ECa remains a leading cause of cancer-related mortality globally (1) and in Portugal (25). Most patients are diagnosed with locally advanced or metastatic disease, entailing poor 5-years survival rate (about 25% and 5%, respectively) (26). Thus, new strategies for early detection of this malignancy are urgently needed. In this context, epigenetic alterations such as DNA methylation have emerged as promising biomarkers in several cancers, including ECa (27, 28).
Herein, we tested three gene promoters as ECa DNA methylation-based biomarkers, following an extensive literature review. Indeed, COL14A1me and GPX3me were previously reported to allow for early ESCC detection (16), whereas ZNF569me was reported in Barrett’s esophagus, a precursor lesion to EA (16, 19). Interestingly, we showed that ZNF569me levels could discriminate between ECa and normal esophagus with high specificity, regardless of the histotype, confirming and extending those previous reports. Furthermore, ZNF569me was shown to play a tumor suppressive role in head and neck squamous cell carcinoma (29) and a DNA-methylation based panel, which included ZNF569, discriminated gastric adenocarcinoma from normal mucosa (30).
Nonetheless, ESCC and EA displayed different cancer-specific methylation patterns. Indeed, in treatment-naïve tumors, the three selected genes disclosed different methylation levels among ESCC and EA, variably comparing to normal esophageal tissue samples. Differential methylation patterns between ESCC and EA have been previously reported and are in line with our results (31, 32). In particular, available data support the value of identification specific ESCC methylation panels to enable early detection (16, 33, 34). COL14A1 aberrant hypermethylation has been reported in ESCC (16), as well as renal cell carcinoma, sarcomas and endometrial carcinoma (35–37), whereas hypomethylation has been shown in coronary artery disease (38). GPX3 promoter hypermethylation has been reported in ESCC and esophageal glandular lesions, including Barrett’s esophagus and EA (17, 18, 39). In the same vein, ZNF569 promoter hypermethylation has been associated with glandular lesions, like Barrett’s esophagus (19) in comparison with normal esophagus. We found ZNF569me in both histosubtypes and, thus, this gene constitutes a promising biomarker for ECa detection, regardless of histological subtype. We found COL14A1 promoter methylation levels slightly higher in ESCC, although not statistically significant. This can be partially explained by variations in population (Asian vs. Caucasian), as previously attested for some genes (40), and different nature of samples tested (plasma vs. FFPE). Notwithstanding, COL14A1 promoter methylation levels were significantly lower in EA compared to normal. To our knowledge, this is the first reported association between COL14A1 methylation levels and EA. Conversely, GPX3 promoter methylation levels did not differ between EA from normal tissues, although a few cases disclosed higher methylation levels (data not shown). Notwithstanding, different sample processing (fresh frozen tissues vs. FFPE) (17, 39), different methodologies to assess GPX3 methylation levels among studies (qMSP vs. MSP vs. methylation ligation-dependent macroarray vs. pyrosequencing) and the smaller size of some cohorts (17, 18) may explain some disparate results.
Overall, we propose two different methylation-based panels, both with high accuracy to early detect ECa according to histological subtype. The ESCC-panel displayed higher specificity (87.5%), whereas the EA-panel disclosed higher sensitivity (97.5%). In fact, the performance of both methylation-based panels compares well with that of other studies (32, 41, 42). For EA detection, Moinova et al reported a two-gene methylation panel, comprising CCNA1 and VIM, with higher specificity (91.7%), similar to TAC1 hypermethylation reported by Jin et al (43, 44). Additionally, for ESCC Li et al. and Wang et al. reported panels with higher performance in Asian populations (16, 45).
Currently, most ECa patients are treated with neoadjuvant therapy followed by surgery, if diagnosed with locally advanced disease (2). The randomized CROSS trial showed that surgery has a major impact in ECa patients’ QoL. Features such as fatigue and physical performance are decreased even in long-term survivors. Those effects are similar in patients undergoing neoadjuvant treatment or surgery only, emphasizing the impact of surgery in QoL (7, 8). Hence, biomarkers enabling the identification of patients complete response to neoadjuvant treatment (whom might be spared surgery) and to early detect disease recurrence are needed (46) to improve QoL without risking the likelihood of cure. Thus, we evaluated methylation levels of candidate genes in 36 samples from non-complete responders after neoadjuvant treatment. In our series, only ZNF569me levels significantly differed in EA comparatively to normal esophagus, whereas GPX3 promoter methylation levels were significantly higher both in ESCC and EA than in normal esophagus. Because GPX3me has been associated with ChT resistance (47), GPX3me levels observed in EA after neoadjuvant treatment might be explained by selective pressure caused upon neoplastic cells, entailing adaptative alterations induced by treatment (48). Several studies have associated DNA methylation with ChT or RT resistance (49–52). However, most used samples before any treatment or in vitro studies with immortalized cell lines (51). Thus, a direct comparison between our results and previously reported data should be made with caution. Nonetheless, the lack of information on methylation status before treatment and the small size of the pre-treated patient cohort along with the retrospective nature of our series are major limitations of this study. Importantly, our series comprised ESCC and EA samples in similar proportions, contrarily to most of the previous studies which evaluated methylation status.