Abstract Background: It is not known which population of estrogen receptor (ER)-positive breast cancer patients should continue endocrine treatment beyond 5 years to overcome late recurrences. The aim of this study was to examine a combination of nuclear grade (NG) and expression level of ER and progesterone receptor (PR) to predict late recurrences. Methods: We assessed retrospectively 1677 consecutive ER-positive/HER2-negative patients who underwent surgical resection between 2004 and 2009. Patients with T2 or larger tumor and/or node-positive received pre- or postoperative chemotherapy following the international consensus panel from the St Gallen Conference, 2003. All patients had received adjuvant endocrine treatment. NG, ER and PR statuses were determined by immunohistochemistory on surgical specimen. We classified the patients into 3 groups as follows; ER-high (+++ or Allred score 7, 8)/PR-high (++ to +++ or 5-8) (n=212), ER-high / PR-low (- to + or 3-6) (n=208), and ER-low (+ to ++ or 3-6) / PR-any (n=255). We compared distant disease-free survival (DDFS) in each cohort based on the NG (1:low, and 2 or 3: high). Results: A median follow-up period was 77.0 months. Four hundred sixty seven patients (27.8%) received neoadjuvant chemotherapy, 208 patients (12.4%) received adjuvant chemotherapy, and 1002 patients (59.8%) did not received chemotherapy. Of the 467 patients with neoadjuvant chemotherapy, 65 patients (13.9%) had developed distant metastasis during study period (before 5 years in 51 (11.0%); and after 5 years in 14 (3.0%)). NG-low had significantly higher risk of late recurrence after 5 years than that of NG-high (p=0.005). According to hormonal receptor expression levels, in patients with NG-low, ER-low/PR-any had significantly higher overall DDFS rate than ER-high/PR-low (p=0.016). A similar trend was found before 5 years (p=0.077). However, ER-high/PR-high turned to have significantly high risk of recurrence after 5 years compared to ER-low/PR-any (p=0.024). Of the 208 patients with adjuvant chemotherapy, 16 patients (7.6%) had developed distant metastasis during study period (before 5 years in 5 (2.4%), and after 5 years in 11 (5.2%)). there was no association between a risk of recurrence and hormone receptor statuses at any study period. In the patients did not received chemotherapy, ER-high/PR-high had a trend of higher DDFS rate than others before 5 years (p=0.067). Of the 1002 patients without chemotherapy, only 36 patients (3.6%) had developed distant metastasis during study period (before 5 years in 27 (2.7%), and after 5 years in 9 (0.9%)). There was no difference of late recurrence after 5 years among the patients regardless of ER and PR expression level and NG with only low recurrence rate (0.9%). Conclusions: Our results demonstrated that, in ER-positive/HER2-negative patients who underwent neoadjuvant chemotherapy, NG-low/ER-high/PR-high should receive extend hormonal treatment over 5 years because of the high risk of late recurrence but NG-high/ER-high might not need. Furthermore, patients with T1 and node-negative may not need extend hormonal treatment because of the extremely low risk of late recurrence regardless of NG and hormone receptor statuses. Citation Format: Iwase M, Hayashi N, Yoshida A, Kajiura Y, Takahashi Y, Takei J, Suzuki K, Nakamura S, Yamauchi H. Hormone receptor expression level and nuclear grade associated with late recurrence in estrogen receptor-positive breast cancer patients. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P5-12-12.