessment (GA) Measures

GA measures were used to describe the clinical characteristics of the sample. Activities of daily living¹⁸ (Katz ADLs), instrumental activities of daily living¹⁹ (Lawton IADLs), functional mobility²⁰ (Short Physical Performance Battery [SPPB]), depression²¹ (Geriatric Depression Scale-15 [GDS-15]), anxiety²² (Generalized Anxiety Disorder-7 [GAD-7]), malnutrition²³ (Mini Nutritional Assessment-Short Form [MNA-SF]), and perceived social support²⁴ (Medical Outcomes Social Support Survey [MOS-SS]) were assessed with validated measures. A description of GA measures discussed in this section can be found in greater detail in Appendix A, Table 2.

Criteria for Cognitive Impairment

Participants were classified as having cognitive impairment based on cognitive testing using four definitions that vary based on standard deviation (SD) below the normative mean: Criteria 1) z-scores at or below −1.5 SD in a single domain (a less conservative criterion); Criteria 2) z-score at or below −1.5 SD in two or more domains; Criteria 3) z-score at or below −2.0 SD in a single domain; or Criteria 4) z-scores at or below −1.5 SD in two or more domains or a single domain with z-score at or below −2.0 SD. Criteria 1 was chosen based on the criteria outlined by Peterson and Morris (2005) to define objective cognitive impairment for mild cognitive impairment in older adults.²⁵ Criterion 2–4 were based on the definition outlined by the ICCTF.¹⁵

Data Recorded

The following data were recorded from medical records when available: age; gender; education; working status (e.g. retired, employed); cancer type; comorbidities including hypertension, diabetes, stroke, and neurological disorders; history of head injury; current or past history of substance use disorder; performance status by Karnofsky Performance Scale; falls; and transplant status.

Statistical Analysis

Raw scores for each neuropsychological test were converted to standardized scores (z-scores) using published normative data adjusted for age, education, and gender when appropriate. The mean, SD, and median values for geriatric assessment and each neuropsychological test were calculated. A t-test was used to compare mean results for each neuropsychological test for the sample that completed the tests as routine clinical care and the sample that completed the tests as a part of the research protocol (Supplementary data). We performed Receiver Operating Characteristics (ROC) analyses to determine the MoCA cut-off score using the Youden Index and separately identifying a point on the curve that maximized both sensitivity and specificity for the detection of cognitive impairment on neuropsychological tests. Often the two methods produced the same optimal cut-off, however in situations when they are different, the Youden Index was used. ²⁶ Sensitivity and specificity were computed with a logistic regression analysis using PROC LOGISTIC from SAS/STAT software, Version 9.4. Area under the curve (AUC) was used to summarize the ROC analysis. Fisher’s exact test was used to compare geriatric assessment measures for patients who have cognitive impairment on neuropsychological tests and those without impairment. For this analysis, we used the definition of z-score at or below −1.5 SD in two or more cognitive domains.

Ethical Considerations

This study was approved by the Institutional Review Board of the University of Nebraska Medical Center. All patients’ records were de-identified to protect the confidentiality of the data.

RESULTS

Participants Demographic and Clinical Characteristics (Table 1)

Table 1.

Demographic and clinical characteristics of patients who completed MoCA and neuropsychological tests.

	Mean (SD)
Age (n=51)	68.31 (4.31)
Karnofsky Performance Status (KPS) >80* (n=44)	81.40 (7.95)
Characteristics	n (%)
Gender (n=51)
Male	34 (66.7%)
Female	17 (33.3%)
Race and ethnicity (n=51)
Caucasian and non-Hispanic	49 (96.1%)
Marital status (n=51)
Married/Life Partner	43 (84.3%)
Single/Never married	4 (7.8%)
Widowed	3 (5.9%)
Divorced/Separated	1 (2.0%)
Work status (n=49)
Employed	9 (18.4%)
Unemployed	1 (2.0%)
Retired	32 (65.3%)
Sick leave/Disability	7 (14.3%)
Education (n=51)
<High school	1 (2.0%)
High school graduate	16 (31.4%)
Some college	10 (19.6%)
Bachelor’s degree	11 (21.6%)
Graduate degree	13 (25.5%)
Cancer Type (n=51)
Leukemia	24 (47.1%)
Lymphoma	6 (11.8%)
Multiple Myeloma	8 (15.7%)
Myelodysplastic Syndrome	13 (25.5%)
Diagnosed Comorbid Conditions (n=51)
Hypertension	29 (56.9%)
Diabetes	12 (23.5%)
Multiple Sclerosis	1 (2.0%)
History of stroke	1 (2.0%)
Dementia♦	1 (2.0%)
Current or past substance use disorder related to alcohol or illicit substances	2 (3.9%)
History of major or moderate head injury (loss of consciousness >60 minutes or structural brain changes on imaging	1 (2.0%)
Uncorrected vision impairment	2 (3.9%)
Transplant (n=51)
Yes (%)	41 (80.4%)
Allogeneic	29 (70.7%)
Autologous	12 (29.3%)

^*KPS >80%: normal activity and no evidence of disease, mild or some signs or symptom

^♦Patient with clinical diagnosis of dementia by neuropsychology and geriatrician.

Fifty-one participants with hematological malignancies were included in the study. The mean age was 68 years. Most participants were males, Caucasian and non-Hispanic, married or had a life partner, and had received at least some post-secondary education. Other health conditions known to affect cognitive functioning were also recorded. Over 80% of participants proceeded to HCT with most receiving allogeneic HCT (70.7%).

Chemotherapy Treatment

All patients had received chemotherapy by the time of cognitive testing. Patients diagnosed with multiple myeloma received either two-drug (lenalidomide/dexamethasone) or three-drug regimens (bortezomib/lenalidomide/dexamethasone, daratumumab/lenalidomide/dexamethasone, or carfilzomib/lenalidomide/dexamethasone). With regards to leukemia/myelodysplasia diagnosis, all patients received induction chemotherapy, and eleven patients had also received consolidation chemotherapy. The induction chemotherapy regimens included: 7+3 therapy (cytarabine and daunorubicin); 7+3 therapy with an additional agent such as gemtuzumab, midostaurin; fludarabine, cytarabine and granulocyte colony-stimulating factor; and single agents (decitabine, azacytidine). Non-Hodgkin lymphoma regimens included cyclophosphamide, doxorubicin, vincristine, prednisone plus rituximab (R-CHOP) and etoposide, doxorubicin, vincristine, cyclophosphamide, and prednisone plus rituximab (EPOCH-R). Acute lymphoblastic leukemia/lymphoma regimens included Hyper CVAD (cycle A: cyclophosphamide, vincristine, doxorubicin, and dexamethasone and cycle B: methotrexate and cytarabine).

Prevalence of Cognitive Impairment

Sample mean z-score and SD for each neuropsychological test are presented in Table 2. The mean z-score for the language, learning and memory, visual spatial skills, and executive functioning domains pre-transplant are lower than expected, with visual spatial skills showing the lowest mean z-score. To assess impairment in each test, the number and proportions of participants who scored ≤ −1.5 SD and ≤ −2.0 SD below the published normative data were recorded. Participants were most impaired in the following tests: Praxis 2D & 3D Figures (visual spatial skills, 45.1%), Word List Learning (learning and memory, 34.7%), Trail Making Test Part A (executive functioning, 32.0%). Table 1S (Supplementary Data) presents a comparison of mean z-score (SD) for the neuropsychological test domains in routine clinical care and research protocol patients. There are no significant differences between the two groups except in the domain of verbal fluency. This is due to one patient with clinical diagnosis of dementia who was seen in the clinic.

Table 2:

Mean Z-score (SD) and percentage of participants impaired in each neuropsychological test domains

Neuropsychological Tests	N	Sample Mean Z-scores (SD)	Impaired (≤ −1.5 SD below normative means) N (%)	Impaired (≤ −2.0 SD below normative means) N (%)
Language
Verbal Fluency (# correct)	N=51	−0.29 (1.01)	8 (15.7%)	2 (3.9%)
Modified Boston Naming Test 15 (#correct)	N=51	−0.77 (1.44)	12 (23.5%)	12 (23.5%)
Learning and Memory
Word List Learning (recall trial #3)	N=49	−0.86 (1.08)	17 (34.7%)	8 (16.3%)
Delayed Recall (#correct)	N=51	−0.66 (1.00)	12 (23.5%)	3 (5.9%)
Word List Recognition (# correct)	N=40	0.11 (0.64)	1 (2.5%)	0 (0.0%)
Visuospatial Skills
Praxis 2D & 3D Figures (copy accuracy)	N=51	−1.62 (1.80)	23 (45.1%)	21 (41.2%)
Executive Functioning
Trail Making Test Part A (time in seconds)	N=50	−1.09 (1.14)	16 (32.0%)	9 (18.0%)
Trail Making Test Part B (time in seconds)	N=46	−0.36 (0.82)	5 (10.9%)	1 (2.8%)

Determination of MoCA Cut-off Score

Figure 1 presents boxplots of raw scores, mean (SD), and median (range) of the MoCA according to the least conservative (Criteria 1) and the criteria recommended by ICCTF (Criteria 4) for cognitive impairment on neuropsychological tests. Table 3 presents the prevalence of impairment as defined by the proposed criteria, areas under the curve (AUC), sensitivity, and specificity for optimal cut-off scores for the MoCA in predicting impairment. Using Criteria 1, the prevalence of cognitive impairment was 86% compared to 53% with Criteria 2, and 70.6% with Criteria 3 and 4.

Figure 1.

A. Criteria 1: ≤−1.5 SD below normative means in at least one domain B. Criteria 4: ≤−1.5 SD in two or more domains or ≤−2.0 SD below normative means in one domain

Table 3:

Optimal cut-off score prediction of impairment measured by neuropsychological tests

Categories	Property
Criteria 1: ≤ −1.5 SD below normative means in at least one domain	Prevalence %	86%
	Optimal MoCA cut-off score	≤25
	AUC	0.745
	Sensitivity	0.795
	Specificity	0.571
Criteria 2: ≤ −1.5 SD below normative means in two or more domains	Prevalence %	53%
	Optimal MoCA cut-off score	≤20
	AUC	0.672
	Sensitivity	0.40
	Specificity	0.96
Criteria 4: ≤ −1.5 SD in two or more domains or ≤ −2.0 SD below normative means in at least one domain*	Prevalence %	70.6%
	Optimal MoCA cut-off score	≤22
	AUC	0.691
	Sensitivity	0.5
	Specificity	0.867

^*Criteria 3: ≤ −2.0 SD below normative means in at least one domain yields the same prevalence and optimal cut-off score as Criteria 4 (data not shown).

For the total MoCA score, a cut-off of 25 or less corresponded to a sensitivity equal to 0.795 and a specificity equal to 0.571 for detecting impairment of ≤ −1.5 SD below normative means in at least one cognitive domain (Criteria 1). A cut-off of 20 or less corresponded to a sensitivity of 0.40 and a specificity of 0.96 for detecting impairment of ≤ −1.5 SD below normative means in two or more cognitive domains (Criteria 2). Finally, a cut-off of 22 or less corresponded to a sensitivity of 0.5 and specificity of 0.867 for detecting ≤ −2 SD in at least one cognitive domain (Criteria 3: data not shown) and for detecting ≤ −1.5 SD in two or more or ≤ −2 SD below normative means in at least one cognitive domain (Criteria 4).

Using neuropsychological testing Criteria 4 to define cognitive impairment, a MoCA cut-off of 29 or higher to screen for normal cognitive function would result in no false negatives in those found to have significant cognitive impairment on the neuropsychological battery (100% sensitivity) and a MoCA cut-off of 20 or lower to screen for cognitive impairment would result in no false positives in those that do not have significant cognitive impairment on neuropsychological testing (100% specificity) (Criteria 4 ROC curve: data not shown).

GA Characteristics of Cognitive Impairment vs No Cognitive Impairment

Table 4 presents data comparing geriatric assessment (GA) domains in individuals with and without cognitive impairment on neuropsychological tests. Cognitive impairment is defined as ≤ −1.5 SD below normative means in two or more cognitive domains (Criteria 2). This criterion was chosen due to having similar sample sizes in each group. There were no significant differences in GA domains between patients with and without cognitive impairment.

Table 4.

Comparing geriatric assessment domains in individuals with and without cognitive impairment on neuropsychological tests

Variables n (%)	Total Sample	Cognitive Impairment (≤ −1.5 SD in two or more domains)	No Cognitive Impairment	P-value
Functional Mobility	N=44	N=23	N=21
SPPB ≤9	21 (47.7%)	11 (47.8%)	10 (47.6%)	1.00
Activities of Daily Living	N=51	N=28	N=23
Katz ADL ≤5	9 (17.6%)	6 (21.4%)	3 (13.0%)	0.20
Instrumental Activities of Daily Living	N=50	N=27	N=23
Lawton IADL ≤ 6	6 (12.0%)	3 (11.1%)	3 (13.0%)	1.00
Falls in the Last 6 Months	N=47	N=25	N=22
≥ 1 Fall	7 (14.9%)	3 (12.0%)	4 (18.2%)	0.69
Cognition	N=51	N=28	N=23
MoCA ≤25	38 (74.5%)	23 (82.1%)	15 (65.2%)	0.21
MoCA ≤22	20 (39.2%)	14 (50.0%)	6 (26.1%)	0.095
Depression	N=44	N=24	N=20
GDS-15 ≥5	3 (6.8%)	2 (8.3%)	1 (5.0%)	1.00
Anxiety	N=48	N=27	N=21
GAD-7 ≥10	3 (6.3%)	2 (7.4%)	1 (4.8%)	1.00
Nutrition	N=50	N=27	N=23
MNA-SF ≤11	35 (70.0%)	17 (62.9%)	18 (78.3%)	0.35
Social Support	N=47	N=26	N=21
MOS-SS ≤75	13 (27.7%)	6 (23.1%)	7 (33.3%)	0.52

SPPB: Short Physical Performance Battery; ADL: Activities of Daily Living; IADL: Instrumental Activities of Daily Living; MoCA: Montreal Cognitive Assessment; GDS: Geriatric Depression Scale; GAD: Generalized Anxiety Disorder; MNA-SF: Mini Nutritional Assessment-Short Form; MOS-SS: Medical Outcomes Social Support Survey

MoCA score of ≤25 is based on the cut-off identified in the original validation study.³³

MoCA score of ≤22 is based on the cut-off identified in community-dwelling older adults. ³⁵

DISCUSSION

Hematological malignancies have a median age at diagnosis of 65 years. There has been a dramatic increase in the number of older adults with hematological malignancies undergoing HCT due to improvements in supportive care and reduced toxicity. However, HCT is an intense treatment associated with many short- and long-term morbidities including cognitive and functional decline. Cognitive impairment prior to HCT predicts delirium, poor cognitive recovery and increased hospital stay. ⁷ Despite this, few studies have characterized cognitive function in older adults considering HCT. Since a growing number of older individuals are receiving HCT and are at risk for treatment-related cognitive decline, there is a pressing need to screen for cognitive impairment in older HCT candidates. To our knowledge, our study is the first to examine the relationship between performance on the MoCA with performance on a battery of neuropsychological tests in this population.

The results of this study demonstrate that the prevalence of cognitive impairment detected by neuropsychological tests is high (53%−86%). All four cognitive domains were affected, specifically the language, learning and memory, visuospatial skills, and executive function cognitive domains. This is consistent with the findings of previous literature on cognitive impairment in younger HCT patients. In a systematic review of 17 studies in adults undergoing HCT (average age ranged from 35 to 53, median 42.59), the prevalence of cognitive impairment before HCT varied widely depending on definition. ¹⁶ With the least restrictive definition, the prevalence was 89% (< −1.0 SD below normative means in one domain).²⁷ The prevalence was 12% using the most restrictive definition of < −1.5 SD in four subtests.²⁸ It is of note that our sample has a high prevalence of cognitive impairment (50%−86%) detected by neuropsychological tests despite using more stringent definitions.

In this study, we determined the participants to have significant cognitive impairment based on normative data for the neuropsychological tests used. We found that the MoCA is an appropriate screening test for detecting significant cognitive impairment on neuropsychological tests. According to the ROC analysis, the universal cut-off of <26 for screening of cognitive deficits in older adults without cancer detects close to 80% patients with impairment of ≤ −1.5 SD below normative means in at least one cognitive domain (Criteria 1), but with low specificity (57%) for detecting significant cognitive impairment on neuropsychological testing. However, using a lower cut-off score of 20 or less, 86 to 100% of patients are correctly classified as having significant cognitive impairment on neuropsychological testing using the criteria recommended by the ICCTF (Criteria 2–4).

While the primary objective of our current study is to describe the prevalence of cognitive impairment and cognitive domains affected, the findings suggest that the universal cut-off score of <26 on the MoCA should be interpreted with caution in older HCT candidates. Future studies with a larger sample size are needed to verify this finding. However, given the limited evidence available, clinicians may consider the following recommendations when evaluating cognitive function in this patient population: 1) a MoCA score of 29 or higher would indicate a very high probability of not having significant cognitive impairment on neuropsychological testing; and 2) a MoCA score of 20 or lower would indicate a very high probability of having significant cognitive impairment on neuropsychological testing.

Despite the high prevalence of cognitive impairment, less than 18% and 12% of patients have limitations in ADLs and IADLs, respectively. Thus, based on these findings, most of the patients in our sample have mild cognitive impairment (MCI). Patients with MCI have subjective cognitive complaints and objective evidence of cognitive impairment on cognitive testing, but preserved function when performing IADLs. Alternatively, these self-reported measures may not adequately capture impairments due to ceiling effects and under reporting of impairments. Future studies evaluating cognitive function in older HCT candidates should consider adding objective measures of functioning and report of functioning from caregivers. This is of clinical relevance because MCI is considered a transitional stage between normal cognitive aging and dementia. In the general population, MCI can be caused by multiple conditions and pathologies, most frequently Alzheimer’s Disease (AD).^29–32 The causes and clinical course of MCI in an older patient diagnosed with cancer are poorly understood and further investigation is needed.

Cognitive impairment in an older patient before HCT has many potential implications, including treatment decision making, ability to understand complex treatment regimen, self-management of symptoms, ability to tolerate the HCT process, and post-HCT outcomes such as cognitive function, physical function, and quality of life. It is important to screen for cognitive impairment with a measure like the MoCA because providers can identify patients who would benefit most from further evaluation including neuropsychological testing.

Cognitive tests represent only one component of the necessary work up for cognitive impairment and should be interpreted in the context of other clinical information including prior cognitive function and a history from a collateral source. Formal neuropsychological evaluation provides the most detailed assessments of cognitive abilities and observations of behavior. Neuropsychological testing compares test performance to normative data based upon age and education, identifies which cognitive domains and associated brain regions are affected, establishes a baseline for possible future comparison, identifies potential modifiable conditions that may be contributing to cognitive impairment (i.e. depression), and informs specific treatment recommendations.

Compared to other screening tests, MoCA is an appropriate screening measure for older patients being evaluated for HCT due to the broader range of cognitive abilities it assesses and its capacity to detect more subtle cognitive changes that occur in MCI. ³³ However, there are some important limitations for clinicians to consider regarding the use of cognitive screening measures, particularly risk of false negative and false positive errors. Although normal performance on a screening test may support a decision not to pursue formal neuropsychological testing, cognitive screening measures may fail to detect more subtle cognitive deficits that can cause distress for patients. ³⁴ Cognitive screening measures also carry a risk of false positive errors, particularly when used with individuals whose education level and/or cultural and linguistic backgrounds differ from that of the normative sample. ³⁴

There are limitations to this research. First, the study did not include a pre-treatment assessment. It is possible that cognitive impairment could be due to pre-existing cognitive impairment or could be attributable to other interventions such as previous intrathecal chemotherapy and cancer treatment for patients with a history of solid cancers. Second, this sample was highly educated, predominantly male and Caucasian, and treated at a tertiary cancer care center, which limits the generalizability of these results. However, the patients’ performances were compared with older adults without cancer using published normative data adjusted for age, education, and gender. Additional limitations include the modest sample size and the fact that half of our patients were accrued as a part of a research study evaluating cognitive function before and after HCT. However, there were no significant differences in the mean and SD of cognitive performances except for the domain of verbal fluency. This is likely due to a lower performance of one patient with clinical diagnosis of dementia who was seen in the clinic.

This study has some important strengths. It is the first study to examine the ability of the MoCA, a widely used cognitive screening tool to detect cognitive impairment on neuropsychological tests in older adults with hematological malignancies. Through this research, experience regarding the challenges of administration of a battery of neuropsychological tests in older adults with hematological malignancies is shared. Future studies with larger sample sizes are needed to confirm these findings and address whether the severity of cognitive impairment correlates with negative outcomes post-HCT in older adults (e.g. normal vs. MCI vs. dementia pre-HCT). In addition, future studies are needed to identify risk factors for older patients at high risk for cognitive decline through longitudinal design and determine the impact of pretransplant cognitive impairment, including MCI, on post-HCT outcomes.

CONCLUSIONS

HCT is a life-prolonging treatment for patients with hematological malignancies. While previously reserved for younger patients, a growing number of adults 65 years and older are receiving this intense treatment. There is a high prevalence of cognitive impairment identified by neuropsychological tests and MoCA in older patients with hematological malignancies evaluated for HCT. MoCA is a screening tool used to assess global cognitive function. Identification of an appropriate MoCA cut-off score in this population is important to identify patients who would benefit from neuropsychological testing.

APPENDIX A

Table 1:

Neuropsychological Tests

Domain	Instrument	Description	Range
Language	Verbal Fluency Test36	Subject is asked to name as many animals as possible in 1 minute.	0-No maximum. Higher is better
	Modified Boston Naming 1537	Subject is asked to name 15 objects presented with high, medium and low frequency of occurrence.	0–15 (Higher is better)
Learning & Memory	Word List Learning38	Subject is asked to read 10 printed words presented at the rate of 1 every 2 seconds on the first trial. Then subject is then asked to recall as many words as possible. The 10 words are presented in a new random order and the subject tries to recall all 10 on two subsequent trials. The score for Trial #3 is used for analysis.	0–10 (Higher is better)
	Delayed Recall38	Free recall of the 10 words presented in Word List Learning	0–10 (Higher is better)
	Word List Recognition38	Recognition of the 10 words presented in Word List Memory mixed with 10 distractor words.	0–10 (Higher is better)
Visuospatial Skills	Praxis 2D and 3D Figures39	Four line-drawings of figures of increasing complexity are presented to the subject for copying.	0–11 (Higher is better)
Executive Functioning	Trail Making Test Part A40	Subject is asked to draw lines to connect consecutively numbered circles quickly as they can.	0-No maximum. (Lower is better)
	Trail Making Test Part B40	Subject is asked to draw lines to connect consecutively numbered and lettered circles on Part B, as quickly as they can.	0-No maximum. (Lower is better)

Table 2:

Geriatric Assessments

Domain	Instrument	Range	Comments
Functional Mobility	Short Physical Performance Battery (SPPB) 20	0–12	A score of <8 indicates at risk for functional decline41
Activities of Daily Living (ADL)	Katz Activities of Daily Living Scale (ADLs)18	0–6	A score of 6 indicates no impairment in ADL18
Instrumental Activities of Daily Living (IADL)	Lawton Instrumental Activities of Daily Living (IADLs)19	0–8	A score of 8 indicates no impairment in IADL42
Falls	Fall Questionnaire43	0-No maximum	Patient reported falls of ≥ 1 in the last 6 months
Cognition	Montreal Cognitive Assessment (MoCA)11	0–30	A score of <26 indicates possible cognitive impairment based on initial validation study11
Depression	Geriatric Depression Scale- 15 Question (GDS-15)21	0–15	A score of >5 suggests depression44
Anxiety	Generalized Anxiety Disorder Screening- 7 (GAD-7)22	0–21	≥10 indicates possible diagnosis of generalized anxiety disorder22
Nutrition	Mini Nutritional Assessment Short Form (MNA)23	0–14	≥ 11 indicates normal status23
Social Support	Medical Outcomes Social Support Survey (MOS) 24	0–100	A score of 100 indicates maximum available support24