The characteristics of the 141 patients with DnIV BC are summarized in Fig. 1; Table 1. The median follow-up time from study enrollment was 47.3 months in the Surgery group (n = 61) and 11.35 months in the Non-surgery group (n = 80). The mean age was 59.9 years in the Surgery group and 63.1 years in the Non-surgery group (p = 0.6). All of the 61 patients in the Surgery group underwent loco-regional PTR surgery, and systemic treatment was performed in 50 of these 61 patients before their surgery. In the Non-surgery group, 74 of the 80 patients had received systemic treatment, and 42 patients were still receiving treatment at the data cutoff point.

Radiotherapy was performed for 10 patients with positive surgical margins and/or multiple metastatic lymph nodes after primary tumor resection and for 10 patients during systemic treatment in each group. The biological subtypes in the Surgery and Non-surgery groups were identified as Luminal in 30 patients (50%) and 46 patients (57.5%) respectively, Luminal Her2 in four (6.7%) and nine patients (11.3%), as pure Her2 in 19 (31.7%) and 10 patients (12.5%), and as TNBC in seven (11.7%) and 14 patients (17.5%) (p = 0.047). The Surgery group had more patients with the Her2 type compared to the Non-surgery group.

There were 25 patients with oligo metastases in the Surgery group (42%), and 66 patients with visceral metastases in the Non-surgery group (82.5%) (p = 0.002). In detail, the sites of oligo metastases in the Surgery and Non-surgery groups were diagnosed in seven and nine patients with bone metastases, 11 and two patients with lung metastases, and four and two patients with lymph node metastases, respectively. Two patients with liver metastases and a single patient with brain metastasis were also observed in the Surgery group.

In contrast, the sites of visceral metastases were identified in the Surgery and Non-surgery groups: eight and 25 patients with liver metastases, 15 and 10 patients with lung metastases, five and two patients with brain metastases, six and one patient with meningeal dissemination, two and five patients with multiple metastases in different organs, respectively, and six patients with pleural dissemination in each group (Table 1).

The treatment (systemic therapies and surgical PTR) and clinical responses in the Surgery and Non-surgery groups.

In the Surgery group, systemic treatment was performed in 14 patients (23.3% by chemotherapy; five patients (8.3%) by endocrine therapy, and in the remaining 38 patients (63.3%) by the combination of chemotherapy followed by endocrine therapy (C + E); three patients (5%) did not receive systemic therapy before or after their surgery. The preoperative treatment regimens using chemotherapies included anthracycline (A) for one patient, taxane (T) for 11 patients, A + T for 26 patients, and other cytotoxic regimens for nine patients with the use of anti-Her2 therapies including trastuzumab with/without pertuzumab combined with a conventional chemotherapy for 17 patients, and using endocrine therapies including an aromatase inhibitor (AI) for 16 patients, tamoxifen for three patients, and fulvestrant for one patient.

In the Non-surgery group, the systemic therapies included chemotherapy for 29 patients (36.7%), endocrine therapy for 27 patients (34.2%), and C + E therapy for 17 patients (31.5%); six patients (7.5%) did not receive any treatment (p < 0.0001). The systemic treatment regimens using chemotherapies included A for one patient, T for 21 patients, A + T for 15 patients, and other cytotoxic regimens such as eribulin, capecitabine, fluorouracil and others for 16 patients. The anti-Her2 therapies included trastuzumab with/without pertuzumab combined with conventional chemotherapy for 16 patients, and the endocrine therapies included an AI for 31 patients, tamoxifen for 12 patients, fulvestrant for 17 patients, and immune checkpoint inhibitors for two patients (atezolizumab for one patient and pemblolizumab for the other).

The clinical responses of the local primary tumor to the systemic therapies in the Surgery group before the surgery were seven (17.5%) complete responses (CRs), 32 (80%) partial responses (PRs) or stable disease (SD) and one patient (2.5%) with progressive disease (PD). The corresponding values in the Non-surgery group were one (1.3%) CR, 12 (15.8%) cases with a PR or SD, and 42 (55.3%) cases with PD (p < 0.0001) (Table 1).

The locoregional PTR surgery with/without axillary lymph node resection after systemic therapies included a total mastectomy for 51 patients and partial mastectomy or tumor resection for eight patients; 55 patients had undergone the axillary lymph node resection simultaneously. The details of the surgical procedure were not recorded for two patients. The pathological diagnoses indicated that five patients with and 27 patients without positive surgical margins. An unclear surgical margin was observed on the surgical specimens of the remaining 29 patients.

The median PFS and OS values were 88 months and 100.1 months in Surgery group and 30.3 months and 31.8 months in the Non-surgery group. Significantly better prognoses were thus achieved by the patients in the Surgery group compared to the Non-surgery group in both PFS (p = 0.004) and OS (p = 0.0002) (Fig. 2A, B). Further, the PFSs of Surgery group after surgery were 92.8 and 59.7 months in patients with Luminal type (n = 30) and Her2 + type (n = 23) DnIV BC, respectively. And the PFS in 7 patients with TNBC was not reached. In contrast, the PFSs of Non-surgery group after first systemic treatment were 44.1, 28.6 and 10 months in patients with Luminal type (n = 36), Her2 + type(n = 17) and triple negative(n = 7) DnIV BC, respectively. In addition, the patients in the Surgery group who had responded to their systemic therapy prior to undergoing a PTR and whose systemic disease was well controlled over an 8.1-month period showed significantly longer OS compared to the patients who had responded to the systemic therapy within < 8.1 months (p = 0.044) (Fig. 2C).

The survival times in the Surgery group and Non-surgery group. Kaplan-Meier curves of progression-free survival (PFS) in each group. A: Kaplan-Meier curves of PFS in each group. B: Kaplan-Meier curves of OS in each group. C: Kaplan-Meier of OS in the Surgery group according to the effective duration of systemic therapies prior to the surgery D: Kaplan-Meier curves of PFS according to the surgical margin of primary tumor resection in the Surgery group. E:Kaplan-Meier curves of OS according to the surgical margin of primary tumor resection in the Surgery group

In contrast, the effective duration of the systemic therapy was not significantly associated with PFS among the patients in the Surgery group (p = 0.297, Suppl. Fig. S1). Similarly, there was no significant association between the effectiveness duration of systemic therapy and the PFS or OS for the patients in the Non-surgery group (data not shown). Compared to the patients with positive surgical margins, those with negative surgical margins had significantly better clinical outcomes: PFS (p = 0.01) and OS (p = 0.008) (Fig. 2D, E).

The associations between clinical factors, the NLR, and the ALC with the Surgery group’s PFS and OS.

The univariate analysis results identified the following as significant factors in the Surgery group: the use of postoperative systemic therapy (p = 0.008), the NLR value prior to the surgery (p < 0.0001), and the post-operative NLR value at 1 year post-surgery (p = 0.034). The results of the multivariate analysis revealed a significant association with the PFS and OS for the use of postoperative systemic therapy (p = 0.0012) and the preoperative NLR value (p = 0.018) (Table 2 ).

The Kaplan-Meier (log-rank test) analysis demonstrated that the patients with a low NLR (≤ 3) prior to surgery or at 1 year after surgery had significantly better prognoses based on both PFS (Fig. 3A, B) and OS (Fig. 3D, E) compared to the patients with a high NLR (> 3) (p < 0.0001 and p = 0.034); in addition, a low NLR was associated with a significantly better PFS until 2 years after PTR surgery (p = 0.024) (Fig. 3C). This OS trend was observed until 2 years post-surgery (p = 0.074) (Fig. 3F). In contrast, the patients’ ALC values prior to surgery and at 1 or 2 years after surgery had no significant association with PFS or OS (Suppl. Fig. S2).

The PFS and OS rates in the Surgery group according to the patients’ NLR values prior to surgery (A, D), at 1 year (B, E), and 2 years (C, F) after the surgery. Upper panel: Kaplan-Meier curves of PFS according to the patients’ NLR values prior to the surgery (A) and at 1 year (B) and 2 years (C).Lower panel: Kaplan-Meier curves of OS according to the patients’ NLR values prior to the surgery (D) and at 1 year (E) and 2 years (F)

The associations between clinical factors, the NLR, and the ALC with the Non-surgery group’s PFS and OS.

The univariate analysis revealed that in the Non-surgery group, the use of systemic therapy (p < 0.0001) and the clinical responses to systemic therapy (p = 0.0012) were significantly associated with prognosis of DnIV BC, whereas other clinical factors including the biological subtype, the histology of primary or metastatic BC, and the metastasis status did not show this association. In contrast, the Non-surgery patients’ ALC value at 6 months (p = 0.026) and the ALC and NLR at 1 year after the start of systemic therapy (p = 0.007 and 0.001) were significantly associated with PFS (Table 3). The results of the multivariate analysis demonstrated significant associations of clinical factors including metastatic lymph node status (p < 0.0001), and the use of systemic therapies and the clinical responses to these therapies (p = 0.0006 and p = 0.0018) were significantly associated with PFS simultaneously. The NLR values at 6 months and 1 year after the start of systemic therapy were also significant (p = 0.025 and 0.0005) (Table2). Moreover, there were also no significant associations between NLR and nuclear grade (p = 0.157) in both groups.

The Kaplan-Meier (log-rank test) analysis showed that there was no significant difference associated with PFS or OS between the patients with a high ALC value (> 1500/µL) and those with a low ALC (< 1500/µL) before and at 6 months or 1 year after the initiation of systemic therapy. In contrast, the limited number of patients with a high ALC at 2 years after the start of systemic therapy (n = 3) had significantly better PFS and OS outcomes (p < 0.0001) (Suppl. Fig. S3). Similarly, the limited number of patients with a low NLR (≤ 3) at 1 year (n = 5) and at 2 years (n = 1) after the start of systemic therapy had significantly better clinical outcomes for both PFS and OS (p = 0.012 and < 0.0001) compared to the patients with a high NLR (> 3) (1 year: n = 5, 2 years: n = 3) (Suppl. Fig. S4).

As illustrated in Fig. 4, the Surgery group’s ALC (number/µL) and NLR values before and at 1, 3, 6, 12, and 24 months after the surgery were 1,534 and 2.37, 1,353 and 2.5, 1,207 and 2.33, 1,315 and 2.11, 1,354 and 2.05, and 1,500 and 2.03, respectively. In the Non-surgery group, the ALC and NLR values before and after systemic treatment were 1,501 and 2.76,1,341 and 2.57,1,341 and 1.99, 1,338 and 2.03, 1,535 and 1.75, and 1,509 and 1.71, respectively (Fig. 4). Our kinetics analyses revealed that both the ALC and the NLR decreased significantly after surgery in the Surgery group and after the start of the systemic therapies in the Non-surgery group at 6 months, and then the ALC re-elevated and recovered at 1 year and 2 years in both groups to the level at the initial treatment, whereas the decreased NLR continued at 1 and 2 years in both groups.

Kinetics of the absolute lymphocyte count (ALC) and neutrophil-to-lymphocyte ratio (NLR) according to the clinical treatment courses: A: The ALC (number/µL) before and at 1, 3, 6, 12, and 24 months after the surgery in the Surgery group (orange curve) or the initial systemic therapy in the Non-surgery group (green curve). B: The NLR before and at 1, 3, 6, 12, and 24 months after the surgery in the Surgery group (orange curve) or the initial systemic therapy in the Non-surgery group (green curve)