This study will use non-invasive neuroimaging (i.e., MRI) to examine whether Mindfulness-Based Stress Reduction (MBSR) improves neural markers of cognitive function for postmenopausal women taking aromatase inhibitor (AI) therapy for breast cancer. The pilot randomized controlled trial will obtain preliminary efficacy of MBSR versus Health Enhancement Program (HEP) active control to improve neural markers of cognitive function. The final sample will include 32 postmenopausal women with breast cancer. MBSR and HEP groups will meet for a matched schedule of 8 weekly 2.5-hour sessions and a one-day weekend retreat. Specimen and data collection will be done at three time points: pre-randomization (i.e., within three weeks before beginning the intervention), within three weeks after completion of the intervention, and approximately three months (+/- three weeks) post intervention. Change scores for neuroimaging parameter estimates will be correlated with change scores for measures of cognitive function and affect. Differential expression of genes will be correlated with neuroimaging parameter estimates.

Adjuvant aromatase inhibitor (AI) therapy improves disease-free and overall survival for postmenopausal women after surgery for hormone receptor-positive breast cancer. Among symptoms associated with AI therapy are changes in cognitive function. Up to 25% of postmenopausal women with breast cancer report that they experience changes in cognitive function during AI therapy. Studies using neuropsychological tests found subtle deteriorations in verbal and visual learning and memory-as well as concentration, working memory, and executive function-for as many as a third of these patients. Changes in cognitive function may be associated with changes in affect (e.g., worry, depressive symptoms). Neural markers of cognitive changes, including changes in brain function and structure, may underlie changes in cognitive function.

The investigators' recent preliminary neuroimaging work to describe neural markers of cognitive changes suggests that postmenopausal women with breast cancer have inefficient cognitive-emotion processing before AI therapy, as evidenced by greater neural activity in the hippocampus (working memory) and amygdala (emotion processing) during task performance compared to controls. During AI therapy, patients show differential activation compared to controls in the dorsolateral prefrontal cortex (executive function and working memory), medial prefrontal cortices (cortical control of amygdala responses), and hippocampus.

Stress responses could partially explain relationships between AI therapy and neural markers of cognitive changes. The Mindfulness Stress-Buffering Account suggests that interventions such as Mindfulness-Based Stress Reduction (MBSR) may improve stress responses by attenuating negative appraisals of stress and reducing reactivity to stressful situations. For example, mindfulness meditation improved psychological stress responses in women with breast cancer. It improved some measures of cognitive function. Mindfulness practices reduced physiological markers of stress responses, including inflammatory markers in women with breast cancer and in stressed community adults, as well as cortisol reactivity for breast cancer survivors and during chemotherapy for colorectal cancer. Although similar neural deficits as were found in the investigators' preliminary work have been shown to improve in stressed adult populations using MBSR, it is not known whether the intervention improves neural deficits in women taking AI therapy (estrogen, production of which is blocked by AI therapy, is neuroprotective and promotes neural plasticity). Genetic variability was previously found to moderate the effect of MBSR on self-reported cognitive function. Therefore, it is possible that inter-individual variability in the expression of genes involved in stress responses could moderate relationships between AI therapy and neural markers of cognitive changes during MBSR. Taken together, MBSR may improve neural markers of cognitive changes shown in preliminary work to be deficient in postmenopausal women before and during AI therapy for breast cancer by targeting stress responses. Changes in these neural markers may correspond to improved self-report and neuropsychological measures of cognitive function.

Hypothesis: Stress reduction, moderated by gene expression, blunts the impact of AI therapy on neural markers of cognitive function, thereby improving cognitive function and affect in women with breast cancer.

Inclusion Criteria:

1. Female
2. <80 years of age by date of baseline assessment visit
3. Able to speak and read English
4. Post-menopausal (defined as [A] amenorrhea persisting for an entire year, [B] oophorectomy or ovarian suppression/ablation, or [C] hysterectomy and age >51 years)
5. Diagnosed with DCIS (stage 0) or stage I, II, or III breast cancer
6. Post lumpectomy or mastectomy and any re-excisions
7. Post neoadjuvant or adjuvant chemotherapy, if prescribed
8. Taking aromatase inhibitor (AI) therapy OR AI therapy scheduled to begin before planned post-intervention assessment visit

Exclusion Criteria:

1. Stage IV (metastatic) breast cancer

2. Diagnosis of a major psychiatric disorder (e.g., bipolar I disorder, schizophrenia, schizoaffective disorder)

3. Suicide attempt within the last 10 years

4. Hospitalization or residential treatment for psychiatric illness, eating disorder, or substance abuse within the last 2 years

5. History of neurological disease (e.g., Parkinson's disease, dementia)

6. History of head trauma

7. Claustrophobia

8. Unable to lie on the back

9. Ever been told not to get an MRI

10. MRI-incompatible metal implant*

    • If a potential participant reports implanted metal objects, which might be affected by MRI magnets, the study personnel and MRI technologist will screen over the phone or in person to determine whether the potential participant would be safe during the MRI scan. A current list of implants compatible with MRI will be consulted (http://www.mrisafety.com/TMDL\_list.php).

Even though the R00 study is not anticipated to reach the threshold for required genomic data sharing, the study will have in place plans for sharing de-identified data for secondary analyses with qualified investigators. The consent language for the R00 study will be worded for possible broad data sharing.

The timeline for submission to an NIH-designated data repository will allow first for publication of findings related to the aims of the R00 study, as well as submission of findings as preliminary data for anticipated grant application(s).

The investigators plan to submit genomic data and relevant phenotypic data (e.g., clinical characteristics of the sample) generated in the R00 study to an NIH-designated data repository in a timely manner. The submission process will likely include registration in the database of Genotypes and Phenotypes (dbGaP) and submission to Gene Expression Omnibus (GEO) for controlled access to the data.

For each participant, adverse event data were collected from the date of enrollment through the final assessment (i.e., approximately 6 months after enrollment).

questionnaire at each follow-up assessment (i.e., post intervention, 3 months post intervention)

The sample was less than the anticipated final sample size of 32; it included fewer minority participants than planned. The advent of the COVID-19 pandemic prohibited further recruitment and conduct of this study, which had unavoidable in-person activities (e.g., neuroimaging).

Participants will complete the baseline assessment and be randomly assigned to study group in a permuted block design stratified by receipt of chemotherapy (i.e., two strata). It will not be possible to blind study participants or interventionists to group assignment. To reduce bias, research staff who conduct assessments, including the PI, will be blinded to group assignment (i.e., single blind). Participants will be reminded not to reveal their group assignment to study staff conducting assessments. The study biostatistician will have no direct contact with study participants and will not be blinded to group assignment. The biostatistician will produce and maintain the randomization codes for the permuted blocks. Randomization may only be unmasked by the biostatistician at the completion of data collection, or for reporting of serious adverse events or unanticipated problems for which it will be essential to provide information to the PI on group assignment.