Prognostic value of pre-treatment serum CA19-9 and lymphocyte-to-monocyte ratio in HR+/HER2- breast cancer: a retrospective cohort study

Abstract

Background

The prognostic significance of carbohydrate antigen 19 − 9 (CA19-9) and the lymphocyte-to-monocyte ratio (LMR) in hormone receptor–positive/human epidermal growth factor receptor 2–negative (HR+/HER2-) breast cancer remains unclear. This study aimed to evaluate their value as pre-treatment prognostic markers.

Methods

We retrospectively analyzed 349 h+/HER2- breast cancer patients who underwent surgery between 2011 and 2012. Patients with other malignancies or incomplete follow-up were excluded. Serum CA19-9 levels and LMR were measured before any treatment, and optimal cut-offs were determined using receiver operating characteristic (ROC) curves.

Results

ROC analysis identified cut-offs of 27.54 U/mL for CA19-9 and 5.34 for LMR. Patients with CA19-9 ≤ 27.54 U/mL had significantly longer OS, and patients with LMR > 5.34 also had significantly longer OS (P = 0.019 and P = 0.021, respectively). Multivariate Cox analysis identified N stage and high pre-treatment CA19-9 as independent predictors of OS, whereas LMR, age, tumor grade, Ki-67, menopausal status, and tumor size were not independently associated with OS. A combined risk score incorporating CA19-9 and LMR was associated with OS stratification within this cohort, with the low-risk group showing more favorable OS (P < 0.05). Nomogram analysis integrating CA19-9, LMR, and clinicopathologic factors showed moderate discrimination in internal validation.

Conclusions

High pre-treatment CA19-9 was independently associated with worse overall survival, whereas the prognostic value of LMR was limited to univariate analysis. The combined assessment of CA19-9 and LMR may provide supplementary prognostic information, although its incremental value beyond established clinicopathologic factors appears limited and requires further validation in independent cohorts.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-026-15941-3.

Introduction

HR+/HER2- breast cancer is the most common molecular subtype, accounting for 60–70% of malignant breast neoplasms [1–3]. Endocrine therapy is the standard systemic treatment for this subtype. Although it is generally considered to have a relatively favorable prognosis, a substantial proportion of patients still experience recurrence, particularly within the first few years after treatment initiation [4–9]. Therefore, identifying practical and reliable prognostic indicators remains clinically important.

Early relapse in HR+/HER2 − breast cancer is often associated with aggressive tumor behavior and potential endocrine resistance [10, 11]. Both tumor-related biological characteristics and host-related immune-inflammatory status have been implicated in disease progression and treatment response. Increasing evidence suggests that systemic immune-inflammatory responses may influence tumor progression, recurrence, and survival outcomes [12, 13]. Recent studies have highlighted the involvement of diverse molecular and therapeutic mechanisms in breast cancer progression and treatment response, including microRNA-mediated signaling pathways, immune checkpoint–targeted therapies, and other regulators of tumor biology [14, 15]. In addition, emerging evidence regarding tumor-associated molecular regulators and potential therapeutic compounds further illustrates the biological complexity of breast cancer and underscores the need for practical and accessible prognostic biomarkers [16, 17]. Consequently, readily available blood-based inflammatory markers have attracted attention as potential prognostic tools.

Several prognostic assays, such as the 21-gene recurrence score (Oncotype DX) and the 70-gene expression signature (MammaPrint), are widely used to guide adjuvant treatment decisions in selected patients [18–20]. However, these gene-expression–based tests are relatively costly and not universally accessible. Therefore, simple blood-based biomarkers may provide a practical complement to existing tools.

The lymphocyte-to-monocyte ratio (LMR), reflecting systemic immune status, has been reported to be associated with survival outcomes in breast cancer, with lower values linked to poorer prognosis [21]. Cancer antigen 19 − 9 (CA19-9), a serum tumor-associated marker primarily used in gastrointestinal malignancies, has occasionally been evaluated in breast cancer patients, although its clinical relevance in this setting remains uncertain [22]. Notably, evidence regarding the prognostic significance of these markers in HR+/HER2- breast cancer is still limited and inconsistent.

In real-world clinical practice, serum tumor markers and routine blood parameters are sometimes measured as part of baseline evaluation or follow-up in certain institutions. However, whether CA19-9 and LMR provide meaningful prognostic information in HR+/HER2- breast cancer remains unclear.

Accordingly, this retrospective study aimed to evaluate the association of baseline LMR and CA19-9 levels with clinicopathological characteristics and survival outcomes in patients with HR+/HER2- breast cancer, and to explore their potential prognostic value, both individually and in combination.

Methods

Study cohort

We retrospectively included 349 patients with HR+/HER2- breast cancer who underwent surgery at Sun Yat-sen University Cancer Center between January 2011 and December 2012.

The exclusion criteria were as follows: (I) patients without HR+/HER2- breast cancer; (II) patients with a history of malignant disease or other primary cancer; (III) patients with missing data or loss to follow-up. The final study population consisted of 349 patients.

Biomarker assessment

The hematological and biochemical profiles of HR+/HER2- breast cancer patients, including serum CA19-9 and the lymphocyte-to-monocyte ratio (LMR), were assessed prior to any treatment. Serum CA19-9 was routinely measured in all patients at our institution during the study period as part of baseline evaluation, and the LMR was calculated from pre-treatment complete blood counts. Receiver operating characteristic (ROC) curve analysis based on overall survival (OS) was performed to determine the optimal cut-off values for CA19-9 and LMR using the Youden index. The optimal cut-off values were 27.54 U/mL for CA19-9 and 5.34 for LMR.

In addition, the ROC-derived cut-offs were used in the primary survival analyses. Patients were then categorized according to these cut-offs, and Kaplan-Meier survival curves were generated to compare outcomes between groups. A prognostic risk score was constructed using regression coefficients derived from the Cox proportional hazards model, and patients were stratified into high- and low-risk groups according to the median risk score for survival analysis.

Postoperatively, patients were followed up every 3 months during the first 2 years, every 6 months during years 3 to 5, and annually thereafter. Follow-up was conducted from diagnosis until death or December 2023 through review of medical records and telephone interviews with patients or their families. The median follow-up time was 11.59 years.

Postoperative adjuvant therapy

Information on postoperative adjuvant therapy was retrospectively collected from medical records. Postoperative adjuvant treatments included endocrine therapy, chemotherapy, and radiotherapy. Endocrine therapy and chemotherapy were administered according to standard clinical practice at the time of treatment, while radiotherapy was delivered based on pathological stage and established clinical indications.

Statistical analysis

For the main multivariate analyses, clinicopathological variables were dichotomized according to clinically established or commonly used cut-off values. To minimize potential information loss associated with dichotomization, additional univariate analyses using multi-category variables were performed and are provided in the Supplementary Materials.

The primary endpoint was overall survival (OS), and the secondary endpoint was disease-free survival (DFS), with detailed definitions provided according to the STEEP system [23]. OS was defined as the interval from the start of observation or randomization to death from any cause, whereas DFS was defined as the time from the start of observation to the first documented tumor recurrence or death from any cause. Categorical variables were compared using the χ² test or Fisher’s exact test, as appropriate. Continuous variables were assessed for normality using the Shapiro–Wilk test. Normally distributed variables were compared using the independent-samples t-test, and non-normally distributed variables were compared using the Mann–Whitney U test. Survival analyses were performed using the predefined ROC-derived cut-offs. Kaplan–Meier survival analysis and log-rank tests were applied to compare OS and DFS between groups. Univariate and multivariate Cox proportional hazards regression models were used to identify independent prognostic factors. Adjuvant chemotherapy was excluded from the OS multivariate model because of sparse events leading to non-estimable hazard ratios. Survival curves and statistical analyses were performed using IBM SPSS Statistics version 17.0 (IBM Corp., Armonk, NY, USA) and R version 3.5.3 (R Foundation for Statistical Computing, Vienna, Austria). For the prognostic nomogram, internal validation was performed using bootstrap resampling (1,000 repetitions) to estimate the optimism-corrected C-index and generate the calibration curve. A P value < 0.05 was considered statistically significant. Proportional hazards assumptions were evaluated using Schoenfeld residuals (cox.zph test). No significant violation was detected (global P = 0.059). Details are provided in Table S1 and Figure S1.

Ethics approval and consent to participate

This study was approved by the Institutional Review Board of Sun Yat-sen University Cancer Center (IRB number: B2025-823-01), and the requirement for informed consent was waived due to the retrospective analysis of de-identified patient data. All procedures performed in this study involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the Declaration of Helsinki.

Results

Patients’ baseline characteristics and overall survival

Table 1 summarizes the baseline clinicopathological characteristics and follow-up distribution of the study cohort. A total of 349 patients were included in the present analysis. The median follow-up time was 11.59 years. ROC curve analysis, combined with the Youden index (Youden index = sensitivity + specificity − 1), was employed to determine optimal cut-off values, which were identified as 27.54 U/mL for pre-treatment CA19-9 and 5.34 for pre-treatment LMR (Fig. 1). The corresponding area under the ROC curve (AUC) was 0.55 (95% CI: 0.40–0.69) for CA19-9 and 0.60 (95% CI: 0.49–0.70) for LMR, indicating relatively modest discriminative ability (Fig. 1A, B). Given the limited predictive performance of the individual markers, a combined model integrating both pre-treatment CA19-9 and LMR was constructed to explore whether discrimination could be modestly improved. ROC analysis showed that the combined model achieved a higher AUC of 0.64 (95% CI: 0.52–0.76) than either marker alone, suggesting modest improvement in discriminative performance within this cohort.

Table 1.

Baseline clinicopathological characteristics of patients with HR+/HER2- breast cancer (N = 349)

Characteristics	Total (n = 349)
Age, years
≤ 50	287(82.23%)
> 50	62(17.77%)
Menopausal status
Postmenopausal	151(43.27%)
Premenopausal	198(56.73%)
Grade
Unknown	18(5.16%)
I	14(4.01%)
II	281(80.52%)
III	36(10.32%)
Tumor size
≤ 2 cm	164(46.99%)
2–5 cm	171(48.99%)
> 5 cm	14(4.01%)
N stage
N0	189(54.15%)
N1	94(26.93%)
N2	43(12.32%)
N3	23(6.59%)
Ki-67
≤ 15%	166(47.56%)
> 15%	183(52.44%)
CA19-9
Pre-treatment CA19-9 ≤ 27.54 U/mL	302(86.53%)
Pre-treatment CA19-9 > 27.54 U/mL	47(13.47%)
LMR
Pre-treatment LMR ≤ 5.34	200(57.31%)
Pre-treatment LMR > 5.34	149(42.69%)
Endocrine therapy
Yes	333(95.42%)
No	16(4.58%)
Chemotherapy
Yes	304(87.11%)
No	45(12.89%)
Radiotherapy
Yes	60(17.19%)
No	289(82.81%)
Overall survival
0–2 years	4(1.15%)
2–5 years	14(4.01%)
> 5 years	331(94.84%)
Disease-free survival
0–2 years	16(4.58%)
2–5 years	24(6.88%)
> 5 years	309(88.54%)

Fig. 1.

Receiver operating characteristic (ROC) curves for serum carbohydrate antigen 19 − 9 (CA19-9), lymphocyte-to-monocyte ratio (LMR), and their combination in predicting overall survival (OS). ROC analysis was conducted to evaluate the discriminative ability of each biomarker. A The area under the ROC curve (AUC) for CA19-9 was 0.55 (95% CI: 0.40–0.69). B The AUC for LMR was 0.60 (95% CI: 0.49–0.70); (C) The combined model incorporating CA19-9 and LMR showed modestly improved discriminative performance, with an AUC of 0.64 (95% CI: 0.52–0.76)

Association of pre-treatment CA19-9 and LMR with clinicopathological characteristics

Patient clinicopathological characteristics and the cut-off values of CA19-9 and LMR are summarized in Table 2. CA19-9 levels were significantly associated with overall survival status (alive vs. dead), pre-treatment LMR was significantly associated with N stage, overall survival status (alive vs. dead), adjuvant endocrine therapy, and radiotherapy (all P < 0.05). No other clinical characteristics were significantly correlated with CA19-9 or LMR (P > 0.05). The median follow-up time for the cohort was 11.59 years. Kaplan-Meier analysis showed significantly longer OS in patients with lower CA19-9 levels and higher LMR values.

Table 2.

Association between clinicopathological characteristics and pre-treatment CA19-9 and LMR levels in patients with HR+/HER2- breast cancer

Demographics		The level of LMR			The level of CA19-9
		≤ 5.34	> 5.34	P	≤ 27.54 U/mL	> 27.54 U/mL	P
Number (n)		200	149		302	47
Age (years), n (%)	< 50	161 (80.5)	121 (81.2)	0.977	244 (80.8)	38 (80.9)	1.000
	> 50	39 (19.5)	28 (18.8)		58 (19.2)	9 (19.1)
T stage, n (%)	T1	100 (50.0)	64 (43.0)	0.081	144 (47.7)	20 (42.6)	0.531
	T2	96 (48.0)	75 (50.3)		147 (48.7)	24 (51.1)
	T3	4 (2.0)	8 (5.4)		9 (3.0)	3 (6.4)
	T4	0 (0.0)	2 (1.3)		2 (0.7)	0 (0.0)
N stage, n (%)	N0	96 (48.0)	93 (62.4)	0.004	165 (54.6)	24 (51.1)	0.673
	N1	54 (27.0)	40 (26.8)		79 (26.2)	15 (31.9)
	N2	31 (15.5)	12 (8.1)		39 (12.9)	4 (8.5)
	N3	19 (9.5)	4 (2.7)		19 (6.3)	4 (8.5)
Grade, n (%)	Unknown	10 (5.0)	8 (5.4)	0.893	12 (4.0)	6 (12.8)	0.088
	G1	9 (4.5)	5 (3.4)		12 (4.0)	2 (4.3)
	G2	162 (81.0)	119 (79.9)		246 (81.5)	35 (74.5)
	G3	19 (9.5)	17 (11.4)		32 (10.6)	4 (8.5)
Overall survival status, n (%)	alive	174 (87.0)	140 (94.0)	0.046	276 (91.4)	38 (80.9)	0.048
	death	26 (13.0)	9 (6.0)		26 (8.6)	9 (19.1)
Adjuvant endocrine therapy, n (%)	Yes	195 (97.5)	138 (92.6)	0.039	287 (95.0)	46 (97.9)	0.071
	No	5 (2.5)	11 (7.4)		15 (5.0)	1 (2.1)
Adjuvant chemotherapy, n (%)	Yes	176 (88.0)	128 (85.9)	0.629	259 (85.8)	45 (95.7)	0.061
	No	24 (12.0)	21 (14.1)		43 (14.2)	2 (4.3)
Radiotherapy, n (%)	Yes	55 (27.5)	5 (3.4)	< 0.001	55 (18.2)	5 (10.6)	0.297
	No	145 (72.5)	144 (96.6)		247 (81.8)	42 (89.4)

P values were calculated using Pearson’s χ² test or Fisher’s exact test, as appropriate

In the univariate Cox regression analyses, higher tumor grade (HR = 2.28, 95% CI: 1.15–4.52, P = 0.019), lymph node metastasis (HR = 1.89, 95% CI: 1.40–2.55, P < 0.001), Ki-67 > 15% (HR = 2.39, 95% CI: 1.15–4.97, P = 0.020), and pre-treatment CA19-9 > 27.54 U/mL (HR = 2.81, 95% CI: 1.09–7.25, P = 0.032) were significantly associated with poorer OS, whereas pre-treatment LMR > 5.34 was associated with reduced mortality (HR = 0.36, 95% CI: 0.15–0.88, P = 0.026). Age, menopausal status, tumor size, adjuvant endocrine therapy, adjuvant chemotherapy, and radiotherapy showed no significant associations with OS (P > 0.05) (Table 3).

Table 3.

Univariate Cox regression analysis of clinicopathological factors associated with OS and DFS

Variable	Coefficient	HR	95% CI	P	Coefficient	HR	95% CI	P
	OS				DFS
Age
> 50 vs. ≤ 50 years	0.233	1.26	0.57–2.78	0.563	0.175	1.19	0.71–2.01	0.512
Menopausal status
Postmenopausal vs. Premenopausal	0.237	1.27	0.65–2.46	0.484	-0.156	0.86	0.55–1.33	0.485
Grade
III vs. I–II	0.822	2.28	1.15–4.52	0.019	0.563	1.76	1.13–2.72	0.012
Tumor size
> 2 cm vs. ≤ 2 cm	0.322	1.38	0.82–2.33	0.228	0.441	1.55	1.11–2.17	0.010
Lymph node metastasis
Node-positive vs. Node-negative	0.638	1.89	1.40–2.55	< 0.001	0.523	1.69	1.38–2.06	< 0.001
Ki-67
> 15% vs. ≤ 15%	0.871	2.39	1.15–4.97	0.020	0.560	1.75	1.12–2.74	0.014
CA19-9
> 27.54 U/mL vs. ≤ 27.54 U/mL	1.034	2.81	1.09–7.25	0.032	0.356	1.43	0.75–2.73	0.282
LMR
> 5.34 vs. ≤ 5.34	-1.027	0.36	0.15–0.88	0.026	-0.430	0.65	0.40–1.05	0.080
Adjuvant endocrine therapy
Yes vs. No	0.486	1.63	0.22–11.88	0.604	0.648	1.91	0.47–7.77	0.314
Adjuvant chemotherapy
Yes vs. No	0.939	2.56	0.61–10.66	0.137	0.901	2.46	0.99–6.08	0.025
Radiotherapy
Yes vs. No	0.532	1.70	0.80–3.63	0.189	0.260	1.30	0.76–2.21	0.352

P values were calculated using univariate Cox proportional hazards models. The first category was used as the reference group

Regarding DFS, higher tumor grade (HR = 1.76, 95% CI: 1.13–2.72, P = 0.012), larger tumor size (HR = 1.55, 95% CI: 1.11–2.17, P = 0.010), lymph node metastasis (HR = 1.69, 95% CI: 1.38–2.06, P < 0.001), and Ki-67 > 15% (HR = 1.75, 95% CI: 1.12–2.74, P = 0.014) were associated with an increased risk of recurrence. Pre-treatment CA19-9 and pre-treatment LMR were not significantly associated with DFS, although pre-treatment LMR demonstrated a borderline association (HR = 0.65, 95% CI: 0.40–1.05, P = 0.080). Adjuvant endocrine therapy and radiotherapy were not significantly associated with DFS. In contrast, adjuvant chemotherapy was associated with an increased risk of recurrence in univariate analysis (HR = 2.46, 95% CI: 0.99–6.08, P = 0.025), likely reflecting confounding by indication (Table 3).

Survival analysis of serum levels of CA19-9 and LMR in HR+/HER2- breast cancer

Kaplan-Meier analyses showed that patients with CA19-9 ≤ 27.54 U/mL had significantly higher OS compared with those with CA19-9 > 27.54 U/mL (5-year OS: 96.8%, 95% CI: 94.5–99.2% vs. 85.3%, 95% CI: 74.2–98.1%; χ² = 5.47, P = 0.019; Fig. 2A, B). Similarly, patients with pre-treatment LMR > 5.34 exhibited improved OS relative to those with LMR ≤ 5.34 (5-year OS: 96.0% vs. 93.5%; χ² = 5.35, P = 0.021; Fig. 3A, B).

Fig. 2.

Kaplan-Meier survival curves according to the CA19-9 cut-off value. A OS in patients stratified by CA19-9 levels (log-rank χ² = 5.47, P = 0.019); (B) DFS in patients stratified by CA19-9 levels (log-rank χ² = 1.49, P = 0.22)

Fig. 3.

Kaplan-Meier survival curves according to the LMR cut-off value. A OS in patients stratified by LMR levels (log-rank χ² = 5.35, P = 0.021); (B) DFS in patients stratified by LMR levels (log-rank χ² = 3.42, P = 0.065)

Patients were further stratified into high- and low-risk groups based on a risk score derived from the Cox regression model (see Methods), and Kaplan-Meier analysis showed that patients in the low-risk group had significantly better OS and DFS than those in the high-risk group (Fig. 4A, B). Combined analysis showed that patients with CA19-9 ≤ 27.54 U/mL and LMR > 5.34 had the most favorable OS within this cohort (Table 4).

Fig. 4.

Kaplan-Meier survival curves according to risk stratification based on the Cox regression model. A OS in patients stratified into high- and low-risk groups according to the Cox risk score (log-rank P = 0.013); (B) Disease-free survival (DFS) in patients stratified into high- and low-risk groups according to the Cox risk score (log-rank P = 0.048)

Table 4.

Association of CA19-9 and LMR groups with mortality and recurrence status

Variable	CA19-9 ≤ 27.54 U/mL	CA19-9 > 27.54 U/mL	Total	X²	P	LMR ≤ 5.34	LMR > 5.34	Total	X²	P
Overall survival				5.410	0.028				6.704	0.015
Survival	276	38	314			174	140	314
Dead	26	9	35			26	9	35
Total (n)	302	47	349			200	149	349
Mortality rate (%)	8.61	19.15				13.00	6.04
Recurrence				1.065	0.37				3.559	0.059
No	227	32	259			142	119	261
Yes	75	15	90			58	30	88
Total (n)	302	47	349			200	149	349
Recurrence rate (%)	24.83	31.91				29.00	20.13

P values in Table 4 were calculated using χ² tests based on status distribution, whereas survival differences in Figs. 2, 3 and 4 were assessed using the log-rank test

To assess whether the prognostic value of pre-treatment CA19-9 and LMR was influenced by postoperative adjuvant therapies, stratified Kaplan-Meier analyses were performed according to postoperative endocrine therapy, chemotherapy, and radiotherapy. Among patients receiving postoperative endocrine therapy, elevated CA19-9 and low LMR were associated with significantly worse OS (P = 0.021 and P = 0.012, respectively). In patients receiving postoperative chemotherapy, high CA19-9 (P = 0.047) and low LMR (P = 0.022) were both associated with poorer OS. Among patients who did not receive postoperative radiotherapy, high CA19-9 remained significantly associated with reduced OS (P = 0.012), whereas low LMR showed a similar trend that did not reach statistical significance (P = 0.095), possibly due to the small sample size (Fig. 5).

Fig. 5.

Kaplan-Meier survival curves for CA19-9 and LMR (high vs. low) stratified by postoperative therapy. A OS stratified by CA19-9 levels in patients receiving postoperative endocrine therapy (P = 0.021); (B) OS stratified by LMR levels in patients receiving postoperative endocrine therapy (P = 0.012); (C) OS stratified by CA19-9 levels in patients receiving postoperative adjuvant chemotherapy (P = 0.047); (D) OS stratified by LMR levels in patients receiving postoperative adjuvant chemotherapy (P = 0.022); (E) OS stratified by CA19-9 levels in patients not receiving postoperative radiotherapy (P = 0.012); (F) OS stratified by LMR levels in patients not receiving postoperative radiotherapy (P = 0.095). Notes: High vs. low groups were defined based on the optimal cut-off values of CA19-9 and LMR determined by ROC analysis. The survival differences observed are consistent across therapy subgroups, suggesting that the observed associations are unlikely to be entirely explained by postoperative treatment differences

Collectively, these exploratory subgroup analyses should be interpreted cautiously given the limited subgroup sample sizes and number of events.

Cox regression analysis of serum CA19-9 and LMR in HR+/HER2- breast cancer

Cox regression analyses were conducted to evaluate the associations of serum CA19-9 and LMR with overall survival (OS) and disease-free survival (DFS), adjusting for potential confounding factors. In univariate analyses, serum CA19-9 (> 27.54 U/mL) was significantly associated with poorer OS (HR = 2.81, 95% CI: 1.09–7.25, P = 0.032), but not with DFS (HR = 1.43, 95% CI: 0.75–2.73, P = 0.282). Higher LMR (> 5.34) was significantly associated with improved OS (HR = 0.36, 95% CI: 0.15–0.88, P = 0.026), with a borderline association for DFS (HR = 0.65, 95% CI: 0.40–1.05, P = 0.080). Among other factors, nodal metastasis (N stage), elevated Ki-67, and higher tumor grade were significantly associated with poorer OS and DFS in univariate analysis. Age, menopausal status, and most adjuvant therapies were not significantly associated with OS or DFS. Tumor size (> 2 cm) was significantly associated with worse DFS in univariate analysis (HR = 1.55, 95% CI: 1.11–2.17, P = 0.010). However, adjuvant chemotherapy was associated with worse DFS in univariate analysis.

Multivariate Cox regression analysis was performed adjusting for Tumor size, N stage, age, pre-treatment CA19-9, pre-treatment LMR, tumor grade, Ki-67 and menopausal status. N stage remained an independent predictor of both OS and DFS (OS: HR = 1.86, 95% CI 1.22–2.84, P = 0.004; DFS: HR = 1.72, 95% CI 1.33–2.22, P < 0.001). Pre-treatment CA19-9 was independently associated with OS (HR = 2.66, 95% CI 1.01–7.04, P = 0.048), but not DFS. Other variables, including tumor size, LMR, age, tumor grade, Ki-67, and menopausal status, were not independently associated with OS or DFS, although tumor size showed a borderline association with DFS (HR = 1.67, 95% CI 0.99–2.79, P = 0.052). Overall, nodal status demonstrated the strongest prognostic impact in this cohort (Table 5). The proportional hazards assumption was verified using Schoenfeld residuals, and no significant violation was detected (global test P = 0.059; all covariates P > 0.05), supporting the validity of the Cox model. Restricted cubic spline analysis demonstrated a significant association between continuous CA19-9 and overall survival (P = 0.023), whereas no evidence of non-linearity was observed (P for non-linearity = 0.1688), suggesting a predominantly monotonic relationship (Figure S2A). In contrast, continuous LMR was not significantly associated with overall survival, and no evidence of non-linearity was detected (Figure S2B).

Table 5.

Multivariate Cox regression analysis of prognostic factors influencing OS and DFS

Variables	HR (OS)	95% CI (OS)	P (OS)	HR (DFS)	95% CI (DFS)	P (DFS)
Tumor size (> 2 cm vs. ≤ 2 cm)	1.20	0.51–2.82	0.683	1.67	0.99–2.79	0.052
N stage (node-positive vs. node-negative)	1.86	1.22–2.84	0.004	1.72	1.33–2.22	< 0.001
Age (> 50 vs. ≤ 50 years)	0.58	0.16–2.14	0.416	1.84	0.76–4.46	0.180
Pre-treatment CA19-9 (> 27.54 U/mL vs. ≤ 27.54 U/mL)	2.66	1.01–7.04	0.048	1.42	0.73–2.75	0.304
Pre-treatment LMR > 5.34 vs. ≤ 5.34	0.57	0.20–1.64	0.296	0.94	0.55–1.63	0.838
Tumor grade (III vs. I–II)	1.67	0.71–3.90	0.238	1.36	0.82–2.26	0.236
Ki-67 > 15% vs. ≤ 15%	1.87	0.73–4.78	0.190	1.53	0.91–2.58	0.109
Menopausal status (postmenopausal vs. premenopausal)	0.58	0.16–2.14	0.415	0.49	0.23–1.01	0.054

Nomogram for predicting prognosis in HR+/HER2- breast cancer

The nomogram developed to predict prognosis in HR+/HER2- breast cancer integrates key clinical and biological factors, including lymph node (N) stage, patient age, histological grade, pre-treatment serum CA19-9 and LMR (Fig. 6). The nomogram demonstrated moderate discriminative ability, with an optimism-corrected C-index of 0.79 based on bootstrap internal validation. These variables were selected based on their associations with survival outcomes in univariate and multivariate analyses. In the nomogram, each factor is assigned a specific point value, and the cumulative total generates a risk score that can be used to estimate the probability of survival for individual patients. Higher pre-treatment CA19-9 levels were associated with higher estimated risk in the model, reflecting potential tumor aggressiveness or systemic tumor burden. Similarly, lower pre-treatment LMR, a marker of systemic immune status and inflammation, was associated with higher estimated risk. Advanced N stage and older age were major contributors to reduced survival probabilities.

Fig. 6.

Nomogram for predicting survival risk in patients with HR+/HER2- breast cancer. The nomogram comprises eight rows. The top row (“Points”) indicates the point assignment for each variable included in the model. Rows 2–6 represent the five prognostic variables. The total score is calculated by summing the points for all variables in the “Total Points” row (Row 7), with higher total points indicating a higher predicted risk of poor survival outcomes. The bottom row (Row 8) shows the predicted survival probability corresponding to the total score. Abbreviations: CA19-9, carbohydrate antigen 19 − 9; LMR, lymphocyte-to-monocyte ratio

Overall, the nomogram identified higher estimated risk in patients with elevated pre-treatment CA19-9, lower LMR, higher histological grade, and more advanced N stage. The calibration curve showed reasonable agreement between predicted and observed overall survival probabilities (Fig. 7). Decision curve analysis showed that the full model provided higher net benefit than the treat-all and treat-none strategies within parts of the evaluated threshold range; however, these findings are based on internal validation only and should be interpreted cautiously without implying established clinical utility (Fig. 8).

Fig. 7.

Calibration curve of the prognostic nomogram for overall survival. The calibration curve shows the agreement between the predicted and observed probabilities of overall survival. The dashed line represents the ideal prediction, while the solid line represents the bias-corrected estimate obtained by bootstrap resampling

Fig. 8.

Decision curve analysis of the prognostic model. The full model showed higher net benefit than the treat-all and treat-none strategies across parts of the evaluated threshold range in internal validation

Discussion

This retrospective cohort study evaluated the prognostic significance of pre-treatment serum CA19-9 and the lymphocyte-to-monocyte ratio (LMR) in patients with HR+/HER2 − breast cancer. We observed that elevated CA19-9 and decreased LMR were significantly associated with poorer OS in univariate analyses, whereas their associations with DFS were weaker and did not reach statistical significance. Using ROC analysis based on OS as the primary endpoint, optimal cut-off values were determined for risk stratification. However, after adjustment for clinicopathologic factors, only CA19-9 remained independently associated with OS, whereas neither biomarker showed independent prognostic significance for DFS. These findings suggest that systemic inflammatory and tumor-associated biomarkers may provide complementary prognostic information, although their independent contribution appears limited.

Although CA19-9 is not routinely recommended in standard breast cancer management, it was consistently measured at our institution during the study period, allowing retrospective evaluation of its prognostic relevance. Biologically, CA19-9 is a sialylated Lewis antigen involved in tumor cell adhesion and tumor-endothelial interactions, and elevated levels may reflect tumor burden, metastatic potential, or tumor-associated inflammatory responses. CA19-9 is a carbohydrate antigen primarily expressed in adenocarcinomas of the gastrointestinal tract [24], but elevated levels have also been reported in breast cancer and correlated with advanced stage, tumor burden, and metastatic potential [25–28]. However, the prognostic significance of CA19-9 in breast cancer remains controversial, with some studies demonstrating an association with adverse outcomes, whereas others report no significant prognostic impact [22]. These inconsistencies may reflect differences in study design, sample size, patient populations, assay methodologies, and cut-off definitions [29]. Acknowledging these conflicting findings underscores the current uncertainty regarding the independent clinical utility of CA19-9 and highlights the need for further validation in diverse cohorts. Emerging evidence suggests that metabolic reprogramming and immune microenvironment remodeling are closely interconnected in cancer progression. In breast cancer, these processes may shape tumor behavior and host inflammatory responses, while studies in other solid tumors further support the broader relevance of metabolic-immune interactions [30, 31]. Several biological mechanisms have been proposed to explain this association, including activation of oncogenic pathways (e.g., Wnt and Notch signaling), promotion of angiogenesis, immune evasion, and pro-inflammatory remodeling of the tumor microenvironment [32–34]. Nevertheless, direct mechanistic evidence specifically in HR+/HER2- breast cancer remains limited. The spline analysis further suggested a dose–response relationship between CA19-9 and overall survival, supporting its potential prognostic relevance. The absence of strong non-linearity suggests a predominantly monotonic association. Consistently, pre-treatment CA19-9 > 27.54 U/mL remained independently associated with worse OS, although the relatively wide confidence interval indicates limited statistical precision and warrants cautious interpretation.

LMR reflects the balance between host antitumor immunity and systemic inflammation. Lymphocytes contribute to tumor immune surveillance and antitumor immunity, whereas monocytes may promote tumor progression through inflammatory signaling and macrophage polarization [35, 36]. Prior studies have demonstrated that decreased LMR is associated with adverse outcomes in several solid tumors, including breast cancer [37]. Inflammation-based hematologic markers and immune-nutritional prognostic indices have been increasingly recognized as prognostic indicators across malignancies, reflecting host immune status and tumor microenvironment interactions [38, 39]. In our cohort, low LMR was associated with poorer OS and showed a borderline trend toward poorer DFS in univariate analysis; however, this association did not persist after multivariate adjustment. This finding indicates that the prognostic impact of LMR may be partly mediated by established clinicopathologic variables and treatment factors, rather than representing an independent determinant of outcome. Therefore, LMR should be interpreted as a complementary biomarker rather than an independent prognostic indicator.

The combined assessment of CA19-9 and LMR showed modestly higher discrimination in this cohort compared with either marker alone. Consistent with these findings, the nomogram integrating CA19-9, LMR, and clinicopathologic variables demonstrated reasonable calibration and a favorable decision-curve profile in internal validation, although external validation is required before any inference regarding clinical usefulness can be made. The C-index of the nomogram was higher than the AUC values observed for individual biomarkers because the nomogram incorporated multiple clinicopathologic variables in addition to CA19-9 and LMR, thereby improving the overall discriminative ability of the model. However, the discriminative performance of CA19-9 and LMR was relatively modest, as reflected by the low AUC values observed in ROC analysis. Therefore, these biomarkers should be interpreted as complementary prognostic indicators rather than standalone predictive tools. The incremental prognostic value beyond traditional clinicopathologic factors (such as tumor stage and nodal status) appears limited, and whether their integration into prognostic models can meaningfully improve risk stratification or clinical decision-making requires validation in independent external cohorts before routine implementation can be considered.

Importantly, after adjusting for clinicopathologic characteristics, the association between CA19-9 and OS remained statistically significant, whereas its relationship with DFS was attenuated. This pattern suggests that CA19-9 may capture aspects of tumor biology or systemic host response that are not fully reflected by conventional staging, but its effect size is modest. LMR did not retain independent prognostic significance after adjustment, further emphasizing the need for cautious interpretation of inflammation-based markers in the context of established prognostic factors [40]. In the present study, adjuvant chemotherapy appeared to be associated with worse DFS in univariate analysis. This finding is likely attributable to confounding by indication, as patients with more aggressive tumor characteristics are more likely to receive chemotherapy in routine clinical practice. Therefore, this association should not be interpreted as evidence that chemotherapy adversely affects prognosis.

Several limitations should be acknowledged. First, this was a retrospective single-center study, which may introduce selection bias and limit the generalizability of the findings. Therefore, external validation in independent multicenter cohorts is required before routine clinical application. Second, peripheral blood biomarkers can be influenced by subclinical inflammatory conditions despite exclusion of overt infections, and single pre-treatment measurements may not fully capture dynamic host-tumor interactions. Third, immunohistochemical validation of tumor CA19-9 expression was not available because archived tissue samples were unavailable. Fourth, the cut-off values for CA19-9 and LMR were derived from ROC analysis within the same dataset, which may introduce optimism and limit their generalizability to other populations. Fifth, the relatively limited number of OS events may have affected the statistical stability of multivariable analyses. To reduce the risk of overfitting, the number of variables included in the multivariable model was restricted. Accordingly, the multivariable analyses should be considered exploratory and hypothesis-generating rather than definitive. Finally, although CA19-9 showed an independent association with OS, the incremental predictive contribution beyond conventional clinicopathologic variables was not formally quantified using discrimination improvement or reclassification metrics. Additionally, because the cohort was treated during 2011–2012, treatment patterns may not fully reflect current standards of care, potentially limiting the contemporary applicability of the findings.

Conclusions

In conclusion, pre-treatment serum CA19-9 was independently associated with overall survival in HR+/HER2- breast cancer, whereas LMR demonstrated prognostic relevance primarily in univariate analysis. Neither biomarker showed independent prognostic value for DFS. Elevated CA19-9 and low LMR were associated with poorer outcomes in this cohort, particularly with respect to overall survival. Although these readily accessible blood-based biomarkers may offer supplementary prognostic information, their incremental value beyond established clinicopathologic factors appears limited in this study. Further validation in independent, preferably multicenter prospective cohorts is required before routine clinical application can be considered.

Abbreviations

CA19-9

Carbohydrate antigen 19 − 9

LMR

Lymphocyte-to-monocyte ratio

HR+

Hormone receptor positive

HER2-

Human epidermal growth factor receptor 2 negative

OS

Overall survival

DFS

Disease-free survival

ROC

Receiver operating characteristic

CI

Confidence interval

HR

Hazard ratio

Authors’ contributions

(I) Conception and design: HJ, WX; (II) Administrative support: All authors; (III) Provision of study materials or patients: All authors; (IV) Collection and assembly of data: HJ, XZC; (V) Data analysis and interpretation: HJ, XZC; (VI) Manuscript writing: All authors; (VII) Final approval of manuscript: All authors.

Funding

No funding.

Data availability

The datasets generated and/or analysed during the current study are not publicly available due to patient privacy and ethical restrictions, but are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Institutional Review Board of Sun Yat-sen University Cancer Center (IRB number: B2025-823-01), and the requirement for informed consent was waived due to the retrospective analysis of de-identified patient data. All procedures performed in this study involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.