This article is mentioned in:

Introduction

Breast cancer (BC) is the most commonly diagnosed cancer in women globally in addition to being the leading cause of cancer mortality in women in >100 countries, with a continuously increasing incidence (1). Moreover, >1/3 of patients with BC will develop distant metastases such as lung, liver, bone and brain metastases (2-5).

Ocular metastasis (OM), an uncommon distant metastasis, is easily neglected because of its obscure clinical symptoms in the early stage (6). When it develops to an advanced stage, OM causes ocular pain, foreign body sensation, vision loss, visual field defects and other symptoms, thus seriously affecting patients' quality of life (7). Currently, positron emission computed tomography/computed tomography (PET/CT), magnetic resonance imaging (MRI) and ultrasonic testing are often used in clinical practice for diagnosis of OM (8,9). However, these approaches have obvious disadvantages, including their expense and damage from high-dose radiation (10). Thus, it is important to explore improved methods for predicting OM in breast cancer. Serum testing for clinical parameters is a practical method to assess the possibility of distant metastases, as it can shed light on the progress of tumors. Among the various parameters that have been used in clinical practice, tumor markers are considered to be reliable indices to predict distant metastases in patients with cancer (11).

A cancer biomarker is a substance or process that indicates the presence of cancer in the body (12). Measured either in the tumor or in blood, tumor biomarkers can be used to evaluate tumor condition and thus predict the possibility of developing distant metastases (11). Traditional cancer biomarkers include embryonic antigens, as well as protein, carbohydrate, enzyme and hormonal markers (13).

Cancer antigen 153 (CA153) is used to detect Mucin 1 (MUC-1), a transmembrane protein consisting of two subunits (14). MUC-1 is expressed at the apical plasma membrane in normal secretory epithelial cells (15). However, it is released into the serum when metastatic BC occurs (16). Numerous other advanced types of cancer, including ovarian, pancreatic, gastric and lung cancer, also result in elevated CA153 levels (17-20). High levels of C153 can also be observed in a number of benign diseases, including chronic active hepatitis, liver cirrhosis, sarcoidosis and metaplastic anemia (17,21,22). Owing to its low specificity, the use of CA153 in diagnosing early BC has been limited, although it provides useful prognostic information. Numerous studies have demonstrated that high preoperative levels of CA153 are associated with shorter disease-free and overall survival times in patients with BC (23,24).

Other cancer biomarkers have also been used in diagnosing tumors and metastases. In patients with colorectal cancer and liver metastasis, higher carcinoembryonic antigen (CEA) level is associated with shorter median progression-free survival and median liver progression-free survival (25). In Zhao's study (26), the sensitivity and specificity of cancer antigen 125 (CA125) for diagnosing ovarian cancer were 88.2 and 67.4%, respectively. Cancer antigen 199 (CA199) level >300 mg/ml is an independent prognostic factor for postoperative survival in Zheng's study (hazard ratio, 3.76; 95% CI, 2.18-6.49) (27). Alkaline phosphatase (ALP) is used to diagnose bone metastases in BC with a high specificity (93.96%) but a low sensitivity (65.14%) (28).

A number of studies have reported altered levels of CA153 in patients with cancer with metastases at various sites and investigated their diagnostic value. However, whether CA153 levels could be used to detect OM in patients with BC remains unknown. The present retrospective study analyzed levels of CA153 and other common tumor biomarkers of patients with BC to determine their value in diagnosing OM.

Materials and methods

Study design

This study was designed to determine the associations between levels of cancer biomarkers and OM in patients with BC. The data range of sample collection is Jan 2005-Jan 2019. All BC were diagnosed pathologically, and the pathologist was independent from the study. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of the First Affiliated Hospital of Nanchang University (approval no. CDYFY-2015-112; date, 2015-05-06; Nanchang, China). All the methods complied with relevant guidelines and regulations. Breast tissues from patients were resected during surgery and the diagnosis of BC was made on the basis of pathological biopsy. CT and MRI were used to make the OM diagnosis. Inclusion criteria were all patients who were diagnosed with BC based on the pathologically. Exclusion criteria were primary ocular malignant tumor, primary ocular benign tumor and secondary BC. As this is a retrospective study, informed consent of the patients was waived, which was approved by the Ethics Committee at The First Affiliated Hospital of Nanchang University.

Data collection

Age, sex, histopathological subtype and menopausal status were recorded as basic information and analyzed for differences among patients (Table I). As well as CA153, axillary lymph node metastasis (ALNM), CEA, CA125, CA199, ALP and calcium were analyzed as these parameters are frequently used in clinical practice and reflect the condition of tumors in patients with cancer. All clinical indices were collected from medical records when patients were first diagnosed with BC. Consecutive data are represented as mean ± standard deviation.

Table I Clinical features of patients with breast cancer.

Table I

Clinical features of patients with breast cancer.

Characteristic	OM	NOM	Whole	P-value
Age, years±SDa	45.85±7.76	48.31±10.58	48.21±10.48	0.241
Sex, nb
Women	26	601	627	1.000
Men	0	2	2
Menopausal status, nc				0.248
Premenopausal	19	373	392
Postmenopausal	7	230	237
Histopathology, nc				0.656
Invasive ductal carcinoma	20	440	460
Other types	6	163	169
Axillary lymph node metastases, nc				<0.001
0	4	259	263
1-4	7	210	217
>4	15	134	149

[i] ^aUnpaired Student's t-test was used;

[ii] ^bFisher's exact test;

[iii] ^cχ² test was used. OM, ocular metastases; NOM, non-ocular metastases.

Statistical analysis

First, basic clinical features including age, sex, histopathological subtype, menopausal status and ALNM number were compared using Fisher's exact test, χ2 test and unpaired Student's t-test. Then, clinical parameters of the ocular metastases (OM) group and non-ocular metastases (NOM) group were compared across different subgroups using the Mann-Whitney U test. Binary logistic regression was carried out to determine independent risk factors for OM. A receiver operating characteristic (ROC) curve was constructed with MedCalc 18.2.1 (MedCalc Software, Ltd.), and the cutoff value, area under the curve (AUC), sensitivity and specificity were obtained. Measurement data are presented as mean ± standard deviation (SD). P<0.05 was considered to indicate a statistically significant difference.

Results

Clinical features of patients with BC

There were 629 patients with BC in total (627 female and two male), of which 26 patients had OM and 603 had NOM. There were no differences in age (P=0.241), sex (P=1.000), menopausal status (P=0.248), or histopathology (P=0.656) between the OM and NOM groups. The average age of patients in the OM group was 45.85±7.76 years, compared with 48.31±10.58 years in the NOM group. In the OM group, 19 patients had premenopausal status and seven were postmenopausal, whereas in the NOM group the numbers were 373 and 230, respectively. Regarding histopathology, the OM group contained 20 cases of invasive ductal carcinoma and six cases of other types. In the NOM group, there were 440 cases of invasive ductal carcinoma and 163 cases of other types. A statistical difference was revealed for ALNM (P<0.001); the OM group contained four cases without ALNM and 22 cases with ALNM, whereas the NOM group contained 259 and 344 cases, respectively. Details are presented in Table I.

Distributions of patients with BC and condition of their distant organ metastases

The top four cities patients came from were Nanchang, Shangrao, Fuzhou, and Jian (marked in red in Fig. 1A). The majority of the patients were from Jiangxi province (marked in red and yellow in Fig. 1A), but there were also 11 patients from other provinces (marked in blue in Fig. 1A). Details of the patient distribution are presented in Fig. 1A.

Figure 1 Distributions of patients with BC and conditions of distant organs metastases. (A) Distributions of patients with BC were counted. (B) Other distant metastases of patients with BC were counted. OM, ocular metastases; NOM, non-ocular metastases.

Numbers of distant organ metastases (bone, liver, lung and brain) in the two groups were counted. Bone was the most common metastatic site for BC, accounting for 25 and 62 cases in the OM and NOM group, respectively (Fig. 1B).

Subgroup analysis and binary logistic regression analysis

To exclude potential confounding by other distant metastases and clarify the diagnostic value of CA153, data were further divided into subgroups. Differences in CA153 levels between the OM group and NOM group were revealed to be statistically significant in each subgroup (Table II). The P-values are summarized in Table III. The binary logistic regression analysis identified CA153 as a risk factor for OM in patients with BC (P<0.001; Table IV).

Table II Subgroup analysis.

Table II

Subgroup analysis.

A, Premenopausal
Clinical features	OM (n=19)	NOM (n=373)	P-value
CEA (ng/ml)	39.59±107.83	2.93±9.99	<0.001
CA125 (µ/ml)	57.23±103.74	22.86±67.94	<0.001
CA153 (µ/ml)	158.59±129.37	17.16±26.55	<0.001
CA199 (µ/ml)	30.89±50.44	13.28±13.76	0.375
ALP (µ/l)	134.32±101.45	59.60±24.64	<0.001
Calcium (mmol/l)	2.30±0.42	2.32±0.55	0.688
B, Postmenopausal
Clinical features	OM (n=7)	NOM (n=230)	P-value
CEA (ng/ml)	43.56±72.41	39.45±531.00	0.009
CA125 (µ/ml)	138.19±203.30	24.63±174.47	0.002
CA153 (µ/ml)	147.14±72.26	19.55±32.91	<0.001
CA199 (µ/ml)	8.37±4.55	27.50±216.90	0.350
ALP (µ/l)	107.29±59.49	82.15±42.56	0.391
Calcium (mmol/l)	2.12±0.60	2.32±0.15	0.622
C, NALNM
Clinical features	OM (n=4)	NOM (n=259)	P-value
CEA (ng/ml)	53.01±101.07	2.39±4.27	0.079
CA125 (µ/ml)	118.43±209.20	14.84±31.73	0.213
CA153 (µ/ml)	217.99±219.97	13.13±9.72	<0.001
CA199 (µ/ml)	11.62±6.95	12.90±11.85	0.920
ALP (µ/l)	97.50±43.61	63.67±22.47	0.053
Calcium (mmol/l)	2.65±0.16	2.31±0.13	<0.001
D, ALNM
Clinical features	OM (n=22)	NOM (n=344)	P-value
CEA (ng/ml)	38.41±99.88	27.75±434.30	<0.001
CA125 (µ/ml)	71.87±126.60	30.08±156.44	<0.001
CA153 (µ/ml)	144.15±89.88	21.79±37.24	<0.001
CA199 (µ/ml)	27.22±47.59	23.08±177.62	0.646
ALP (µ/l)	132.41±97.68	71.61±40.89	<0.001
Calcium (mmol/l)	2.18±0.47	2.33±0.57	0.235
E, Bone metastases
Clinical features	OM (n=25)	NOM (n=62)	P-value
CEA (ng/ml)	42.14±99.88	141.30±1021.78	0.095
CA125 (µ/ml)	81.67±140.05	105.98±365.58	0.019
CA153 (µ/ml)	156.81±117.64	54.64±77.02	<0.001
CA199 (µ/ml)	25.33±44.89	72.86±416.87	0.704
ALP (µ/l)	130.00±92.31	102.05±77.10	0.222
Calcium (mmol/l)	2.26±0.48	2.49±1.30	0.980
F, Liver metastases
Clinical features	OM (n=14)	NOM (n=33)	P-value
CEA (ng/ml)	36.01±56.65	252.70±1400.68	0.009
CA125 (µ/ml)	53.37±109.41	122.59±462.49	0.061
CA153 (µ/ml)	140.13±127.64	36.24±53.15	<0.001
CA199 (µ/ml)	31.53±55.19	117.97±570.94	0.735
ALP (µ/l)	140.36±101.37	93.33±81.60	0.036
Calcium (mmol/l)	2.32±0.42	2.30±0.25	0.761
G, Lung metastases
Clinical features	OM (n=14)	NOM (n=39)	P-value
CEA (ng/ml)	53.26±131.52	8.04±29.49	0.047
CA125 (µ/ml)	78.84±148.35	35.70±90.15	0.023
CA153 (µ/ml)	131.96±125.65	20.37±21.18	<0.001
CA199 (µ/ml)	37.96±57.54	14.28±17.47	0.175
ALP (µ/l)	90.50±42.53	70.69±20.57	0.102
Calcium (mmol/l)	2.45±0.26	2.31±0.15	0.068
H, Brain metastases
Clinical features	OM (n=12)	NOM (n=12)	P-value
CEA (ng/ml)	20.99±31.48	6.06±8.68	0.160
CA125 (µ/ml)	77.45±131.97	16.19±12.78	0.012
CA153 (µ/ml)	185.05±151.17	28.40±21.97	<0.001
CA199 (µ/ml)	44.28±60.21	10.18±7.11	0.052
ALP (µ/l)	137.25±105.89	64.42±22.92	0.045
Calcium (mmol/l)	2.41±0.29	2.24±0.12	0.155
I, Whole
Clinical features	OM (n=26)	NOM (n=603)	P-value
CEA (ng/ml)	40.66±98.15	16.86±328.08	<0.001
CA125 (µ/ml)	79.03±137.87	23.53±120.14	<0.001
CA153 (µ/ml)	155.51±115.46	18.07±29.14	<0.001
CA199 (µ/ml)	24.82±44.05	18.71±134.39	0.792
ALP (µ/l)	127.04±91.70	68.20±34.42	<0.001
Calcium (mmol/l)	2.25±0.47	2.32±0.44	0.958

[i] Mann-Whitney U test was used. OM, ocular metastases; NOM, non-ocular metastases; ALNM, axillary lymph node metastases; NALNM, non-axillary lymph node metastases; CA, Cancer antigen.

Table III Summary of P-values.

Table III

Summary of P-values.

Clinical features	PRE	POST	NALNM	ALNM	BM	LIM	LUM	BRM	Whole
CEA	<0.001	0.009	0.079	<0.001	0.095	0.009	0.047	0.160	<0.001
CA125	<0.001	0.002	0.213	<0.001	0.019	0.061	0.023	0.012	<0.001
CA153	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
CA199	0.375	0.350	0.920	0.646	0.704	0.735	0.175	0.052	0.792
ALP	<0.001	0.391	0.053	0.000	0.222	0.036	0.102	0.045	<0.001
Calcium	0.688	0.622	<0.001	0.235	0.980	0.761	0.068	0.155	0.958

[i] These were the P-values of all subgroup analyses. PRE, premenopausal; POST, postmenopausal; NALNM, non-axillary lymph node metastases; ALNM, axillary lymph node metastases; BM, bone metastases; LIM, liver metastases; LUM, lung metastases; BRM, brain metastases; CEA, carcinoembryonic antigen; CA, cancer antigen; ALP, alkaline phosphatase.

Table IV Binary logistic regression analysis.

Table IV

Binary logistic regression analysis.

Factors	B	Exp(B)	Exp(B) 95% CI	P-value
CEA	-0.020	0.998	0.992-1.003	0.458
CA125	0.001	1.001	0.998-1.004	0.464
CA153	0.028	1.029	1.018-1.039	<0.001
CA199	0.002	1.002	0.989-1.015	0.759
ALP	0.001	1.001	0.991-1.010	0.892
Calcium	-0.064	0.938	0.266-3.311	0.921

[i] Enter method was used in the binary logistic regression analysis. CI, confidence interval; CEA, carcinoembryonic antigen; CA, cancer antigen; ALP, alkaline phosphatase.

Sensitivity, specificity, AUC and cutoff value for CA153

A ROC curve for CA153 was established to evaluate its diagnostic value. The AUC for CA153 was 0.977, with a 95% CI of 0.958-0.995. The cutoff value was 43.3 µ/ml, with a corresponding sensitivity of 96.15% and specificity of 96.02% (Fig. 2). Details are presented in Table V.

Figure 2 Diagnostic of ocular metastases based on CA153 value. AUC was 0.977 (P<0.001; 95% confidence interval, 0.958-0.995). The cutoff value was 43.3 µ/ml, with a sensitivity of 96.15% and a specificity of 96.02%. AUC, areas under the curve.

Table V Receiver operating characteristic curve analysis.

Table V

Receiver operating characteristic curve analysis.

Variable	Cutoff value	Sensitivity (%)	Specificity (%)	AUC	95% CI	P-value
CA153	43.3	96.15	96.02	0.977	0.958-0.995	<0.001

[i] Sensitivity and specificity were obtained at the point of cutoff value. AUC, area under the curve; CI, confidence interval.

Discussion

BC has a high incidence of distant metastases and has increased the medical cost burden of our society. Although the widespread use of mammography, PET/CT and MRI has increased the detection rate, there are still some disadvantages (29). Various risk factors and clinical parameters associated with distant metastases of BC have been taken into consideration to make up for the limitations of these techniques (28,30-32) (Table VI).

Table VI Literature summary of risk factors for common distant metastases in patients with BC.

Table VI

Literature summary of risk factors for common distant metastases in patients with BC.

Author (Refs.)	Year	Distant metastasis	Risk factors
Slimane et al (30)	2004	Brain	Negative hormone receptor status
Cao et al (31)	2012	Liver	Lactate dehydrogenase + γ- glutamyltransferase + CA153
Chen et al (28)	2017	Bone	Axillary lymph node metastases + CA125 + CA153 + alkaline phosphatase + hemoglobin
Hu et al (32)	2017	Lung	CD44v

[i] CA153, cancer antigen 153; CA125, cancer antigen 125.

Tumor marker serum tests are widely used in patients with cancer to evaluate tumor occurrence and condition. Several studies have already investigated the use of tumor markers in a number of types of cancer. Brand et al (33) revealed higher levels of CA199 in patients with pancreatic ductal carcinoma (PDC) compared with healthy controls, indicating that CA199 can be used to diagnose PDC. In Stojkovic's study (34), higher CEA levels are associated with advanced stage in patients with colorectal carcinoma (CRC); the study also demonstrates the use of CEA as a diagnostic factor to predict the severity of CRC. In another study (35), CA125 has been used to diagnose ovarian cancer; the sensitivity and specificity of CA125 according to the ROC curve were 79.6 and 82.5%, respectively. Also, in work by Tang et al (36), CA153 has been used in the diagnosis of BC with a sensitivity of 63% and a specificity of 82%.

Notably, tumor markers are also helpful to assess a patient with cancer's risk of distant metastasis. Yuan et al (37) observed that CA125 can increase A2780 and OVCAR-3 cell migration, indicating that CA125 may play an important part in tumor metastasis. Moreover, Zhou et al (38) revealed CA125 to be an independent risk factor of bone metastases in patients with lung cancer. Cao et al (31) revealed that CA153 is useful to predict liver metastasis in patients with BC.

OM is often associated with poor prognosis and low quality of life (39). The incidence of OM in BC has varied among different studies, with rates between 5-30% (40,41). The choroid is the most common site of OM owing to its rich blood supply, followed by the orbit, iris, ciliary body, optic nerve, conjunctiva and eyelid (42). When tumor cells grow in these locations, the substances they produce and release enter the blood and can be tested (43). Thus, tumor markers can well reflect the stage of an OM tumor. However, the diagnostic value of tumor markers in OM of patients with BC remains unclear.

The present study analyzed the differences in basic clinical features and common tumor markers between the OM group and NOM group. Among the basic features, ALNM demonstrated significant differences between the two groups. There were also statistically significant differences in CA153 levels in all subgroups and in the whole group, indicating its diagnostic value as a risk factor for OM in patients with BC.

The present study was the first to evaluate the diagnostic value of CA153 in patients with BC with OM. Compared with patients with NOM, patients with OM had higher CA153 levels. The cutoff value was 43.3 µ/ml, suggesting that patients with BC with CA153 levels higher than this value were at risk of OM. Moreover, the sensitivity and specificity for diagnosing OM in patients with BC were 0.96 and 0.96, respectively. Given the high AUC value (0.977), CA153 represents a relatively high predictive accuracy. As an important specific marker of BC, it is widely used in diagnosis of BC and in assessing prognosis and metastasis (44). In a meta-analysis performed by Tang et al (36), the sensitivity and specificity of CA153 in breast secretions for predicting BC were 0.63 and 0.82, respectively. Another meta-analysis conducted by Li et al (45), demonstrated that elevated CA153 is associated with shorter disease-free survival times in patients with BC. In a retrospective study performed by Chen et al (28), CA153 was revealed to be an independent risk factor for BC bone metastases, with sensitivity and specificity of 0.77 and 0.87, respectively.

Although CA153 is an independent risk factor for both bone metastases and OM, there are differences between the two conditions. First, in the present study, bone metastases were more common, whereas OM was relatively rare. Second, OM was more likely to occur at an advanced stage, indicating the progression of the disease, whereas bone metastases could be found in early BC. Third, when bone metastases occurred, elevated CA153 and ALP levels were detected, whereas OM was not associated with high levels of ALP. Regarding the diagnostic value of CA153, high levels of CA153 were detected in both metastases, but with some differences. On the one hand, the cutoff values were different. For diagnosing bone metastases, the cutoff value was 25.42 µ/ml, whereas that for OM was 43.3 µ/ml. On the other hand, the sensitivity and specificity for OM were both >95%. Given these differences and clinical symptoms, it is not difficult to distinguish bone metastases from OM.

However, the present study had some limitations. First, as OM is rare, the sample size was relatively small. Second, the majority of participants were from the same province; more studies of patients from different places are needed. Third, although the present study indicated an association between CA153 and OM, it could not identify their causal relationship. Hence, tests from more institutions are necessary to validate the conclusions of the present study. It would also be appropriate to explore the diagnostic value of CA153 in other metastases. Furthermore, associated molecular experiments could be carried out to explore the specific correlations.

The present study clarified CA153 as an independent risk factor for OM in patients with BC. Patients with BC with CA153 >43.3 µ/ml were more likely to have OM. Although CA153 levels had high predictive value for OM according to the present study, CA153 alone was insufficient to diagnose OM in patients with BC. Combining relevant clinical parameters with clinical symptoms would be an improved strategy. However, the present study indicated that physicians should be vigilant when CA153 levels are over the cutoff value. Treatment should be performed in time to avoid severe metastases. These results will help physicians to improve prediction of OM in patients with BC and provide some insight into the specific mechanisms of tumor markers in cancer metastases.

Acknowledgements

Not applicable.

Funding

Funding: This work was supported by the National Natural Science Foundation of China (grant no. 82160195), Jiangxi Double-Thousand Plan High-Level Talent Project of Science and Technology Innovation (grant no. jxsq2023201036) and the Key R & D Program of Jiangxi Province (grant no. 20223BBH80014).

Data availability and materials

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Authors' contributions

YS designed the study. QG, QH and JL analyzed and interpreted the patient data. YS and QG confirm the authenticity of all the raw data. QG, QH, JL, QYL and YS made major contributions writing the manuscript. YM, BL, and RL performed the statistical analyses. WS, QL, QY, and QYL collected the clinical data. All authors read and approved the final manuscript.

Ethics approval and consent to participate

The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of the First Affiliated Hospital of Nanchang University (approval no. CDYFY-2015-112). All patients provided written consent to participate.

Consent for publication

Not applicable.

Conflict of interest

The authors declare that they have no competing interests.

References