Cognition in Patients Treated with Targeted Therapy for Chronic Myeloid Leukemia: A Controlled Comparison

Abstract

This controlled comparison study evaluated objective and subjective cognitive function and their relationships with patient-reported symptoms (depression, fatigue, insomnia) in patients receiving tyrosine kinase inhibitors (TKIs) for chronic myeloid leukemia (CML) and non-cancer controls. Patients with CML in chronic phase treated with the same oral TKI for ≥6 months (n=90) and non-cancer controls (n=87) completed a neurocognitive battery and self-report measures. Patients demonstrated worse overall neuropsychological performance (p=.05) and verbal memory (p=.02) compared to controls. Patients were not more likely to meet criteria for impaired cognitive performance compared to controls(ps>.26). Patients reported worse subjective global and domain-specific cognitive complaints and less satisfaction with cognitive function compared to controls(ps<.05). Patients also reported greater fatigue and insomnia symptoms(ps<.001). In both groups, greater fatigue, insomnia, and depressive symptoms were associated with worse subjective cognition (ps<.01). Longitudinal studies are needed to examine changes in cognitive function in patients before and during TKI treatment.

Background

Chronic myeloid leukemia (CML) accounts for 20% of new adult leukemias annually, with an estimated 9,110 cases to be diagnosed in the United States in 2021.¹ Treatment of CML using BCR-ABL tyrosine kinase inhibitors (TKIs) is one of the first and most successful examples of targeted therapies in cancer treatment. Following the approval of imatinib in 2001, the eight-year survival rate for CML patients has improved from <20% to 82%.² Several second- and third-generation TKIs have since been approved for use in CML^3,4 and other oral medications targeting tyrosine kinase pathways continue to be developed for many types of cancer.⁵

Due to the high potential for disease relapse,^6,7 daily use of TKIs in CML can occur for years and may be life-long.⁸ Although imatinib and similar TKIs are better tolerated than regimens they replaced,⁹ they are not without side effects.¹⁰ Studies assessing patient-reported symptoms among CML patients treated with TKIs suggest that cognitive problems may be common. In one study, difficulty concentrating was endorsed by 50% of CML patients being treated with a TKI and was the third most commonly endorsed symptom of 26 symptoms assessed.¹¹ In another study, problems with remembering were among the top five most severe symptom among 20 symptoms assessed in CML patients being treated with TKIs.¹² Memory impairment has been reported as an adverse event in a small number of CML patients receiving the TKI dasatinib on clinical trials;¹³ however, these reports have relied on retrospective record review and case reports.^13,14 Although these findings suggest the importance of evaluating cognition in CML patients treated with TKIs, no published reports to our knowledge have comprehensively assessed cognition using standardized objective measures of neuropsychological functioning. This limits clinicians’ ability to describe potential effects of TKIs on cognitive function and limits researchers’ ability to develop interventions to prevent or reduce these effects.

Research suggests that fatigue, insomnia severity, and depressive symptoms are also commonly reported symptoms in CML patients receiving TKIs.^11,15,16 Given that these symptoms have been shown to be associated with subjective cognition in people being treated with anticancer agents,^17,18 it is important to evaluate their relationship to objectively and subjectively measured cognition in CML patients being treated with TKIs.

The goals of the current cross-sectional study were to: 1) compare cognitive functioning in CML patients treated with TKIs with similar non-cancer controls, 2) describe the rate of impaired cognitive performance in CML patients receiving TKIs, 3) examine the prevalence and demographic and clinical correlates of impaired cognitive performance in CML patients treated with TKIs, and 4) .explore the relationship of patient-reported symptoms to cognitive performance in patients.

Methods

Participants

Eligibility criteria for patients were as follows: age ≥ 18 years, ability to speak and read English, educational level ≥ 6^th grade, diagnosed with CML in chronic phase and treated with the same oral TKI for at least six months, no reported history of stroke or brain injury, and no reported history of head trauma with loss of consciousness within the past five years.

Eligibility criteria for non-cancer controls were as follows: age ≥ 18 years, ability to speak and read English, educational level ≥ 6^th grade, no history of cancer with the exception of non-melanoma skin cancer, no reported history of stroke or brain injury, no head trauma with loss of consciousness within the past five years, and have a valid mailing address and telephone number.

Procedures

Data for patients were collected between December 2015 and May 2019. These procedures were approved by the Chesapeake IRB. Written informed consent was obtained before initiation of study procedures. Patients being treated for CML at Moffitt Cancer Center (MCC) and Memorial Sloan Kettering Cancer Center (MSKCC) were identified through medical record review and approached for study participation in person or via postal mail. All study participants were scheduled for a 90-minute in-person assessment. The assessment included administration of standardized neuropsychological measures of cognitive performance, collection of demographic and clinical data, and completion of self-report questionnaires. Upon completion of the study assessment, MCC patients were paid $25 and MSKCC patients were paid $40.

Data for non-cancer control participants were collected between November 2011 and August 2014 as part of a Moffitt Cancer Center study on neurocognitive functioning among patients receiving hematopoietic stem cell transplant.¹⁹ These procedures were approved by the University of South Florida’s Institutional Review Board. Controls were identified using a national marketing database. Potential control participants who matched to transplant patients on gender and age (within 5 years) were randomly selected from zip codes in the greater Tampa metropolitan area and contacted via postal mail and telephone to assess interest in the study and screen for eligibility. Controls were also recruited via media outreach and flyers. Written informed consent was obtained and participants gave permission to use their data for future research.

Controls completed an assessment that included administration of standardized neuropsychological measures of cognitive performance, collection of demographic and clinical data, and completion of self-report questionnaires. Upon completion of the assessment, controls were paid $20. Controls from this study were used for the current analysis due to similarities in eligibility criteria, study methods, and types of data collected.

Measures

Demographic and Clinical Characteristics

A self-report form was used to collect demographic information for all participants. Clinical information about patients was abstracted from medical records.

Objective Cognitive Function

The neurocognitive test battery was selected for administration based on previous research of cognitive functioning in people with cancer^20,21 and recommendations put forth by the International Cognition and Cancer Task Force (ICCTF).²² Tests included measures of verbal memory (i.e., Hopkins Verbal Learning Test- Revised^23,24), visual memory (i.e., Brief Visuospatial Memory Test-Revised²⁵), attention and concentration (i.e., Digit Span subtest of the Wechsler Memory Scale,²⁶ Color Trails Test Part 1²⁷), and executive function (i.e., Color Trails Part 2,²⁷ Stroop Neuropsychological Screening Test Color-Word Task,²⁸ Controlled Oral Word Association Test²⁹). All measures have demonstrated reliability, validity, and published normative data for adults. Neuropsychological testing was administered and scored by trained research personnel supervised by experienced clinicians (H.S.L.J., M.B.J., P.B.J., J.R.). Premorbid intellectual functioning was estimated using the Wechsler Test of Adult Reading.³⁰

Subjective Cognitive Function

The Everyday Cognition Questionnaire was used to assess participants’ subjective evaluations of cognitive function in routine daily activities.³¹ Items are rated on a 4-point scale, then averaged to form seven domain scores: memory, language, visuospatial abilities, planning, organizing, divided attention, and satisfaction with cognitive function. Scores on the satisfaction with cognitive function domain are reverse scored, such that higher scores indicate more satisfaction. For all other domains, higher scores indicate worse subjective cognition. A Global Cognitive Complaints (GCC) score consists of all but the satisfaction item (Total score range: 1-4).

Self-Reported Symptoms

Participants completed the Fatigue Symptom Inventory,³² the Center for Epidemiological Studies – Depression Scale,³³ and the Insomnia Severity Index³⁴ to measure fatigue, depressive, and insomnia symptoms, respectively.

Statistical Methods

Sample Size Justification

The original planned sample size (N=45) was based on consideration of the number of potential participants available for recruitment.³⁴ This study was able to recruit a larger sample than 45 due to recruitment at multiple sites, thereby improving statistical power. The final sample size of 90 CML recipients and 87 controls yielded 80% power with a two-tailed alpha of .05 to detect mean group differences on cognitive performance in independent group t-tests of d=0.43 (a medium effect size). Given the exploratory nature of the current study, we did not employ corrections for multiple statistical tests.

Evaluation of Sample Characteristics and Study Outcomes

Analyses were conducted using SAS 9.4 (Cary, NC). Chi-square and t-tests were used to assess group differences in demographic and self-reported symptoms. Continuous raw scores were calculated for each neurocognitive test index. Neurocognitive test scores were converted to z-scores and adjusted for age, sex, and/or education, based on published normative data and the standard scoring metrics for each test. Z-scores were converted to t-scores for ease of interpretation (T score = 10(z score) + 50). Composite indices for the domains of memory (i.e., verbal, visuospatial), attention/concentration, and executive functioning were derived by averaging the t-scores of the test indices within each domain. A total neuropsychological performance (TNP) score was calculated by averaging the t-scores for all domains.

T-tests were used to identify group differences in neurocognitive performance at the test-level, domain-level, and total performance score-level. Multivariate models were used to identify group differences in subjective cognitive function, controlling for age and education. Cohen’s d effect sizes were calculated by dividing group differences in objective and subjective cognitive outcomes by the pooled standard deviation (SD) of both groups. Effect sizes were interpreted as follows: small (0.2 to 0.49), medium (0.5 to 0.79), or large (≥0.8).³⁶

Consistent with ICCTF guidelines²² and published literature,²¹ impaired cognitive performance was defined as scoring ≥ 1.5 SDs below published normative data on ≥ 2 tests or scoring ≥ 2.0 SDs below published normative data on ≥ 1 test. Chi-square tests were conducted to compare groups on rates of impaired cognitive performance. Pearson correlations and chi-square tests were conducted to examine the relationship of demographic and clinical characteristics and symptoms with impaired cognitive performance among TKI recipients.

Pearson’s r correlation coefficients were used to evaluate relationships between objective cognitive function (i.e., total neuropsychological performance), subjective cognitive function (i.e., global cognitive complaints), and patient-reported symptoms. Independent samples t-tests were conducted to compare groups on cognition and symptoms. An alpha level of 0.05 was used for all statistical analyses.

Results

Demographics

Participants included 90 patients and 87 non-cancer controls (see Supplementary Figure 1). Demographic, clinical, and self-reported symptoms for patients and controls are reported in Table 1. The groups did not differ on age, sex, education, employment status, income level, or marital status (ps≥.10). More patients identified as non-White and/or Hispanic (25%) compared to controls (13%) (p=.05). Patients reported greater symptoms of fatigue and insomnia severity compared to controls (ps<.001). Patients also reported greater symptoms of depression, with average scores more than 2 points higher on the CES-D compared to controls, however this did not reach statistical significance (p=.06). Groups did not differ on estimated IQ (p=.58), with mean scores in the average range.

Table 1.

Demographic characteristics, patient-reported outcomes, and estimated IQ of all participants (N=177).

Characteristic	TKI Recipients N=90	Healthy Controls N=87	p
	n (%)	n (%)
Age, M (SD) [Range]	50.71 (13.52) [24–76]	53.46 (14.29) [21–75]	.19
Gender			.29
Female	43 (48.31)	49 (56.32)
Male	46 (51.69)	38 (43.68)
Racial/Ethnic Background			.05
White and Non-Hispanic	63 (79.07)	73 (86.90)
Non-White and/or Hispanic	21 (25.00)	11 (13.10)
Marital Status			.28
Not Married	20 (22.47)	25 (29.76)
Married	69 (77.53)	59 (70.24)
Educational Attainment			.14
≤ Some College	39 (43.82)	28 (32.94)
≥ College Graduate	50 (56.18)	57 (67.06)
Work Status			.79
Not Working	33 (37.08)	25 (29.41)
Working	56 (62.92)	60 (70.59)
Annual Household Income			.53
≤ $39,999	10 (12.66)	7 (9.46)
≥ $40,000	69 (87.34)	67 (90.54)
Time since Diagnosis (years), M (SD) [Range]	6.11 (4.48) [.55–16.94]	-	-
Time on current TKI (years), M (SD) [Range]	4.12 (3.90) [.50–16.30]	-	-
Current TKI
Imatinib	25 (27.78)	-	-
Nilotinib	20 (22.22)	-	-
Dasatinib	33 (36.67)	-	-
Ponatinib	5 (5.56)	-	-
Bosutinib	7 (7.78)	-	-
Previous TKI Therapy	39 (44.32)	-	-
Depressive Symptoms, M (SD)	9.56 (7.49)	7.37 (7.49)	.06
Insomnia Severity, M (SD)	8.02 (5.28)	5.50 (4.66)	<.001
Fatigue Severity, M (SD)	3.34 (2.17)	2.35 (1.77)	<.001
Estimated IQ, M (SD)	108.94 (10.57)	109.77 (9.02)	.58

Note: TKI: tyrosine kinase inhibitor. IQ: intelligence quotient, estimated using the Wechsler Test of Adult Reading. M: mean. SD: standard deviation. p values calculated using t-tests for continuous characteristics and chi-square analyses for categorical characteristics.

Cognitive Function

Differences in Objective Cognitive Function

Mean-level scores by group for individual neuropsychological tests, cognitive function domains, and total neuropsychological performance are shown in Table 2. Patients had significantly lower total TNP scores (d=0.29, p=.05) and performed significantly worse than controls in the domain of verbal memory (d=0.36, p=.02). Groups did not differ in the visual memory, attention, or executive function domains (ps≥.09). At the test-level, patients performed worse than controls on the HVLT tests of immediate and delayed verbal memory (d’s=.36, 31, ps <.05) and Color Trails Test 1 (d=.43, p<.01).

Table 2.

Comparing TKI recipients and controls on objective and subjective cognitive function among patients (n=90) and healthy controls (n=87).

Objective Cognitive Function	TKI Recipients M (SD)	Healthy Controls M (SD)	p	Effect Size d
Verbal Memory	43.49 (9.60)	46.97 (9.74)	.02	0.36
HVLT: Immediate Verbal Memory	43.33 (9.55)	47.00 (10.66)	.02	0.36
HVLT: Delayed Verbal Memory	43.70 (10.97)	47.07 (10.93)	.04	0.31
Visual Memory	50.21 (11.21)	52.20 (11.04)	.24	0.18
BVMT-R: Immediate Visual Memory	48.96 (11.75)	51.84 (11.85)	.11	0.24
BVMT-R: Delayed Visual Memory	51.74 (11.79)	52.55 (11.26)	.64	0.07
Attention, Concentration	51.39 (6.62)	53.05 (6.39)	.09	0.26
Color Trails Test 1: Visual Attention, Psychomotor Speed, Simple Sequencing	49.19 (9.95)	52.90 (6.95)	<.01	0.43
Digit Span: Auditory Attention, Working Memory	53.48 (9.31)	53.22 (10.18)	.86	0.03
Executive Function	51.62 (7.17)	51.95 (7.41)	.76	0.05
Color Trails Test 2: Complex Sequencing	52.73 (11.51)	55.18 (7.54)	.10	0.25
Stroop Color-Word: Ability to Overcome Cognitive Interference	51.68 (9.46)	50.35 (10.63)	.39	0.13
COWA: Verbal Fluency, Sustained Effortful Output	50.42 (11.92)	50.01 (12.43)	.82	0.03
Total Neuropsychological Performance (TNP)	49.18 (6.08)	51.04 (6.65)	.05	0.29
Subjective Cognitive Function	TKI Recipients M (SD)	Healthy Controls M (SD)	p	Effect Size d
Global Cognitive Complaints (GCC)	1.45 (0.44)	1.21 (0.22)	<.001	0.69
Memory	1.88 (0.71)	1.44 (0.40)	<.001	0.76
Language	1.48 (0.62)	1.25 (0.33)	<.01	0.46
Visuospatial	1.16 (0.27)	1.09 (0.20)	.06	0.29
Planning	1.26 (0.45)	1.10 (0.22)	.01	0.45
Organization	1.35 (0.57)	1.20 (0.31)	.04	0.33
Divided Attention	1.66 (0.81)	1.28 (0.38)	<.001	0.60
Satisfaction with Cognitive Function	2.86 (1.31)	3.96 (1.11)	<.001	0.90

Note: HVLT: Hopkins Verbal Learning Test. BVMT-R: Brief Visuospatial Memory Test – Revised. COWA: Controlled Oral Word Association Test. TNP: Total Neuropsychological Performance. GCC: Global Cognitive Complaints Total Score from the Everyday Cognition Questionnaire. p values for objective cognitive function calculated using t-tests. p values for subjective cognitive function calculated using multivariate models controlling for age and education.

Impairment in Objective Cognitive Function

Among patients, 40% met criteria for impaired cognitive performance based on the criteria described above. Among controls, 32% met criteria for impaired cognitive performance. Group differences in rates of meeting criteria for impaired cognitive performance did not reach statistical significance (p=.28). The percentage of patients meeting criteria for impaired cognitive performance at the individual test level by group are shown in Table 3. Patients had higher rates of impairment on the Color Trials 2 test at both the 1.5 and 2 SD below the mean thresholds (ps .03 and <.02, respectively).

Table 3.

Test-level rates of impaired objective cognitive function by group

	Percentage with impaired performance 1.5 SD			Percentage with impaired performance 2 SD
Cognitive Test	Patients	Controls	p	Patients	Controls	p
Verbal memory
HVLT-R Total Recall	20.00	13.79	0.27	9.52	5.75	0.35
HVLT-R Delayed Recall	22.73	15.12	0.20	13.64	8.14	0.25
Visual memory
BVMT-R Total Recall	16.67	9.2	0.14	10.71	4.6	0.13
BVMT-R Delayed Recall	12.36	11.49	0.86	8.99	6.9	0.61
Attention
Color Trails 1	8.99	2.3	0.1	6.74	1.15	0.12
Digit Span	0	1.16	1	0	0	-
Executive function
Color Trails 2	10.00	2.3	0.03	7.78	0	<.02
Stroop color word	6.74	7.23	0.9	4.49	3.61	1
COWA	11.24	12.64	0.77	5.62	6.9	0.73

Patients who earned less than a college degree (X²(1, N=89) = 4.16, p=.04) and were fewer years since diagnosis (X²(1, N=90) = 4.63, p=.03) were more likely to meet criteria for impaired cognitive performance by scoring ≥1.5 SDs below norms on 2 test or ≥2 SDs below norms on 1 test. Time on TKI treatment was not associated with impairment status (p=.14). Patient-reported symptoms were not associated with impaired status (ps≥.21).

Differences in Subjective Cognitive Function

Mean-level scores for subjective cognitive function are shown in Table 2. Patients reported worse global cognitive function (d=0.69, p<.001) and less satisfaction with cognitive function (d=0.90, p<.001) than controls. Patients also reported worse subjective functioning in five of six domain-specific areas of memory, language, planning, organizing, and divided attention (ps<.05).

Relationship between Cognitive Function and Symptoms

In both groups, worse subjective cognitive function was associated with greater fatigue, insomnia severity, and depressive symptoms (Table 4). Correlation coefficients ranged from r =.34 to .49 among patients and from r = .33 to .51 among controls. Objective cognitive function was not associated with patient-reported symptoms among patients (ps≥.31). Objective cognitive function was associated with more depressive symptoms among controls (r=−.22, p=.04), but was not associated with fatigue or insomnia (ps≥.39).

Table 4.

The relationship of total neuropsychological performance, global cognitive complaints, and symptoms in TKI recipients and healthy controls

	Subjective Cognition		Fatigue		Insomnia		Depression
PATIENTS	r	p	r	p	r	p	r	p
Objective Cognition (TNP)	−0.0001	0.99	0.11	0.31	−0.08	0.43	−0.09	0.40
Subjective Cognition (GCC)			0.39	<.001	0.34	<.01	0.49	<.001
CONTROLS
Objective Cognition (TNP)	−0.01	0.95	−0.05	0.66	−0.10	0.39	−0.22	0.04
Subjective Cognition (GCC)			0.37	<.001	0.33	<.01	0.51	<.001

Note: TNP: Total Neuropsychological Performance. GCC: Global Cognitive Complaints Total Score from the Everyday Cognition Questionnaire

Discussion

To the best of our knowledge, the current study is the first to systematically evaluate objective and subjective cognitive function in CML patients receiving TKIs. The principal findings were that patients demonstrated worse overall neuropsychological performance compared to non-cancer controls; however, groups did not differ on rates of impaired cognitive performance using established cutoffs.^21,22 Patients endorsed significantly more global cognitive complaints and less satisfaction with cognition compared to controls. Patients also endorsed worse symptoms (fatigue, insomnia severity, depression), and symptoms were correlated with greater subjective cognitive complaints, but not with objective cognitive performance.

At the domain-level, patients demonstrated poorer verbal memory compared to controls, with small but clinically and statistically significant effect sizes. The significant group difference in performance on Color Trails Test 1, which measures visual attention, simple sequencing, and motor speed, is also notable. Given the high level of educational attainment and baseline intelligence in the sample, we would anticipate that patients would perform well on neuropsychological testing. Subtle differences between patients and controls may be quite meaningful, as even slight decrements in cognitive function may be noticeable and distressing to the individual. This is reflected in the subjective cognition results, in which patients reported significantly greater global cognitive complaints and less satisfaction with cognition than controls.

Non-statistically significant group differences in meeting criteria for impairment were in the anticipated direction, with more patients meeting the criteria than controls. The rate of impairment in the patient group (40%) is consistent with other populations of patients receiving anticancer therapies (i.e., chemotherapy, endocrine therapies),^20,21,37 although estimates range widely.³⁸ On average, more patients had impaired performance on Color Trails Test 2 than controls, although the overall proportion was small (<10% of patients). Findings are consistent with broader literature suggesting that cancer patients demonstrate deficits in cognitive function compared to healthy controls, but cancer-related cognitive impairment is typically mild to moderate in nature.³⁹

Demographic correlates of impairment in patients (e.g., education, estimated IQ) are consistent with those observed in general and cancer samples.^40,41 The observed association between male gender and cognitive impairment is consistent with findings in colon cancer patients, suggesting that men are at higher risk for cognitive decline compared to women.⁴² The correlation between fewer years since diagnosis and meeting criteria for impairment stands in contrast to previous findings of a positive relationship between amount of time on TKI treatment and cognitive deficits in patients with metastatic renal cell carcinoma or GIST tumors treated with VEGFR inhibitors.⁴³ Larger, longitudinal studies examining the impact of TKI treatment (e.g. type(s) of treatment, length of treatment) on cognition are warranted due to our findings and the relatively high rate of cognitive impairment we observed among cancer-free controls.

Patients endorsed significantly worse subjective cognition across multiple domains. In particular, patients reported substantially worse satisfaction with their cognition compared to controls (d=.90). It is possible that subjective cognition is more sensitive to change compared to cross-sectional neuropsychological testing. The lack of a baseline for neuropsychological testing may have made it difficult to detect group differences in objective cognition, whereas subjective reports allow individuals to compare their functioning to their own previous state. Further, worse subjective cognition may have more significant real-world implications for patients as compared to scores on neuropsychological testing. Recent literature has emphasized the clinical relevance of subjective cognition and patients’ dissatisfaction with cognition given the negative implications for functional status and quality of life.^44–46 This highlights the importance of assessing both objective and subjective cognition in cancer patients taking targeted therapies.

The current findings extend previous literature suggesting that behavioral symptoms (e.g., fatigue, insomnia), are associated with subjective cognition,^45,47 but not objective cognition in cancer patients. Given limited research on the pathophysiology of cognition in CML patients receiving TKIs and the cross-sectional nature of the current study, we cannot conclude the directionality of this relationship. It is possible that cognitive complaints are secondary to behavioral symptoms (e.g., insomnia, fatigue).⁴⁸ Alternatively, shared neuroendocrine-immune responses to cancer and its treatment may account for this cluster of behavioral comorbidities.^46,49 Regardless, interdependence of behavioral symptoms suggests that biobehavioral interventions targeting potential contributing factors (e.g., symptom distress, inflammation) may improve cognitive complaints.⁵⁰ This is supported by findings that interventions targeting fatigue can also improve subjective cognition, but not objective cognitive performance, in people with cancer.⁵¹

Patients endorsed significantly more subjective cognitive complaints and less satisfaction with cognition compared to controls. Subjective cognition was also associated with patient-reported symptoms. Mixed methods research is warranted to better describe patients’ cognitive complaints and expectations around cognitive changes in the context of TKI treatment.⁵²

Study findings are an important contribution to what little is currently known about the cognitive effects of targeted therapies, and represent the first comprehensive evaluation of the neuropsychological effects of TKIs in patients with CML. Strengths of the study include the novel focus on CML patients treated with BCR-ABL TKIs and the inclusion of a well-selected control group. Use of a guideline-recommended cognitive testing battery and established thresholds for evaluating cognitive impairment in cancer patients²² facilitated comparison of findings with the broader cancer and cognition literature. Limitations include the cross-sectional study design with no pre-TKI assessment, which limits the ability to draw causal conclusions or to evaluate change in cognitive performance over time. In addition, physiologic factors that may impact cognition were not assessed and neuroimaging studies were not conducted. Literature describing cognition in other populations of cancer patients taking TKIs suggests several potential mechanisms by which disease, TKI treatment, and immune-related process may impact memory and executive function.^43,53–55 Future studies should further examine the pharmacodynamics of BCR-ABL TKIs and mechanisms by which CML and TKIs might influence cognition, particularly neuroendocrine-immune pathways. Lastly, although the historical controls we used were well matched with patients on demographic factors, they were not explicitly matched, and it is possible that residual confounding may impact findings. We observed a higher rate of cognitive impairment than expected among controls. Recruiting younger controls can be challenging, and it is possible that the population of people that are available to participate are those with more time, which may be self-selecting in terms of lifestyle variables. We also noted a correlation between depressive symptoms and worse TNP in controls, but not patients. It is possible that depressive symptoms influenced control motivation and performance on neuropsychological testing. Future controlled, longitudinal studies with multiple controls groups (e.g., non-cancer controls, patients with cancer not receiving TKI treatment for CML) would help to elucidate the effects of TKIs on cognition.

In summary, findings suggest that patients with CML taking TKIs experience both objective and subjective deficits in cognitive function compared to healthy controls. Oncology providers should routinely assess patients for cognitive complaints and self-reported symptoms (e.g., fatigue) that may impact perceived cognitive functioning. TKI recipients who endorse changes in cognitive function should be assessed for other biobehavioral factors that may explain these changes and referred for neuropsychological testing as appropriate. Interventions targeting associated symptoms (fatigue, insomnia) may improve subjective cognitive complaints.