Abstract Little is known about the dynamics of CTCs during treatment and its clinical significance. We examined the predictive utility of serial CTC analysis in ER+HER2- MBC patients (pts) treated with chemotherapy in the CALGB 40502/NCCTG N063H study, a randomized phase III trial of weekly paclitaxel compared to weekly nanoparticle albumin bound nab-paclitaxel or ixabepilone +/- bevacizumab as first-line therapy (ClinicalTrials.gov Identifier: NCT00785291, Support: U10CA180821, U10CA180882). Methods: Of the 783 pts treated, 469 had ≥3 serial blood samples (including baseline) successfully analyzed for CTCs by CellSearch® and were included in this analysis (n=2,202). Samples with ≥5 CTCs per 7.5 mLs of blood were considered CTC+. The prognostic and predictive performance of baseline CTCs (bCTC) and CTC status from baseline to cycle 2 (b2CTC) were compared to a novel latent mixture model classification based on trajectory of CTCs (tCTC). Akaike Information Criterion (AIC) was used to select the model (bCTC vs b2CTC vs tCTC) that best predicts overall survival (OS), progression-free survival (PFS), and time-to-treatment failure (TTF). Results: 53% of the pts were CTC+ at baseline. b2CTC status changed in 36% of the pts, most of whom were CTC+CTC- (35%), and very few CTC-CTC+ (1%); the rest of the pts did not experience a change in b2CTC status (46% CTC-CTC- and 19% CTC+CTC+). Mixture model analysis revealed 4 groups of pts that show distinct tCTC patterns over the course of treatment: consitently very low/undectectable CTCs (tCTCneg, 56%), low (tCTClo, 24%), intermediate (tCTCmid, 15%), or high (tCTChi, 5%). bCTC, b2CTC, and tCTC were significantly correlated with tumor subtype (all p <0.0022) and presence of bone metastasis (all p <0.0001). Multivariate analysis showed that pts who were CTC+ at baseline, and those whose b2CTC status remained positive (CTC+CTC+) had significantly reduced OS, PFS and TTF. OSPFSTTFModelsHR (95% CI)p-valueHR (95% CI)p-valueHR (95% CI)p-valuebCTC (vs CTC-) → CTC+2.5(1.8-3.3)<0.00011.6(1.3-2.0)<0.00011.3(1.1-1.6)0.0046b2CTC (vs CTC+CTC-) → CTC-CTC+1.6(0.5-5.4)0.41491.6(0.6-4.5)0.39051.6(0.6-4.3)0.3961→ CTC+CTC+2.7(1.9-3.8)<0.00011.8(1.4-2.5)<0.00011.8(1.3-2.4)<0.0001→ CTC-CTC-0.5(0.4-0.8)0.00020.8(0.6-0.9)0.01600.9(0.7-1.1)0.2771tCTC (vs tCTCneg) → tCTClo2.6(1.9-3.7)<0.00011.9(1.4-2.4)<0.00010.9(0.7-1.1)0.0033→ tCTCmid5.3(3.6-8.0)<0.00012.5(1.8-3.4)<0.00011.8(1.4-2.5)0.0001→ tCTChi10.8(6.1-19)<0.00013.0(1.8-5.0)<0.00012.3(1.4-3.7)0.0009CTC- (<5 CTCs per 7.5 mLs); CTC+ (≥5 CTCs per 7.5 mLs) Pts with tCTClo, tCTCmid and tCTChi had significantly shorter OS, PFS and TTF compared to those with tCTCneg. After adjustment for potential confounders, AIC analysis revealed that the tCTC model best predicts OS and PFS, while b2CTC best predicts TTF. AIC Score*ModelsOSPFSTTFbCTC243240514199b2CTC240540384186tCTC237940264188*The lowest AIC score indicates the best model. Conclusions: Analysis of CTC trajectory patterns identified pts with poor outcome who could potentially benefit from more effective treatment. Validation in independent cohorts is warranted to confirm the findings in this study. Citation Format: Magbanua MJ, Hendrix L, Hyslop T, Barry WT, Winer EP, Hudis C, Toppmeyer D, Burnstein H, Qadir M, Ma C, Scott JH, Park JW, Rugo HS. Trajectory patterns of circulating tumor cells (CTC) in chemotherapy-treated metastatic breast cancer (MBC) patients predict poor clinical outcomes: CALGB 40502 (Alliance)/NCCTG N063H study [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P2-01-01.