Breast cancer (BC) diagnosis and treatment expose patients to a 5-fold higher risk of depression compared with the general population, with an estimated prevalence of 10% to 25%. A depressive episode in patients with BC has implications for the tolerance of and adherence to treatment, impairing quality of life and reducing life expectancy.To identify and characterize distinct longitudinal patterns of depressive symptoms in patients with BC from diagnosis to 3 years after treatment.The CANTO-DEePRESS (Deeper in the Understanding and Prevention of Depression in Breast Cancer Patients) cohort study included women in the French multicenter CANTO (CANcer TOxicities) cohort study (conducted between March 20, 2012 and December 11, 2018), who were 18 years or older with invasive stage I to III BC and no previous BC treatment. The study aimed to characterize toxicities over a 5-year period following stage I to III primary BC treatment. Assessments of depressive symptoms were performed on a subset of patients with available data at diagnosis and at least 2 other time points. All data were extracted from the CANTO database on October 1, 2020.The primary outcome was the level of depressive symptoms at each assessment time point measured with the Hospital Anxiety and Depression Scale and depression subscale at BC diagnosis and at 3 to 6, 12, and 36 months after the end of treatment. The group-based trajectory modeling was used to identify trajectory groups, and multinomial logistic regression models were used to characterize the following factors associated with trajectory group affiliation: demographic, socioeconomic, clinical, lifestyle, and quality-of-life data.A total of 4803 women (mean [SD] age, 56.2 [11.2] years; 2441 patients [50.8%] with stage I BC) were included in the study. Six trajectory groups that described the heterogeneity in the expression of depressive symptoms were identified: noncases with no expression of symptoms (n = 2634 [54.8%]), intermediate worsening (1076 [22.4%]), intermediate improvement (480 [10.0%]), remission (261 [5.4%]), delayed occurrence (200 [4.2%]), and stable depression (152 [3.2%]). HADS-D scores at diagnosis were consistently associated with the 5 depressive trajectory group affiliations, with an estimated higher probability per point increase of experiencing subthreshold or clinically significant depressive symptoms between diagnosis and the 3 years after the end of BC treatment. The higher probabilities ranged from 1.49 (95% CI, 1.43-1.54) for the intermediate worsening group to 10.53 (95% CI, 8.84-12.55) for the stable depression group. Trajectory groups with depressive symptoms differed from the noncases group without symptoms by demographic and clinical factors, such as having dependent children, lower household income, cancer stage, family history of BC, previous psychiatric hospitalizations, obesity, smoking status, higher levels of fatigue, and depression at diagnosis.In this cohort study, nearly a third of patients with BC experienced temporary or lasting significant depressive symptoms during and after treatment. Improving early identification of women at risk of developing long-term or delayed depression is therefore critical to increase quality of life and overall survival. Subjected to validation, this study is an important first step toward personalized care of patients with BC at risk of depression.< Réduire