e12557 Background: Atezolizumab was approved for PD-L1-positive advanced recurrent triple negative breast cancer. However, it is controversial about when and when to measure PD-L1. Here we retrospectively reaearched how PD-L1 positivity rates in biopsy, surgical, and recurrent specimens for triple-negative breast cancer (TNBC) treated with neo adjuvant chemotherapy (NAC). We also investigated the relationship between androgen receptor (AR) and PD-L1, and researched the significance of PD-L1 positivity on TNBC. Methods: 28 TNBC biopsy samples before NAC, 16 surgical specimens with residual tumor after NAC, and 5 cases reccurence specimens were obtained subjected to stain for PDL1 using the SP142 antibody. AR staining was also performed on biopsy specimens. PD-L1 score determination was performed according to the Ventana OptiView PD-L1 (SP142) inspection guide, and AR positive cuttoff was setted 1% or more. Results: The PD-L1 positivity rates of the biopsy samples were IC0: 46.4%, IC1: 25.0%, IC2: 21.4%, IC3: 7.1%, TC0: 28.6%, TC1: 28.6%, TC2: 42.9%, TC: 0%. In the surgical specimens after NAC, PD-L1 showed almost no change in IC ( P = 0.38), but the score significantly decreased in TC ( P < 0.001). The recurrence specimesn that could be collected were IC0: 60%, IC1: 20%, IC2: 20%, IC3: 0%, TC0: 60%, TC1: 20%, TC2: 20%. There were four AR-positive cases, one of which showed TC2, but the others were negative.In terms of prognosis, both the DFS and OS in the PD-L1 strongly positive group (IC≥2 and TC≥2) and other groups Poor in PD-L1 strongly positive group ( P < 0.01, P = 0.041, respectively). Conclusions: TC was affected by NAC, the score decreased markedly, and the positive rate at the site of recurrence also tended to decrease. The PD-L1 strong positive had a significantly worse prognosis than the others. It was suggested that PD-L1 immunostaining may be effective in predicting prognosis by considering TC as well as IC.