Differences in drug efficacy and prognosis between primary and metastatic sites for de novo stage IV breast cancer: an exploratory analysis of a phase III trial, JCOG1017

Abstract

Background

Breast cancer is a highly heterogeneous disease, with biological factors like estrogen receptor, progesterone receptor, and HER2 receptor differing between primary and metastatic sites, potentially affecting treatment response.

This exploratory analysis aims to differentiate drug efficacy and long-term prognosis between these sites in stage IV breast cancer.

Methods

Patients from JCOG1017, a phase III trial evaluating the role of primary tumor resection, who received primary systemic therapy (PST) were evaluated at three months. In this analysis, treatment response was assessed separately in primary and metastatic sites. Patients were categorized into four groups: discordant I (primary non-PD, metastatic PD), concordant I (both PD), discordant II (primary PD, metastatic non-PD), and concordant II (both non-PD).

Results

Among 271 patients, overall discordance proportion of treatment response between primary and metastatic sites was 25.1%. Group distribution was 24.7% (discordant I), 9.6% (concordant I), 0.4% (discordant II), and 65.3% (concordant II). Discordance was more frequent in luminal (28.9%), triple-negative (25.0%), and luminal-HER2 (22.0%) subtypes than in HER2-enriched (11.1%). Survival analysis showed prognostic differences: concordant II, with both sites non-PD, demonstrated the most favorable outcome compared with discordant I (HR 0.556; 95% confidence interval, 0.396–0.782).

Conclusions

One-fourth of patients exhibited discordant responses between primary and metastatic sites in early treatment phases. These discrepancies were associated with survival differences, emphasizing the importance of evaluating both primary and metastatic lesions when assessing efficacy and determining treatment strategies in de novo stage IV breast cancer.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12282-026-01846-3.

Introduction

Breast cancer is a highly heterogeneous disease, and there is a biological diversity. Biological factors such as estrogen receptor, progesterone receptor, and HER2 receptor may vary between primary and metastatic sites in breast cancer. This variation could influence treatment responses in stage IV breast cancer. Previous studies[1–3] have indicated that primary and metastatic lesions may respond differently to systemic therapy due to factors such as clonal evolution and microenvironmental influences. In particular, systemic therapy for patients with de novo stage IV breast cancer who have both primary and metastatic lesions is often selected based on the biology of the primary tumor, which is more readily accessible for biopsy and evaluation. However, it is not uncommon that such treatment proves ineffective against metastatic sites. Whether these biological discrepancies and the resulting choice of systemic therapy affect prognosis remains unclear.

The JCOG1017 trial[4], a phase III study, assessed the role of primary tumor resection in de novo stage IV breast cancer. In JCOG1017, patients diagnosed with de novo stage IV breast cancer received primary systemic therapy (PST) according to subtypes after first registration. Primary tumor resection could not significantly prolong the overall survival (OS) of de-novo stage IV breast cancer patients with sensitivity to PST[5].

This exploratory analysis aims to differentiate drug efficacy and long-term prognosis between primary and metastatic sites in patients with de novo stage IV breast cancer.

Patients and methods

Study design and population

This study is an exploratory analysis of JCOG1017, which included patients diagnosed with de novo stage IV breast cancer.

The key eligibility criteria for the first registration in JCOG1017 were aged 20–80 years with an Eastern Cooperative Oncology Group performance status (PS) of 0 or 1 (PS 2 caused by the symptoms of bone metastasis is also eligible.), histologically proven invasive breast cancer and without bilateral breast cancer or extension to the contralateral breast. The patients with at least one measurable metastatic lesion other than the breast tumor and axillary lymph nodes, and without brain metastasis, were included. Other eligibility criteria were no prior chemotherapy for breast cancer or radiotherapy to the ipsilateral breast, no surgery, chemotherapy or radiotherapy for any other malignancies within the past 5 years, no prior history of invasive breast cancer, and adequate organ function.

This analysis included participants who had not undergone primary tumor resection because primary tumor resection could affect prognosis and obscure differences in treatment effect of systemic therapy. Consequently, the analysis comprised patients who were evaluated as progressive disease (PD) after PST and not eligible for second registration, as well as patients who were enrolled at second registration and assigned to systemic therapy only (arm A) (Fig. 1).

Fig. 1.

Study design of JCOG1017

Treatment and response evaluation

Patients received PST based on their subtype:

• ER-positive, no life-threatening disease: Tamoxifen + goserelin (premenopausal) or letrozole (postmenopausal).

• HER2-positive: Paclitaxel + trastuzumab (before 2014) or docetaxel + trastuzumab + pertuzumab (after 2014).

• HER2-negative, life-threatening disease: Paclitaxel.

Treatment response was assessed approximately three months post-initiation of PST (median, 12 weeks; interquartile range, 10–14 weeks) using both JCOG1017 criteria and RECIST v1.1[6]. In the JCOG1017 criteria, PD was defined as a ≥ 10% increase in tumor size. This differs from the RECIST v1.1 definition of PD, which requires a ≥ 20% increase in the sum of diameters of target lesions, accompanied by an absolute increase of at least 5 mm. The assessments were performed by the investigators at each participating institution without central review.　In addition, the tumor response at the primary and metastatic sites approximately six months of PST (three months after second registration) was evaluated in patients of arm A.

Patients were categorized into four groups based on response at primary and metastatic sites:

1. Discordant group I: Primary non-PD, metastatic PD.

2. Concordant group I: Both primary and metastatic sites PD.

3. Discordant group II: Primary PD, metastatic non-PD.

4. Concordant group II: Both primary and metastatic sites non-PD.

*Discordance proportion = (a + c)/(a + b+c + d)

Primary PD was defined as an increase in the longest diameter of the primary breast tumor by ≥ 10% according to the JCOG1017 criteria. Metastatic PD is defined as target lesion PD or non-target lesion PD or appearance of new lesions. Metastatic non-PD is defined as the absence of PD in target lesions and non-target lesions, with no new lesion appearance. Not evaluable at tumor assessment was excluded from analysis. For patients with bone-dominant disease, PD was defined by the appearance of new bone lesions or unequivocal progression on bone scan or PET/CT, or by progression of measurable soft-tissue components when present.

Statistical analysis

Clinicopathologic characteristics and subtype-specific discordance were analyzed to confirm the distribution of concordant and discordant treatment responses.

OS was defined as the time from the date of first registration to the date of death from any cause. Patients who were alive or lost to follow-up were censored at the date of last confirmation of survival. Survival curves were estimated using the Kaplan–Meier method, and survival probabilities at each year were calculated with corresponding 95% confidence intervals (CIs).

To assess the association between the four predefined response groups (discordant group I, concordant group I, discordant group II, and concordant group II) and OS, hazard ratios (HRs) were estimated using univariable and multivariable Cox proportional hazards regression model. Statistical significance was defined as a two-sided p-value < 0.05.

Covariates included in the univariable and multivariable models were menopausal status (pre- vs. postmenopausal), performance status (PS: 0, 1, 2), presence of visceral metastasis (yes/no), histological subtype (invasive ductal carcinoma vs. special type), clinical T category (cT1–4), clinical N category (cN0–3), and breast cancer subtype based on ER and HER2 status (ER-negative/HER2-negative, ER-positive/HER2-negative, ER-negative/HER2-positive, ER-positive/HER2-positive).

All analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).

Results

Of 570 patients who enrolled in JCOG1017, 315 patients were included in this analysis. This cohort consisted of 110 patients who were evaluated as PD at tumor assessment before second registration and 205 patients who were assigned to arm A (systemic therapy only) after second registration.

A total of 14.0% (44/315) of patients were not evaluable, primarily due to the absence of CT image evaluations at the 3-month assessment. The total number of patients included in the final analysis was 271, which was derived by excluding the 44 unevaluable patients from the initial 315 patients. The overall discordance proportion of treatment response between primary and metastatic sites was 25.1% (68/271). The proportions of each group were 24.7% (67/271) for discordant group I, 9.6% (26/271) for concordant group I, 0.4% (1/271) for discordant group II, and 65.3% (177/271) for concordant group II (Table 1). Subtype-specific discordance proportions were notably higher in luminal (28.9%), triple-negative (25.0%) and luminal-HER2 (22.0%) subtypes compared to HER2-enriched (11.1%), with all subtypes classified based on the primary tumor. Of 68 patients with discordant responses between primary and metastatic sites, 50 (73.5%) were without life-threatening disease and received hormone therapy, 5 (7.4%) had life-threatening disease and received weekly paclitaxel, and 13 (19.1%) were otherwise.

Table 1.

Treatment response at 3 months of primary systemic therapy by subtype

 PD:progressive disease; PST:Primary systemic therapy

Menopausal status did not significantly impact discordance proportions; pre 25.3% (24/95) and post 25.0% (44/176) (Table 2 , 3). However, among the triple-negative subtype, the discordance proportion is much higher in the premenopausal population (45.5%, 5/11) than in the postmenopausal population (7.7%, 1/13). The number of metastatic organs did not significantly impact discordance proportions (one metastatic organ: 24.8% (38/153), two or more metastatic organs: 23.8% (28/118)). For each subtype, there was no difference between one organ and two or more organs. A detailed distribution of metastatic sites, categorized as bone, liver, lung, pleura, brain, and others, is provided in Supplementary Table S1. In a treatment-adjusted sensitivity analysis that included chemotherapy use at initial systemic therapy, the association between response discordance and overall survival remained directionally consistent with the primary analysis (Supplementary Table S2). Baseline distributions of metastatic organ involvement according to response concordance status are summarized in Supplementary Table S3.

Table 2.

Treatment response at 3 months of PST in premenopausal patients

Table 3.

Treatment response at 3 months of PST in postmenopausal patients

 PD:progressive disease; PST:Primary systemic therapy

Median survival times were 3.8 years for discordant group I, 4.6 years for concordant group I, 1.7 years for discordant group II, and 5.7 years for concordant group II (Fig. 2). Because only one patient was classified as discordant group II (primary PD/metastatic non-PD), statistical analysis for this group was not feasible. Therefore, group II was included descriptively but was not interpreted as an independent prognostic subgroup. In the univariable analysis, no differences in prognosis were observed between discordant and concordant groups I (HR, 0.749; 95%CI, 0.432–1.301). However, differences in prognosis between discordant group I and concordant groups II were observed. Concordant group II demonstrated the most favorable prognosis (HR, 0.556; 95% CI, 0.396–0.782). This trend remained significant even after multivariable adjustment using the Cox proportional hazards model (HR, 0.517; 95% CI, 0.359–0.744). Significant prognostic factors identified in the univariable analysis included PS 1 (HR, 1.664; 95% CI, 1.060–2.610), histological subtype (special type) (HR, 1.787; 95% CI, 1.139–2.804), and breast cancer subtypes: luminal type (HR, 0.326; 95% CI, 0.205–0.516), HER2-enriched (HR, 0.127; 95% CI, 0.062–0.258), and luminal-HER2 (HR, 0.158; 95% CI, 0.085–0.294). In the multivariable analysis, significant factors were PS 1 (HR, 1.854; 95% CI, 1.161–2.959), clinical N3 (HR, 2.212; 95% CI, 1.184–4.134), and subtypes: luminal type (HR: 0.233; 95% CI, 0.142–0.381), HER2-enriched (HR: 0.108; 95% CI, 0.051–0.229), and luminal-HER2 (HR, 0.094; 95% CI: 0.047–0.189) (Table 4).

Fig. 2.

Kaplan–Meier curves for overall survival by treatment response group

Table 4.

Univariable and multivariable Cox proportional hazards models for overall survival

Factors		event/N	Univariate analysis		Multivariate analysis
			HR (95% CI)	P value	HR (95% CI)	P value
Menopausal status	premenopausal	58/95	1		1
	postmenopausal	110/176	1.066 (0.775–1.466)	0.694	1.047 (0.738–1.485)	0.797
Performance status (PS)	0	142/235	1		1
	1	22/29	1.664 (1.060–2.610)	0.027	1.854 (1.161–2.959)	0.010
	2	4/7	0.910 (0.336–2.463)	0.852	1.415 (0.427–4.695)	0.570
Visceral metastasis	No	68/107	1		1
	Yes	100/164	1.019 (0.749–1.387)	0.904	1.205 (0.867–1.673)	0.266
pathological tissue	Invasive ductal carcinoma (IDC)	146/242	1		1
	Special type	22/29	1.787 (1.139–2.804)	0.012	1.588 (0.975–2.586)	0.063
cT	cT1	8/15	1		1
	cT2	74/121	1.190 (0.573–2.469)	0.641	1.331 (0.600–2.953.600.953)	0.482
	cT3	28/37	1.709 (0.778–3.752)	0.182	2.269 (0.981–5.244)	0.055
	cT4	58/98	1.178 (0.562–2.468)	0.665	1.372 (0.624–3.015)	0.431
cN	cN0	13/22	1		1
	cN1	70/120	0.995 (0.551–1.799)	0.987	1.451 (0.776–2.714)	0.244
	cN2	25/40	1.237 (0.633–2.419)	0.534	1.784 (0.883–3.601)	0.107
	cN3	60/89	1.351 (0.741–2.461)	0.326	2.212 (1.184–4.134)	0.013
Subtype	ER (-) & HER2 (-)	22/24	1		1
	ER (+) & HER2 (-)	115/170	0.326 (0.205–0.516)	< 0.0001	0.233 (0.142–0.381)	< 0.0001
	ER (-) & HER2 (+)	12/36	0.127 (0.062–0.258)	< 0.0001	0.108 (0.051–0.229)	< 0.0001
	ER (+) & HER2 (+)	19/41	0.158 (0.085–0.294)	< 0.0001	0.094 (0.047–0.189)	< 0.0001
Treatment response between primary and metastatic sites	(a) discordant I	50/67	1		1
	(b) concordant I	17/26	0.749 (0.432–1.301)	0.305	1.106 (0.601–2.035)	0.746
	(c) discordant II	1/1	5.507 (0.743–40.841)	0.095	8.897 (0.797–99.294)	0.076
	(d) concordant II	100/177	0.556 (0.396–0.782)	0.001	0.517 (0.359–0.744)	< 0.0001

 HR: hazard ratio; CI: confidence interval; PD: progressive disease; ER: estrogen receptor; HER2: human epidermal growth factor receptor 2; PST: primary systemic therapy

The Kaplan-Meier curve for OS is similar based on both evaluation by JCOG1017 criteria and RECIST v1.1 (Fig. 3).

Fig. 3.

Overall survival by PD versus non-PD status (JCOG1017 criteria)

In the evaluation at about six months of PST (three months after second registration) in patients of arm A. Using JCOG1017 criteria, only one case of luminal type was identified in the discordant group I (Table 5). This observation indicates that in patients who were eligible for second registration due to effective PST, treatment response at about six months after initiation of PST was almost entirely both primary and metastatic site non-PD.

Table 5.

Treatment response at 3 months of PST in postmenopausal patients

 PD:progressive disease; PST:Primary systemic therapy

Discussion

This exploratory analysis of JCOG1017 revealed that approximately one-fourth of patients with de novo stage IV breast cancer exhibited discordant treatment responses between primary and metastatic sites in the early phase of PST. Most discordant cases involved non-PD status in the primary tumor but PD in metastatic lesions. This discordance was particularly frequent in luminal and triple-negative breast cancer subtypes, whereas it was relatively uncommon in HER2-enriched tumors.

These findings support the clinical response heterogeneity between primary and metastatic lesions previously reported in breast cancer[1–3]. Treatment decisions in de novo stage IV breast cancer are often based on the biology of the primary tumor, as it is more readily accessible for biopsy. However, our results suggest that such an approach may overlook important changes in metastatic lesions that could influence treatment response. This is particularly relevant for hormone receptor–positive tumors, where endocrine therapy may be insufficient for controlling metastatic disease despite apparent control of the primary site.

Notably, patients with discordant responses—specifically discordant group I—tended to have worse OS compared to concordant group II. This difference in prognosis remained significant even after adjustment for known clinical and pathological prognostic factors. This indicates that even among patients whose primary tumor remained controlled (primary non-PD), those with progressive metastatic disease had substantially worse survival compared with those without metastatic progression. In contrast, cases of primary PD with non-PD metastases were extremely rare, precluding meaningful analysis. Only one case was categorized as discordant group II. Because of this extremely small sample size, no statistical comparisons were performed for this group. This limitation should be considered when interpreting the overall group comparisons. Therefore, our findings suggest that assessment of metastatic sites carries major prognostic weight, but evaluation of both primary and metastatic sites remains necessary in clinical practice. It should also be noted that patients in arm B of JCOG1017 were not eligible for this analysis because the primary tumor had been surgically removed, making assessment of primary-site response inherently impossible. Therefore, the analytic cohort reflects the population in whom treatment response at both primary and metastatic sites can be objectively evaluated. These findings highlight the potential clinical impact of response discordance and reinforce the importance of evaluating metastatic lesions independently, rather than relying solely on the primary tumor[7].

HER2-enriched tumors showed the lowest discordance proportion, suggesting a more clinically consistent treatment response pattern to anti-HER2 therapy across disease sites. This does not necessarily indicate biological concordance. As biopsy of metastatic sites was optional in this study, pathological confirmation of receptor status concordance could not be performed. This observation may partly explain the lower discordance proportion in HER2-enriched tumors compared with luminal or triple-negative subtypes.

However, because metastatic site biopsy and receptor reassessment were not mandated in JCOG1017, biological discordance between primary and metastatic lesions could not be directly evaluated. Therefore, the discordant or concordant clinical responses observed in this study should be interpreted as differences in treatment response patterns rather than evidence of underlying biological concordance or heterogeneity.

Interestingly, within the triple-negative subtype, a higher discordance proportion was observed in premenopausal patients. Although the sample size was limited, this may reflect underlying age-related biological differences and warrants further investigation in future studies.

From a clinical perspective, these results emphasize the need for careful monitoring of both primary and metastatic sites when evaluating treatment response in de novo stage IV breast cancer. In particular, patients with discordant responses may require earlier consideration of treatment modification, even if the primary tumor appears stable. This is especially important given that RECIST v1.1 guidelines combine primary and metastatic sites into a single response evaluation, potentially obscuring site-specific progression. In this analysis, patients showing discordant treatment responses between the primary and metastatic sites had poorer outcomes. Particularly when the primary tumor remains controlled but metastatic lesions progress, early reassessment of metastatic disease should be performed. Treatment strategies may include switching to novel agents such as molecular targeted therapies, CDK4/6 inhibitors, or antibody–drug conjugates, combining endocrine therapy with chemotherapy, and considering local treatment for progressing metastatic sites.

This is consistent with findings from the EA2108 trial, which also questioned the survival benefit of primary tumor resection in triple-negative breast cancer patients[8].

This study has several limitations. First, the sample size of discordant group II was extremely small, limiting statistical analysis. Second, biopsy and receptor reassessment of metastatic sites were not mandatory in JCOG1017, and therefore biological differences between primary and metastatic lesions could not be evaluated. The observed response concordance and discordance reflect clinical treatment response rather than underlying clinical response heterogeneity. Third, treatment response evaluation was based on investigator assessment without central imaging review, which may have introduced variability in interpretation. Fourth, organ-specific survival analyses could not be performed because patients who experienced progressive disease after the initial systemic therapy were not followed longitudinally in JCOG1017. As a result, OS data for these patients were not available, precluding site-specific prognostic comparisons.

Despite these limitations, our findings provide important insight into the clinical significance of treatment response discordance in de novo stage IV breast cancer. Future advances in response evaluation, such as the use of liquid biopsy and advanced imaging techniques, may enable earlier and more accurate detection of site-specific tumor heterogeneity and progression. Incorporating these modalities into clinical trials may help refine treatment strategies for patients with de novo stage IV breast cancer.

Conclusions

In de novo stage IV breast cancer, discrepancies in treatment response between primary and metastatic sites were not uncommon and were associated with poorer outcomes. These findings highlight the importance of evaluating both primary and metastatic lesions when determining systemic therapy.

Supplementary Information

Below is the link to the electronic supplementary material.

Funding

Open Access funding provided by Okayama University. the National Cancer Center Research and Development Fund: 26-A-4, 29-A-3, 2020-J-3, 2023-J-03, AMED: JP22ck0106593, JP19ck0106307, JP16ck106046

Declarations

Conflict of interest

Dr. Shien reported grants from the Ministry of Health, Labour and Welfare of Japan and from the Japan Agency for Medical Research and Development (AMED) during the conduct of the study, as well as personal fees from Pfizer, Chugai, Lilly, Daiichi Sankyo, MSD, Eizai, AstraZeneca, Taiho, Nihonkayaku, Gilead. Dr. Hara reported the personal fees from Pfizer, Ili Lilly, Daiichi Sankyo, Kyowa Kirin, Chugai, MSD. Dr. Ogawa reported the grants from Japan Agency for Medical Research and Development and Ministry of Health, Labour and Welfare, Japan. Dr. Fukuda reported the grants from National Cancer Center and AstraZeneca, Ono Pharmaceutical Co., Ltd, Chugai Pharmaceutical Co., Ltd, Takeda Pharmaceutical Company Limited and the personal fees from Chugai Pharmaceutical Co., Ltd., Kyowa Kirin Co., Ltd., CMIC Co., Ltd. Dr. Iwata reported the grants from Pfizer and Chugai, Daiichi Sankyo, AstraZeneca and personal fees from Daiichi Sankyo, Chugai, AstraZeneca, Lilly, MSD, Pfizer, Taiho, Kyowa Kirin. No other disclosures were reported.

Ethical approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed consent

Informed consent including the secondary use of data was obtained from all individual participants enrolled in JCOG1017.