Cancer-Related Cognitive Outcomes Among Older Breast Cancer Survivors in the Thinking and Living With Cancer Study

Abstract

Purpose

To determine treatment and aging-related effects on longitudinal cognitive function in older breast cancer survivors.

Methods

Newly diagnosed nonmetastatic breast cancer survivors (n = 344) and matched controls without cancer (n = 347) 60 years of age and older without dementia or neurologic disease were recruited between August 2010 and December 2015. Data collection occurred during presystemic treatment/control enrollment and at 12 and 24 months through biospecimens; surveys; self-reported Functional Assessment of Cancer Therapy-Cognitive Function; and neuropsychological tests that measured attention, processing speed, and executive function (APE) and learning and memory (LM). Linear mixed-effects models tested two-way interactions of treatment group (control, chemotherapy with or without hormonal therapy, and hormonal therapy) and time and explored three-way interactions of ApoE (ε4+ v not) by group by time; covariates included baseline age, frailty, race, and cognitive reserve.

Results

Survivors and controls were 60 to 98 years of age, were well educated, and had similar baseline cognitive scores. Treatment was related to longitudinal cognition scores, with survivors who received chemotherapy having increasingly worse APE scores (P = .05) and those initiating hormonal therapy having lower LM scores at 12 months (P = .03) than other groups. These group-by-time differences varied by ApoE genotype, where only ε4+ survivors receiving hormone therapy had short-term decreases in adjusted LM scores (three-way interaction P = .03). For APE, the three-way interaction was not significant (P = .14), but scores were significantly lower for ε4+ survivors exposed to chemotherapy (−0.40; 95% CI, −0.79 to −0.01) at 24 months than ε4+ controls (0.01; 95% CI, 0.16 to 0.18; P < .05). Increasing age was associated with lower baseline scores on all cognitive measures (P < .001); frailty was associated with baseline APE and self-reported decline (P < .001).

Conclusion

Breast cancer systemic treatment and aging-related phenotypes and genotypes are associated with longitudinal decreases in cognitive function scores in older survivors. These data could inform treatment decision making and survivorship care planning.

INTRODUCTION

Cognitive problems commonly have been reported among breast cancer survivors before and after systemic therapy.^1-12 However, these declines are not universal,^8,13,14 can be subtle, can vary by treatment regimen, and may only affect certain subgroups.¹⁵ Older survivors have not been well studied, but should be at risk for cancer-related cognitive decline^1,2,6,15,16 because aging is associated with an increasing incidence of neurodegenerative disease and shares many common biologic pathways with putative mechanisms of cancer-related cognitive decline.¹⁵ Furthermore, chemotherapy produces changes in biomarkers¹⁷ and brain structure that mimic aging.^18-21 Chronologic age and aging phenotypes, such as frailty²² and/or high comorbidity burden, may be markers for risk of cognitive decline.³ Genotypes associated with neurodegenerative disease, including polymorphisms in the apolipoprotein E (ApoE) gene, a risk factor for Alzheimer’s disease, also have been reported to be associated with cancer-related cognitive decline.^2,23,24

Identification of specific risk factors for cancer-related cognitive decline in older populations has important implications for oncology care¹ because 75% of breast cancer survivors in the United States are 60 years of age and older²⁵ and because detection of subtle cognitive problems can be challenging in practice.^2,26,27 Among the few prospective studies that have examined risk of cognitive decline in older survivors,^3,26 few included a contemporaneous noncancer control group to assess the effect of aging²⁸ or examine whether risk factors vary in their effects by treatment regimen.^2,26

Thinking and Living With Cancer (TLC) is a multisite prospective study designed to fill this clinical gap. We used data from older breast cancer survivors and matched controls without cancer followed for 24 months to evaluate cognition after breast cancer and its therapies relative to that seen with aging alone. We focused on two cognitive domains related to aging and commonly affected in cancer-related cognitive decline: attention, processing, and executive function (APE) and learning and memory (LM).^1,2 We tested the hypothesis that older survivors exposed to chemotherapy (with or without hormonal therapy) would have lower neuropsychological domain and self-reported cognitive scores over time than survivors who received hormonal therapy only or controls. We also examined whether age, frailty, or comorbidity was independently related to cognitive scores and explored whether ApoE gene polymorphisms affected the differences in cognitive domain scores among treatment groups over time. The results are intended to inform clinical practice.

METHODS

This study was conducted at Georgetown University and affiliated practices in the Washington, DC, area; Memorial Sloan Kettering Cancer Center; Moffitt Cancer Center; City of Hope Comprehensive Cancer Center; Hackensack University Medical Center; Indiana University (IU) School of Medicine; and University of California, Los Angeles. IU and University of California, Los Angeles, joined the study for laboratory support and IU for participant recruitment in 2016, so data in this report are from the five other sites. All institutional review boards approved the protocol.

Setting and Population

We included participants recruited between August 1, 2010, and December 31, 2015; the study is ongoing. Eligible survivors were 60 years of age or older, newly diagnosed with primary nonmetastatic breast cancer, and English speaking. Those with stroke, head injury, major axis I psychiatric disorders, and neurodegenerative disorders were ineligible. Survivors with a history of other cancers were excluded if active treatment was for less than 5 years or they had systemic therapy. Among eligible survivors, 355 consented (36.5%; consent rate across sites, 17.2% to 72.7%; median, 62.5%; Fig 1). Consenting survivors were similar in age to nonparticipants.

Fig 1.

(A) Sample for evaluation of cognition in older breast cancer survivors and (B) matched controls without cancer. Participants were excluded if they failed the cognitive screen (at baseline). The percentage who consented and refused was calculated among those alive and eligible to continue the study at each time point. Eligibility for continuing in the study was the same as enrollment eligibility and included development of a neurologic disease (eg, stroke, Parkinson’s disease) and a diagnosis of cancer. Survivors who were diagnosed with breast cancer recurrence were excluded from assessment for the 6 months before diagnosis of recurrence. Participants may have refused an interview at one time point but then completed later interviews. Most participants completed two or three assessments (64% completed three, 21% completed two, and 15% completed baseline only). No significant differences existed in age, frailty, apolipoprotein E ε4 status, or self-reported cognition by number of completed assessments. Those completing baseline only tended to have slightly lower baseline attention, processing speed, and executive function and learning and memory scores than those completing two or more assessments and to be a survivor versus a control, which potentially underestimated declines in mean post-treatment 12- and 24-month scores. SD, standard deviation.

There were 362 consenting controls without cancer, including 88 friends. When no friend was available, we recruited age-, race-, education-, and site-frequency–matched controls. All controls met the same eligibility criteria as survivors.

Participants were screened using the Mini-Mental State Examination and the Wide Range Achievement Test 4 (WRAT4) Word Reading subtest; those with scores less than 24 or less than third-grade–equivalent reading level were ineligible (one control and one survivor, respectively). Controls who scored more than 3 standard deviations (SDs) below the control mean baseline neuropsychological scores for their age- and education group were ineligible post hoc (n = 8). Data for survivors who experienced a recurrence (n = 1) were excluded for the 6 months before recurrence. Nine consenting survivors and six controls did not complete baseline assessments. The final sample included 344 survivors and 347 controls. Among participants remaining alive and eligible, 73.6% and 70.5% of survivors and 89.5% and 77.2% of controls completed 12- and 24-month assessments, respectively (Fig 1).

Data Collection

Assessments included neuropsychological testing, a structured survey, and biospecimens for ApoE genotyping. Staff members were certified bi-annually on neuropsychological test administration. ApoE genotype was batch tested using TaqMan assays (rs429358 assay identifier: C_3084793_20; rs7412 assay identifier: C_904973_10; Life Technologies, Carlsbad, CA) on a 7900HT Fast Real-Time PCR System (Thermo Fisher Scientific, Waltham, MA); analyses were blinded to group and used TaqMan Genotyper Software version 1.3 (Thermo Fisher Scientific).

Measures

Outcomes.

The primary cognitive outcome was the domain-specific scores on neuropsychological tests of APE (six tests)^2,6,23; verbal LM (five tests) was the other outcome of interest. Visuospatial ability (two tests) was a secondary domain. We used recommended tests with established reliability and validity in older populations^29-31 and included instruments with equivalent forms³¹ where possible to minimize practice effects.

Factor analysis confirmed that domain structure and reliability were consistent for survivors and controls at all time points (Appendix Table A1, online only). A language domain included in earlier reports³ was dropped because it was not a separate factor and had limited variation. The visuospatial domain was not reported because of poor reliability. Secondary outcomes included cognitive subdomain scores and self-reported cognition on the basis of the Functional Assessment of Cancer Therapy-Cognitive Function (FACT-Cog^32,33; Cronbach’s α = 0.96); declines of 5% to 7%, or 7 to 10 points, were considered clinically meaningful.³⁴

Variables.

The primary predictor was treatment group (chemotherapy [with or without hormonal therapy], hormonal therapy only, or noncancer control). We explored whether ApoE genotype (ε4+ v not) affected treatment group differences in cognitive scores over time and examined whether baseline age, frailty, comorbidity burden (two or fewer v more than two illnesses), or diabetes were independently related to cognition.³ Frailty was measured using the Searle’s deficits accumulation index.^35-41 Our 40-item adapted index excluded cognition but included baseline comorbidity; prediagnosis/pre-enrollment physical, social, role, and emotional function using the 12-Item Short-Form Health Survey (Cronbach’s α = 0.85)⁴²; prediagnosis/pre-enrollment activities of daily living and instrumental activities of daily living⁴³; and baseline Timed Up and Go score.⁴⁴ Scores were categorized using established cut points (robust, 0 to less than 0.2; prefrail, 0.2 to less than 0.35; frail, 0.35 or greater).⁴⁰

Several covariates were examined as possible confounders of the effects of group on cognition. Sociodemographic measures included race (white v nonwhite), cognitive reserve (WRAT4 score), self-reported family history of dementia (first-degree relative, yes v no), married versus not married, and years of education. Lifestyle habits included self-reported ever use of any type of hormonal replacement therapy (excluding oral contraceptives), cigarette smoking (ever/current v never), and current alcohol use. Baseline function was assessed using the FACT-General (Cronbach’s α = 0.71).^45,46

Scores of 16 or greater on the Center for Epidemiologic Studies Depression Scale defined clinical depression.⁴⁷ The State-Trait Anxiety Inventory was used to measure state anxiety (Cronbach’s α = 0.86).⁴⁸ Fatigue was assessed using the FACT-Fatigue scale (Cronbach’s α = 0.90).⁴⁹ Clinical variables included surgery type, breast irradiation, biomarkers, and stage.

Statistical Analysis

Raw neuropsychological test results (Appendix Table A2, online only) were standardized to z scores using the baseline means and SDs of age- and education group–matched controls without cancer.⁵⁰ Standardized z scores were calculated for domains (Appendix Table A3, online only). Univariable tests compared characteristics by group and evaluated potential confounders. All participants with complete baseline data were included in the analyses, and the characteristics of those with two to three assessments (v one) were evaluated for relationships to key variables.

Linear mixed-effects models tested the protocol-specified analyses: the presence of group-by-time and group-by-time-by-ApoE interactions for cognitive domain, subdomain, and self-reported scores (and 95% CIs). These models included a participant-specific random effect. Age, race, WRAT4 score, and site were included as fixed effects to adjust for potential confounding effects. Given strong correlations, frailty, comorbidity, or diabetes was included in the models one at a time. We estimated that there was 80% power to detect a group-by-time effect size equivalent to a Cohen’s d of 0.3 when 30% to 40% of 342 survivors received chemotherapy.

In sensitivity analyses, we evaluated effects of group on specific tests or subdomain scores and whether fatigue, anxiety, depression, or smoking changed conclusions about interactions. Finally, we explored two- and three-way interactions of age with group and time. Analyses were conducted using SAS 9.4.b statistical software (SAS Institute, Cary, NC).

RESULTS

Participants ranged in age from 60 to 98 years and had an average of 15 years of education. There were no baseline differences in sociodemographic factors, cognitive scores, or ApoE genotypes between survivors and controls, except survivors were more often married and had a higher proportion who were frail (Table 1). Twenty-seven percent of survivors received chemotherapy (with or without hormonal therapy); the majority of chemotherapy regimens were anthracycline based, and most hormonal treatment was initiated with aromatase inhibitors (Table 1).

Table 1.

Baseline Characteristics of Older Breast Cancer Survivor Versus Controls Without Cancer

APE Domain

Cognitive scores tended to improve over time, consistent with expected practice effects. However, there was a significant group-by-time interaction (P = .05) where survivors exposed to chemotherapy did not show practice effects and actually had declines in adjusted mean APE scores, whereas the other groups increased over time (Fig 2). Baseline frailty was an independent predictor of baseline APE scores (P < .001; Table 2). In models that included comorbidity or diabetes instead of frailty, those with more than two comorbid conditions (v two or fewer; P < .01) had significantly lower baseline mean APE scores independent of other effects (Appendix Table A4, online only), and diabetes (v no diabetes) was borderline significantly associated with lower scores (P = .09; Appendix Table A5, online only).

Fig 2.

Adjusted mean cognitive scores over time for older breast cancer survivors and controls without cancer. Adjusted mean cognitive domain scores on the basis of least squares means from linear mixed-effects models show scores at baseline, 12 months, and 24 months for three treatment groups, including survivors who received chemotherapy with or without hormonal therapy, survivors who received only hormonal therapy, and controls. The models included as fixed effects time; group; apolipoprotein E genotype; all two- and three-way interactions for group, apolipoprotein E, and time; baseline age; frailty; standardized Wide Range Achievement Test 4 score; race; and recruitment site. Adjusted mean scores are shown by treatment group and time for the (A) attention, processing speed, and executive function (APE) domain (P = .05 for group-by-time interaction) and (B) learning and memory (LM) domain (P = .03 for group-by-time interaction) for the genotypes combined. (C) Self-reported cognition scores on the Functional Assessment of Cancer Therapy-Cognitive Function (FACT-Cog), with higher scores indicating better cognitive function. Declines of 5% to 7%, or 7 to 10 points, on this 148-point scale are considered clinically meaningful.³⁴ Tables 2 and 3 include the 95% CIs for each mean score at each time point and for each outcome.

Table 2.

Adjusted Mean APE and LM Domain Cognitive Scores Among Older Breast Cancer Survivors and Controls Without Cancer by Time, Treatment, and Genotype

The three-way group-by-time-by-ApoE interaction was not statistically significant for APE scores (P = .14; Table 2; Fig 3A). Despite this, an inspection of the means indicated that the small number of survivors who were ApoE ε4+ and exposed to chemotherapy had lower adjusted APE mean scores at 24 months (−0.40; 95% CI, −0.79 to −0.01) than the ApoE ε4+ controls (0.01; 95% CI, 0.16 to 0.18; P < .05).

Fig 3.

Adjusted mean cognitive scores over time for older breast cancer survivors and controls without cancer by apolipoprotein E (ApoE) status. Adjusted mean cognitive domain scores on the basis of least squares means from linear mixed-effects models show scores at baseline, 12 months, and 24 months for three treatment groups, including survivors who received chemotherapy with or without hormonal therapy, survivors who received only hormonal therapy, and controls. The models included as fixed effects time; group; ApoE genotype; all two- and three-way interactions for group, ApoE, and time; baseline age; frailty; standardized Wide Range Achievement Test 4 score; race; and recruitment site. (A) Results for the attention, processing speed, and executive function (APE) domain for group-by-time-by-ApoE ε4 positivity, where adjusted means are plotted for participants who are ApoE ε4+ and ApoE ε4− (P = .14 for three-way interaction). (B) Results for the learning and memory (LM) domain for group-by-time-by-ApoE ε4 positivity, where adjusted means are plotted for participants who are ApoE ε4+ and ApoE ε4− (P = .03 for three-way interaction). (C) Results for self-reported cognition on the Functional Assessment of Cancer Therapy-Cognitive Function (FACT-Cog) scale for group-by-time-by-ApoE ε4 positivity, where adjusted means are plotted for participants who are ApoE ε4+ and ApoE ε4− (P not significant for three-way interaction). Declines of 5% to 7%, or 7 to 10 points, on this 148-point scale are considered clinically meaningful.³⁴ Tables 2 and 3 include the 95% CIs for each mean score at each time point for each outcome.

Depression, anxiety, fatigue, or smoking and other lifestyle factors did not affect the mean scores for the various group-by-time or group-by-time-by-ApoE combinations (data not shown). Results for the APE subdomain and individual test scores followed a similar pattern as the overall domain results (data not shown). Older age was significantly associated with lower baseline cognitive scores, but the effects of treatment over time did not vary by age (data not shown).

LM Domain

A significant group-by-time interaction was found where survivors taking hormonal therapy alone had less improvement in cognitive scores at 12 months than other groups but improved by 24 months (P = .03 for two-way interaction; Fig 2). There was also a statistically significant group-by-time-by-ApoE interaction (P = .03), where differences in LM scores between treatment groups over time were largely confined to those who were ApoE ε4+ and initiated hormonal therapy. This group had a small LM decline at 12 months but subsequent improvement at 24 months, whereas other genotypes and groups showed early improvements (Table 2; Fig 3B). Frailty, comorbidity, diabetes, and other covariates did not change the conclusions with regard to interaction effects. LM subdomain and individual test scores followed a similar pattern as the overall domain results, and there was no interaction of age with longitudinal treatment effects (data not shown).

Self-Reported Cognition

Self-reported cognition was moderately but significantly correlated with APE and LM (Pearson’s r = 0.40 to 0.41 at baseline, 12, and 24 months; all P < .001). Adjusted mean self-reported cognitive scores for survivors exposed to chemotherapy decreased nonsignificantly, whereas the other groups did not change over time (Fig 2). Baseline frailty was independently related to baseline self-reported cognitive scores. ApoE ε4+ survivors who received chemotherapy showed a clinically meaningful decrease in adjusted mean self-reported score (from 133.1 [95% CI, 123.1 to 143.0] at baseline to 126.0 [95% CI, 114.0 to 138.0] at 24 months, a 7-point mean decrease), but the group-by-time-by-ApoE interaction was not statistically significant (Table 3; Fig 3C).

Table 3.

Adjusted Mean Self-Reported Cognitive Function Scores Among Older Breast Cancer Survivors and Matched Controls by Time, Treatment, and Genotype

DISCUSSION

The TLC is one of the largest prospective, controlled studies of cognitive function among older breast cancer survivors. Our results for the first 2 years after diagnosis indicate that systemic treatment and aging-related genotype and phenotypes are associated with cognitive decline. Older survivors exposed to chemotherapy had significantly lower longitudinal cognitive function scores on the APE domain than other groups, and this effect was largely confined to those with the ApoE ε4 allele. This genotype also was associated with having lower LM scores after hormonal therapy initiation. Older age and frailty were independently related to lower baseline cognitive scores.

Most^{2,5,6,8,9,26,51} but not all^13,14,52 studies of breast cancer–related cognitive decline have reported cognitive problems after chemotherapy among predominately younger survivors. The current findings confirm an adverse effect of chemotherapy on APE scores in older survivors on the basis of not only failure to show the expected practice effects⁷⁴ but also declining scores. In other studies, the ApoE ε4 genotype was linked to postchemotherapy decline in similar domains and with reductions in gray matter in younger breast cancer^23,53 and testicular cancer survivors.^54-57 The current data suggest a similar selective deficit in APE among ApoE ε4 carriers exposed to chemotherapy, but the overall interaction effect was not statistically significant.

In our cohort, the ApoE ε4 genotype also was associated with small, nonpersistent decreases in LM scores after hormonal therapy initiation. Some reports have noted decrements in LM after hormonal therapy,⁵⁸ but others have shown inconsistent results.^10-12,59 Because even short-term cognitive deficits are meaningful to survivors, confirmation of our results and extension of follow-up will be important. Additional knowledge about genotype-treatment interactions could suggest mechanistic pathways, be used in decisions to recommend extended hormonal therapy duration,⁶⁰ and could potentially affect the use of direct-to-consumer ApoE testing.⁶¹

Chronologic age and aging phenotypes also were associated with lower baseline APE and self-reported cognitive function scores. Baseline cognitive scores have been shown to be a predictor of cognitive trajectories in other older cancer cohorts.³⁹ These results, together with the growing body of evidence from other studies,^1,2,15,28,62 support the idea that chemotherapy (and possibly hormonal therapy) can lead to cancer-related cognitive declines through acceleration of aging processes. Aging processes, ApoE ε4 genotype, and insulin resistance (seen with diabetes, a common comorbidity and component of frailty) each has been related to inflammation, which in turn is one of the putative risks for cancer-related cognitive decline and Alzheimer’s disease.^63,64 Aging and the ApoE ε4 genotype also reduce brain plasticity and repair, another possible mechanism of cancer-related cognitive decline.⁶⁵ Biomarker and imaging studies may provide additional insights into the role of aging processes in cancer-related cognitive decline. For instance, Sanoff et al¹⁷ reported that chemotherapy exposure in younger breast cancer survivors was associated with increased expression of p16^INK4a, a marker of cellular senescence, at levels equivalent to 10 to 15 years of chronologic aging. Neuroimaging studies of younger survivors have shown postchemotherapy decreases in frontal gray matter volume, abnormalities in brain network structure, and lower hippocampal volume consistent with aging.^18-21 We suggest that future research on mechanistic pathways focus on areas of overlap among aging processes, Alzheimer’s disease, and risks for cancer-related cognitive decline.

Despite the strength of the evidence and rigor of our design, several limitations should be considered in interpreting the findings. First, the functional effect of the observed cognitive declines is uncertain, and we do not know whether survivors will develop dementia-related diagnoses. Prior research with older survivors found that accelerated self-reported cognitive decline is associated with lower physical and emotional function over 7 years postdiagnosis.³⁹ Second, the study population was well educated, cognitively intact at baseline, and recruited primarily from academic centers and affiliated community hospitals, which potentially underestimates cognitive declines in general populations. Third, despite the large sample, statistical power was low to detect a significant three-way interaction of group-by-time-by-genotype effects because relatively few underwent chemotherapy, and rates of ApoE ε4+ were low. Furthermore, too few participants had two copies of the ε4 allele to test dose-response relationships. Other genotypes, such as COMT and BDNF, also may be important in cancer-related cognitive decline.^24,66 Pooling of samples from other studies of older survivors using similar eligibility and assessments could be used to confirm results and increase statistical power for detecting significant gene-treatment effects on cognition over time. Fourth, although we used recommended tests,²⁹ drawing conclusions about whether specific subdomains were affected differentially by treatment was difficult because many tests capture multiple cognitive constructs, and some tests did not have alternative forms and might show greater improvement with practice than tests with alternative forms. Fifth, there was limited treatment variability, so the evaluation of specific agents was not possible.⁹ Hormonal therapy effects were based on treatment initiation and assumed adherence for the first 24 months. Early discontinuation as a result of cognitive problems could underestimate the effects of this modality on outcomes. Finally, too few survivors had human epidermal growth factor receptor 2–positive or hormone receptor–negative tumors to assess the respective effects of trastuzumab or chemotherapy alone. Future clinical studies and preclinical experiments^67,68 will be necessary to examine the separate and combined effects and mechanisms of specific agents.

In summary, older breast cancer survivors with aging-related phenotypes and genotypes may be at risk for cognitive decline, especially after chemotherapy. These results could be useful in several ways to clinicians who care for older adults. First, information about risks for cognitive decline could help clinicians to discuss treatment options when chemotherapy is discretionary because many older cancer survivors are concerned about cognitive problems related to their cancer and its treatment.^69-72 The low percentage of older survivors with cognitive decline also could provide some reassurance if chemotherapy is clinically indicated. Second, knowledge of the cognitive effects of systemic therapy could prompt plans for monitoring during survivorship care to facilitate adherence to long-term cancer and other medical therapies.⁷³ Cognitive function monitoring also could be useful to flag survivors at risk for impaired daily functioning as a consequence of cognitive decline.^39,51 Finally, geriatric assessments that measure cognitive function before treatment and during the survivorship phase of cancer care could provide data for risk prediction tools and assist clinicians in identifying older adults for preventive or other interventions to maximize function and healthy lifespans.

Appendix

Table A1.

Neuropsychological Tests and Domains Used to Assess Cognition in Older Breast Cancer Survivors and Matched Controls Without Cancer

Table A2.

Raw Baseline Neuropsychological Testing Scores of Pretreatment Cognitive Performance for Older Breast Cancer Survivors and Matched Controls Without Cancer

Table A3.

Unadjusted Standardized Mean z Scores for Older Breast Cancer Survivors and Matched Controls Without Cancer by Assessment Time and Treatment Group

Table A4.

Adjusted Mean Domain Scores Among Older Breast Cancer Survivors and Matched Controls Without Cancer by Time and Genotype: Comorbidity Effects

Table A5.

Covariate-Adjusted Mean Domain Scores Among Older Breast Cancer Survivors and Matched Controls Without Cancer by Time and Genotype-Diabetes Effects

AUTHOR CONTRIBUTIONS

Conception and design: Jeanne S. Mandelblatt, Gheorghe Luta, Deena Graham, Jonathan Clapp, Claudine Isaacs, Paul B. Jacobsen, Robert A. Stern, Andrew J. Saykin, Tim Ahles

Financial support: Jeanne S. Mandelblatt

Administrative support: Jeanne S. Mandelblatt, Raymond Turner, Andrew J. Saykin, Tim Ahles

Provision of study materials or patients: Heather Jim, Brenna C. McDonald, Asma Dilawari, Andrew J. Saykin, Tim Ahles

Collection and assembly of data: Jeanne S. Mandelblatt, Arti Hurria, Heather Jim, Brenna C. McDonald, Deena Graham, Jonathan Clapp, Neelima Denduluri, Paul B. Jacobsen, Kelly Holohan Nudelman, Andrew J. Saykin, Tim Ahles

Data analysis and interpretation: Jeanne S. Mandelblatt, Brent J. Small, Gheorghe Luta, Arti Hurria, Heather Jim, Brenna C. McDonald, Deena Graham, Xingtao Zhou, Jonathan Clapp, Wanting Zhai, Elizabeth Breen, Judith E. Carroll, Neelima Denduluri, Asma Dilawari, Martine Extermann, Claudine Isaacs, Paul B. Jacobsen, Lindsay C. Kobayashi, James Root, Danielle Tometich, Raymond Turner, John W. VanMeter, Andrew J. Saykin, Tim Ahles

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Cancer-Related Cognitive Outcomes Among Older Breast Cancer Survivors in the Thinking and Living With Cancer Study

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/site/ifc.

Jeanne S. Mandelblatt

No relationship to disclose

Brent J. Small

No relationship to disclose

Gheorghe Luta

No relationship to disclose

Arti Hurria

Consulting or Advisory Role: GTx, Boehringer Ingelheim, On Q Health, Sanofi, OptumHealth, Pierian Biosciences, MJH Healthcare Holdings

Research Funding: GlaxoSmithKline, Celgene, Novartis

Heather Jim

Consulting or Advisory Role: Janssen Pharmaceuticals, RedHill Biopharma

Brenna C. McDonald

No relationship to disclose

Deena Graham

Stock and Other Ownership Interests: Cota

Xingtao Zhou

No relationship to disclose

Jonathan Clapp

No relationship to disclose

Wanting Zhai

No relationship to disclose

Elizabeth Breen

No relationship to disclose

Judith E. Carroll

No relationship to disclose

Neelima Denduluri

Research Funding: Amgen (Inst), Novartis (Inst), Genentech (Inst)

Asma Dilawari

Consulting or Advisory Role: Cardinal Health

Travel, Accommodations, Expenses: Cardinal Health

Martine Extermann

Research Funding: GTx

Claudine Isaacs

Honoraria: Genentech, Roche, AstraZeneca, Pfizer

Consulting or Advisory Role: Pfizer, Genentech, Roche, Novartis, AstraZeneca, Medivation, NanoString Technologies, Syndax

Speakers’ Bureau: Genentech, Pfizer, AstraZeneca

Research Funding: Novartis (Inst), Pfizer (Inst), Genentech (Inst), Tesaro (Inst)

Patents, Royalties, Other Intellectual Property: UpToDate, McGraw-Hill

Paul B. Jacobsen

No relationship to disclose

Lindsay C. Kobayashi

No relationship to disclose

Kelly Holohan Nudelman

No relationship to disclose

James Root

No relationship to disclose

Robert A. Stern

Stock and Other Ownership Interests: King-Devick Technologies

Consulting or Advisory Role: Eli Lilly, Avanir Pharmaceuticals

Patents, Royalties, Other Intellectual Property: Royalties for creating neuropsychological tests and published commercially by Psychological Assessment Resources

Danielle Tometich

No relationship to disclose

Raymond Turner

Research Funding: Biogen (Inst), Eli Lilly (Inst), Novartis (Inst), Merck (Inst), Acadia Pharmaceuticals (Inst)

John W. VanMeter

No relationship to disclose

Andrew J. Saykin

Consulting or Advisory Role: Bayer AG

Research Funding: Eli Lilly (Inst)

Tim Ahles

No relationship to disclose