Assessment of Serum Tumor Markers for Predicting Ocular Metastasis in Lung Adenocarcinoma: A Retrospective Study

The purpose of this study was to detect clinical variations between lung adenocarcinoma patients with and without ocular metastasis (OM) to identify risk factors for OM and assess the diagnostic values. We included 1153 patients with lung adenocarcinoma in this study. Independent t-tests and chi-square tests were used to compare patients' clinical characteristics. Statistically significant parameters were analyzed by binary logistic regression to detect risk factors of OM. The results showed that the OM group had increased alpha-fetoprotein (AFP), carcinoembryonic antigen (CEA), cytokeratin fragment 19 (CYFRA 21-1), carbohydrate antigen- (CA-) 125, CA-153, and total prostate-specific antigen (TPSA) compared with the NOM group. CYFRA21-1 is the most useful biomarker for detecting OM in this population.


Introduction
Lung cancer is a common malignant tumor worldwide. The incidence and mortality of lung cancer rank first among all cancer types in males, accounting for 17% of all new cancer cases and 23% of cancer-related deaths [1]. The female incidence of lung cancer has continually increased in recent years, and it is now the second most common type of cancer in women [2]. Causes of lung cancer include air pollution and exposure to occupational and environmental carcinogens [3,4]. Normal epithelial cells experience multiple genetic insults and eventually undergo abnormal growth and develop invasive behaviors. Lung cancer can be divided into small cell and nonsmall cell according to the histological type. The latter group includes squamous cell cancer, adenocarcinoma, and large cell cancer. Most lung adenocarcinomas originate from the bronchial mucosal epithelium, while a few develop from the mucous glands of bronchia. However, the prognosis of lung adenocarcinoma is poor [5,6].
The eye is a rare metastatic site for malignant tumors. Ocular metastasis (OM) is the most often secondary to, for example, breast, lung, kidney, or prostate cancer [7,8]. Lung adenocarcinoma is usually diagnosed at an advanced stage, and patients often have distant metastases [9]. The most common sites are the brain, bone, liver, and adrenal gland [10,11]. In the early stage, OM may have not been accompanied by ophthalmic symptoms. Over time, patients with OM may have blurred vision, pain, visual field defects, flashes, and diplopia [12]. OM can also result in pupil deformation and secondary glaucoma [13]. These symptoms seriously affect the quality of life and shorten the survival time of patients with lung adenocarcinoma. Some studies have also reported asymptomatic cases of choroidal metastasis [14,15]. Given the shortcomings of conventional computed tomography (CT) and magnetic resonance imaging (MRI) in detecting eye metastases, a simple and economical technique for predicting eye metastases would be clinically useful.
Tumor markers are widely used for clinical screening. Detection of markers in the blood, body fluid, or cells can facilitate diagnosis, help clarify the pathogenesis, and inform prognosis. Many serum tumor markers are associated with lung cancer. For example, Ma et al. [16] found increased levels of carcinoembryonic antigen (CEA), carbohydrate antigen-(CA-) 125, and cytokeratin fragment 19 (CYFRA21-1) in lung cancer patients, which could be used for diagnostic purposes. Previous studies have also shown that CEA, CYFRA21-1, and CA-125 are associated with worse prognosis in non-small-cell lung cancer [17].
Tumor markers may be effective markers for predicting metastasis. However, it is not clear whether there are differences in tumor markers used in the diagnosis of OM and NOM patients with lung adenocarcinoma. In this study, we collected medical records of lung adenocarcinoma patients treated at the First Affiliated Hospital of Nanchang University and retrospectively analyzed the diagnostic value of several tumor markers, which might provide a medical basis for predicting OM in this population.

Statistical Analysis.
We performed independent t-tests (age, tumor markers) and chi-square tests (sex) to compare the OM and NOM groups. Binary logistic regression models were then applied to identify independent risk factors for OM. We then constructed receiver operating characteristic (ROC) curves and areas under the curves (AUCs). Excel 2010 software (Microsoft, Washington, USA) was used to calculate the cutoff value, sensitivity, and specificity of each biomarker. Differences were considered significant at p < 0:05. All statistical analyses were performed using SPSS 20.0 software (SPSS, IBM, USA) and Excel 2010 software.

Discussion
Lung adenocarcinoma has become one of the most common diseases with high morbidity and mortality. OM is rare in lung adenocarcinoma patients, but the clinical symptoms are serious. OM seriously affects the quality of life of patients and may suggest poor prognosis, so early prediction is of great significance.

Disease Markers
OM that appears in lung cancer patients is caused by hematogenous dissemination of tumor cells [18]. In a study including 229 eyes of 194 patients with uveal metastasis [19], tumors were located in the choroid (88%), iris (10%), and ciliary body (2%), with bilateral uveal (18%). The choroid is supplied by several thick, short posterior ciliary arteries in the rear of the eye, and there are extensive anastomotic branches between the choroidal vessels. The lumen is larger, and blood flow is abundant but slow [20]. For these reasons, choroidal metastasis is the most common clinical presentation. Metastasis to the orbit, eyelid, conjunctiva, retina, and optic nerve has also been reported [21,22]. Interestingly, OM is more common in the left eye. This is because the left common carotid artery branches directly from the aortic arch, allowing tumor emboli to easily enter the left eye, whereas right eye access requires bypassing the anonymous artery.
Traditional histopathological and immunohistochemical examinations and cytogenetic analyses are tedious and time-consuming. Once a diagnosis of lung cancer is made, physicians can choose multimodal treatment including surgery, radiotherapy, and chemotherapy. Tumor markers mainly exist in serum and can be detected by immunological, biological, and chemical methods [23]. Compared with CT and MRI, they are economical and convenient. Most importantly, tumor markers have a predictive value. Table 6 shows the risk factors for distant metastasis of lung cancer reported in previous studies.
After analyzing clinical data from 1153 patients with lung adenocarcinoma, we found that the concentrations of AFP, CEA, CYFRA 21-1, CA-125, CA-153, and TPSA were significantly elevated in patients with OM. However, AFP is associated with liver metastasis [24], and TPSA is associated with benign prostatic hyperplasia, prostate cancer, and bone metastasis [25,26]. We therefore excluded AFP and TPSA from the tumor markers under consideration as OM markers. According to the binary logistic regression result,   3 Disease Markers CYFRA 21-1, CA-125, and CA-153 were identified as possible independent risk factors for OM in patients with lung adenocarcinoma. CYFRA 21-1 is a fragment of cytokeratin-19 (an epithelial cell filament) produced during cancer cell differentiation. This marker has been associated with lung cancer, colorectal tumors, and bladder cancer [27][28][29], as well as with distant metastasis in patients with lung cancer [30]. In a retrospective study, Chen et al. [31] found that CYFRA 21-1 was an independent risk factor for predicting lymph node metastasis in lung cancer, with a sensitivity and specificity of 0.375 and 0.85, respectively. CA-125, also known as tumor antigen 125, is a high molecular weight glycoprotein produced by the MUC16 gene that is expressed on the surface of epithelial ovarian tumors and mesotheliumderived cells. CA-125 has been widely used to diagnose and assess progression and treatment effects for multiple types of cancer [32,33] CA-125 levels were also found to be useful for predicting lymph node metastasis in patients with ovarian cancer [34]. CA-153 was first identified on breast cancer cell membranes with a relative molecular weight of 4000. Abnormal CA-153 levels may also be measured in patients with many types of tumors including lung cancer [35]. It can be used to predict bone metastasis in breast cancer patients [36]. Based on the analyses of CYFRA 21-1, CA-125, and CA-153 levels and the high AUCs of the three biomarkers, we concluded that they are independent risk factors for OM in patients with lung adenocarcinoma.
The cutoff values showed that CYFRA 21-1 > 6:785 ng/ mL, CA-125 > 66:295 U/mL, and CA-125 > 13:005 U/mL were associated with OM in patients with lung adenocarcinoma. The highest AUC was observed for CYFRA 21-1, suggesting its superior diagnostic value in predicting OM. We also analyzed diagnostic accuracy rates for different combinations of these risk factors. However, the AUC values of these combinations were lower than that of CYFRA 21-1.
Some limitations of this study should be mentioned. First, the group of patients with OM was small, and these findings should be replicated in a larger cohort. Secondly, the participants were from the same hospital, which may limit the generalizability of our results.     Disease Markers     [38] 2004 Adenocarcinoma Liver AFP Cabreraalarcon [39] 2011 NS NS CYFRA21-1 Lee [30] 2012 NSCLC Brain CEA Chen [31] 2015 NS Lymph node CYFRA21-1, CEA Chen [40] 2015 NSCLC Brain NSE Zhou [41] 2017 NS Bone CA-125, ALP Morita [42] 2019 NSCLC Intertrabecular vertebral CEA NS: not specific; NSCLC: non-small-cell lung cancer.

Disease Markers
Based on the analysis of 1153 patients with lung adenocarcinoma, we conclude that the serum concentrations of CYFRA 21-1, CA-125, and CA-153 are independent risk factors for OM. Among these, CYFRA 21-1 has the highest accuracy in predicting OM.

Data Availability
All data is available on request.

Conflicts of Interest
This was not an industry-supported study. The authors report no conflicts of interest in this work.

Authors' Contributions
Wen-Qing Shi and Wen-Feng Liu have contributed equally to this work.