Carbohydrate antigen 72-4 and tumor-associated glycoprotein 72 - tumor marker: Past, present and future

Abstract

Carbohydrate antigen 72-4 (CA72-4) is a serum tumor marker linked to tumor-associated glycoprotein 72 (TAG-72), a high-molecular-weight mucin-like antigen identified during the monoclonal-antibody era. Although developed later than CA19-9 and most closely associated with gastric cancer, CA72-4 has since attracted broader interest across gastrointestinal oncology, peritoneal disease, malignant effusions, cytology, benign false-positive states and TAG-72-directed translational research. This review examines the history, biology, clinical utility, diagnostic limitations and future applications of CA72-4 and TAG-72, using gastric cancer as the principal clinical anchor while considering wider esophagogastric, colorectal, appendiceal, pancreaticobiliary, peritoneal and fluid-based applications. Current evidence supports CA72-4 as an adjunctive biomarker rather than a population screening test or stand-alone diagnostic tool. Its most defensible clinical roles are in gastric cancer risk stratification, peritoneal disease assessment, selected multimarker panels and malignant effusion work-up. Looking ahead, TAG-72 may ultimately prove more clinically consequential than serum CA72-4 itself if imaging, cytology-based diagnostics and targeted therapeutic platforms continue to mature.

Key Words: Carbohydrate antigen 72-4; Tumor-associated glycoprotein 72; Gastric cancer; Tumor markers; Peritoneal carcinomatosis

Core Tip: Carbohydrate antigen 72-4 (CA72-4) refers mainly to the clinical assay or serum readout of circulating tumor-associated glycoprotein 72 (TAG-72)-related antigenic determinants; while TAG-72 refers to the broader TAG target. Gastric cancer remains the main clinical setting for CA72-4, especially in prognosis, recurrence assessment and peritoneal dissemination. Pleural-fluid CA72-4 has high specificity but modest sensitivity for malignant pleural effusion, supporting a rule-in adjunctive role. TAG-72 also links CA72-4 to cytology, immunohistochemistry, molecular imaging and cellular therapy research. Future studies should prioritize assay standardization, disease-specific thresholds, marker kinetics and patient-management outcomes.

INTRODUCTION

Tumor markers remain attractive in gastroenterology and gastrointestinal oncology as they are minimally invasive, easily measured repeatedly, generally understood by patients when monitored serially, and abnormal values may appear to offer an early signal of occult disease. When interpreted within a defined clinical context, they may support prognosis, surveillance, treatment response or clarification of a suspected diagnosis. Their value is much more limited when used as broad screening tests in low-prevalence populations, where abnormal results often have poor positive predictive value and may lead to patient anxiety, repeated imaging, endoscopy, incidental findings and prolonged surveillance without a clear management benefit[1]. This principle is especially relevant to carbohydrate antigen 72-4 (CA72-4) as the marker is available, easily repeated and often included in health-screening panels, despite its performance being highly dependent on disease prevalence, pre-test probability and the clinical question being asked.

CA72-4 is best known as a gastric cancer-associated marker, but its literature now extends beyond gastric malignancy. Studies have examined its utility in esophagogastric cancer, colorectal cancer, appendiceal mucinous disease, pancreatic cyst fluid, gallbladder cancer, malignant pleural effusion, ascites, cytology and tumor-associated glycoprotein 72 (TAG-72)-directed translational oncology. This wider body of work reflects the biological importance of aberrant glycosylation and mucin expression in epithelial cells. Cancer-associated glycosylation can alter cell adhesion, receptor signaling, immune recognition, epithelial plasticity and metastatic behavior[2]. Mucins further contribute to the tumor cell surface architecture, protect tumor cells from hostile microenvironments and influence interactions with stromal and immune compartments[3]. Glycosylated mucins may help tumor clones adapt to hostile microenvironments by altering immune visibility, anoikis resistance and serosal spread[2,3]. In this sense, circulating CA72-4 represents only one measurable expression of a broader adaptive phenotype.

Conceptually, it is important to distinguish CA72-4 from TAG-72. TAG-72 is a high-molecular-weight mucin-like tumor-associated glycoprotein first identified through monoclonal antibody studies, while CA72-4 refers to a clinical immunoassay used to detect circulating TAG-72-related antigenic determinants[4-6]. Subsequent antibody platforms, including CC49, improved recognition of TAG-72-associated epitopes and supported further assay developments, molecular imaging and targeting studies[7]. TAG-72 expression has been documented across several epithelial malignancies, reinforcing its biological breadth and explaining why serum, tissue and fluid studies have evolved in parallel[8,9].

In the era of precision medicine, the potential clinical utility of CA72-4 include disease-specific thresholds, longitudinal marker kinetics, fluid-specific interpretation, peritoneal-risk prediction and integration with endoscopy, cytology, radiomics, circulating tumor DNA (ctDNA), methylation panels and artificial intelligence-supported models. Serum CA72-4 may rise when tumor burden, antigen secretion, and host clearance align. This review therefore reappraises CA72-4 and TAG-72 from a gastroenterology and gastrointestinal oncology perspective. It focuses on where CA72-4 has practical clinical value and where TAG-72 biology may still continue to create diagnostic, imaging or therapeutic value.

METHODS

This narrative review was conducted through a targeted search of PubMed using terms related to CA72-4, TAG-72 and their historical antibody platforms. Search terms included “CA72-4”, “CA 72-4”, “CA724”, “CA 724”, “carbohydrate antigen 72-4”, “cancer antigen 72-4”, “TAG-72”, “tumor-associated glycoprotein 72”, “tumour-associated glycoprotein 72”, “B72.3” and “CC49”. These terms were combined with disease and application specific keywords, including “gastric cancer”, “esophageal cancer”, “gastroesophageal junction”, “colorectal cancer”, “appendiceal”, “pseudomyxoma peritonei”, “pancreatic cyst”, “gallbladder cancer”, “peritoneal dissemination”, “ascites”, “pleural effusion”, “cytology”, “immunohistochemistry”, “false positive”, “benign disease”, “CAR-T”, “CAR-NK”, “radioimmunoguided” and “theranostics”. Citation tracking was also used to identify historical antibody and assay literature. Recent systematic reviews, meta-analyses, large observational studies and clinically relevant translational studies were prioritized, while historically important antibody and assay papers were included where they clarified the development of the field.

For this review, evidence was interpreted according to four levels of clinical relevance (Figure 1). These include: Biological detectability, diagnostic or prognostic association, incremental value, and clinical consequence. Biological detectability refers to whether TAG-72 or CA72-4 can be measured in serum, tissue or fluid. Diagnostic or prognostic association refers to whether marker expression or levels differ according to the disease presence, stage, recurrence or prognosis. Incremental value refers to whether CA72-4 improves diagnostic or prognostic performance beyond existing clinical variables, imaging, cytology, endoscopy or other markers. Clinical consequence refers to whether testing changes patient management, including staging, biopsy, surveillance, treatment intent, avoidance of non-therapeutic surgery or patient outcome. This framework was used as much of the CA72-4 literature demonstrates biological detectability or clinical association, whereas fewer studies directly establish incremental value or management impact.

Figure 1 Interpretation of evidence according to levels of clinical relevance. CA72-4: Carbohydrate antigen 72-4; TAG-72: Tumor-associated glycoprotein 72.

THE PAST

Historical background

CA72-4 emerged from the monoclonal-antibody era of tumor-marker discovery. CA19-9 was first described in 1979 by Koprowski et al[10], using hybridoma antibodies generated against colorectal carcinoma cells. Shortly thereafter, monoclonal antibody B72.3 was reported in studies of human mammary carcinoma and shown to recognize a distinct tumor-associated antigen distribution pattern[4]. Johnson et al[5] subsequently characterized the B72.3-bound antigen as a high-molecular-weight tumor-associated glycoprotein, later termed TAG-72, in 1986. Building on this antibody-based work, Gero et al[6] then reported the CA72-4 radioimmunoassay in 1989 as a method for detecting circulating TAG-72 related carcinoma-associated antigen.

CA72-4 therefore emerged approximately a decade after CA19-9, but it did not acquire the same dominant disease identity. CA19-9 became strongly associated with pancreatic ductal adenocarcinoma, despite well-recognized limitations in sensitivity, specificity and interpretation. By contrast, CA72-4 remains distributed across several clinical and translational domains, including gastric, ovarian, colorectal, pancreatic, effusion and antibody-targeting research. CA72-4 has sustained scientific interest but continues to be unevenly integrated into routine clinical practice. Early reviews have recognized CA72-4 as a potentially useful serum marker across several carcinomas, particularly gastrointestinal and gynecological malignancies[11].

Ovarian cancer illustrates this broader non-gastrointestinal biology of TAG-72 and CA72-4. In a study of 106 patients with primary ovarian cancer, serum CA72-4 measured at a cutoff of 4.0 U/mL showed a sensitivity of 63.2% for overall primary ovarian carcinomas, including 67.6% for mucinous cystadenocarcinoma and 66.7% for serous cystadenocarcinoma, with 91.1% specificity for benign ovarian tumors[12]. TAG-72 localization was also demonstrated in ovarian carcinoma tissue using immunohistochemistry and electron microscopy[12]. These findings support the biological breadth of TAG-72 expression across epithelial malignancies. However, modern clinical reviews have increasingly emphasized gastric cancer as the principal setting for CA72-4 use[13], while more recent work has broadened the field again through fluid diagnostics, cytology and TAG-72-directed imaging and therapeutic platforms.

Biomarker vs therapeutic target

CA72-4 and TAG-72 should be interpreted through two distinct evidence pathways. TAG-72 is a high-molecular-weight mucin-like glycoprotein expressed by several epithelial tumors[5,8,9]. It is not a single gene-encoded protein. It denotes a family of tumor-associated truncated O-glycan epitopes, including sialyl-Tn-related antigenic determinants, displayed on high-molecular weight mucins and mucin like glycoprotein. These epitopes have been reported on mucins including MUC1, MUC2, and MUC5AC, among others. It was originally defined through monoclonal antibody recognition, particularly by B72.3, with subsequent antibodies such as CC49 developed to improve binding characteristics and clinical utility[7]. The evidence for TAG-72 directed treatment remains preclinical or early phase. CA72-4 should therefore not be treated as synonymous with TAG-72. Rather, CA72-4 represents a clinical immunoassay readout of circulating TAG-72-related antigenic determinants. Its evidence base consists mainly of diagnostic accuracy studies, observational cohorts and meta-analyses. Its most defensible roles are context specific applications in gastric cancer, assessment of peritoneal dissemination, selected malignant effusions and multimarker models. This distinction is important because TAG-72 may be studied as a tissue antigen, cytological marker, imaging target or therapeutic target, whereas CA72-4 usually refers to a measurable serum or fluid biomarker (Figure 2).

Figure 2 Biological pathway of tumor-associated glycoprotein 72 and carbohydrate antigen 72-4. CA72-4: Carbohydrate antigen 72-4; TAG-72: Tumor-associated glycoprotein 72.

Biology and assay interpretation

The biological basis of CA72-4 measurement is closely linked to antibody recognition of TAG-72 associated epitopes. In a colorectal tissue-serum study, the quantitative CA72-4 immunoassay used B72.3 and CC49 in a double-determinant radioimmunoassay to detect the sialyl-Tn epitope on TAG-72, and tumor TAG-72 levels correlated with corresponding serum CA72-4 levels[14]. This assay biology has practical implications. In gastric cancer, serum CA72-4 may support clinical assessment when interpreted alongside tumor staging, imaging, endoscopy, histology, treatment status and other biomarkers. In contrast, tissue TAG-72 expression may be more relevant for immunohistochemistry, cytology, molecular imaging or targeted therapeutic development. Assay platform differences also affect interpretation, as CA72-4 thresholds vary across studies, populations and laboratory methods[11,13]. Serial monitoring is therefore most meaningful when performed on the same assay platform and interpreted in relation to the disease status, treatment timing and possible confounders.

The link between TAG-72, mucin biology and abnormal glycosylation is also clinically relevant. Truncated O-glycans and abnormal mucin glycosylation are recurrent features of epithelial cancers and may influence adhesion, immune recognition and metastatic behavior[2,3]. CA72-4 is therefore best interpreted as a serum or fluid readout of a broader glycoprotein phenotype that may differ across tissue, ascites, pleural fluid and cytology specimens. Serum CA72-4 reflects circulating antigen burden and clearance, fluid CA72-4 reflects local tumor shedding, and tissue TAG-72 reflects antigen expression and targetability. TAG-72 expression is not specific to invasive carcinoma, where TAG-72 related epitopes have also been detected in inflammatory and premalignant epithelial lesions. TAG-72 should therefore be interpreted as a marker of aberrant glycosylation and epithelial change rather than evidence of a direct tumor driving mechanism.

THE PRESENT

Gastric cancer: The clinical anchor

Gastric cancer remains the principal clinical setting in which CA72-4 has been most consistently studied. Early studies suggested that serum TAG-72, measured using the CA72-4 assay, was more frequently elevated in advanced gastric cancer than in early stage disease and could complement CA19-9 or carcinoembryonic antigen (CEA) during follow-up[11]. Subsequent reviews have continued to position CA72-4 among the conventional serum markers for gastric cancer, alongside CEA and CA19-9[13,15]. A systematic review by the Japanese Gastric Cancer Association screened 4925 reports, identified 187 publications with CEA or CA19-9 data, and found only 19 publications evaluating all three markers: CEA, CA19-9 and CA72-4[16]. The review concluded that marker combinations may assist with staging before surgery or chemotherapy and may support monitoring for recurrence or treatment response, while also emphasizing the limited availability of prospective evidence[16].

Diagnostic performance is limited when CA72-4 is used alone. In a meta-analysis of 10 studies including 6574 participants, CA72-4 alone showed sensitivity 0.58, specificity 0.86 and area under the curve (AUC) of 0.84 for gastric cancer diagnosis[17]. When testing is combined with CEA, CA19-9 and CA72-4, there was an improved sensitivity to 0.67, specificity to 0.89 and AUC to 0.87[17]. Liang et al[18] similarly found that individual serum markers had low sensitivity, while combined testing improved performance but remained incomplete. These findings support CA72-4 as an adjunctive marker in gastric cancer rather than a substitute for endoscopy, histology or staging imaging.

The evidence is weak in population screening. In a multicentre study of 7757 healthy adults, only three gastric cancers were detected. CA72-4 had a sensitivity of 33.3%, specificity 92.8%, positive predictive value 0.18% and negative predictive value 99.97%[19]. Elevated CA72-4 was also associated with benign endoscopic findings, including gastric ulcer, gastric polyps and atrophic gastritis[19]. These data illustrate the issue when applying tumor markers in low-prevalence populations, where it may generate a low positive predictive value when pre-test probability is low. These findings preclude its routine use in population screening.

The most clinically coherent gastric cancer signal relates to peritoneal dissemination. A 2025 systematic review found that CA125 and CA72-4 were useful for detecting peritoneal dissemination in gastric cancer, while CEA and CA125 had stronger independent prognostic evidence[20]. In patients with gastric cancer and peritoneal dissemination, Emoto et al[21] reported sensitivities of 19% for CEA, 36% for CA19-9, 46% for CA125 and 45% for CA72-4 at initial diagnosis. These value remain insufficient for exclusion of disease, but they suggest that CA72-4 may contribute to risk assessment when combined with other clinical, radiological and operative findings.

This is clinically important as peritoneal disease is often occult on conventional imaging. Staging laparoscopy remains a key tool for detecting radiologically occult metastasis. In one series, staging laparoscopy upstaged 48% of patients with gastric malignancy, avoided non-therapeutic laparotomy in those with laparoscopically detected metastasis and had no procedure-related complications[22]. More recently, a 653-patient study of newly diagnosed cT1-2N+/T3-4bNx, cM0 gastric or gastroesophageal junction adenocarcinoma found occult peritoneal metastasis in 22.2% at initial laparoscopy[23]. CA72-4 ≥ 6.9 U/mL, sT4a/b stage and minimal ascites were among the associated factors[23]. Patients with occult peritoneal metastasis had significantly worse median overall survival than those without occult peritoneal metastasis (15.9 months vs not reached; P < 0.001)[23]. These findings suggest that CA72-4 may be most useful as a component of a broader peritoneal-risk assessment strategy.

Prognostic evidence for CA72-4 in gastric cancer is supportive but heterogeneous. In 184 gastric cancer patients treated with neoadjuvant chemotherapy and surgery, Sun and Zhang[24] found that pretreatment CA72-4 positivity was associated with shorter median overall survival and remained an independent prognostic factor. Louhimo et al[25] also reported prognostic relevance for preoperative CA72-4. Overall, these studies support the use of CA72-4 for risk stratification, particularly in advanced disease or in patients at risk of peritoneal dissemination. However, variation in assay cutoff, stage distribution, treatment setting and adjustment strategy limit direct translation into treatment decisions.

Early gastric cancer and lymph-node metastasis

The evidence supporting CA72-4 for decision making in early gastric cancer remains limited. In particular, there is limited evidence in the utilisation of CA72-4 for guiding the choice between endoscopic resection and radical gastrectomy in early gastric cancer. A 2026 systematic review and meta-analysis identified lymphovascular invasion, submucosal invasion, tumor size, histology, perineural invasion and ulceration as stronger predictors of lymph-node metastasis[26]. In the same analysis, CA72-4 above 7 U/mL showed only a weak, non-significant association with lymph-node metastasis, with an unadjusted odds ratio 1.33 and a 95% confidence interval of 0.84-2.11[26]. These findings suggest that CA72-4 has insufficient discriminatory value for selecting endoscopic submucosal dissection vs gastrectomy. Its future role, if any, is more likely to lie within validated multimarker prediction models that test whether CA72-4 adds incremental value beyond established clinicopathological variables, including endoscopic morphology, histology, lymphovascular invasion, invasion depth, radiomics and molecular assays.

Esophageal and gastroesophageal junction cancers

CA72-4 has a less established role in esophageal cancer than in gastric cancer. In esophageal squamous cell carcinoma, conventional biomarkers such as squamous cell carcinoma antigen, CEA, cytokeratin fragments and p53 antibody have been evaluated more frequently[20]. Matsumoto et al[20] reported that squamous cell carcinoma antigen had the highest sensitivity among the esophageal squamous cell carcinoma markers assessed. Although its overall sensitivity remained limited, CA72-4 did not emerge as the central marker in this disease. Current evidence therefore does not support CA72-4 as a stand-alone diagnostic or surveillance biomarker for esophageal squamous cell carcinoma.

More recent biomarker-panel studies have begun to include CA72-4, suggesting a possible role in integrated risk models rather than isolated interpretation. Pang et al[27] developed an integrated biomarker signature incorporating CA72-4, VEGF-C and the pepsinogen I/II ratio to predict lymph-node metastasis after R0 resection in esophageal squamous cell carcinoma. The model outperformed the American Joint Committee on Cancer staging in their cohorts and generated treatment-response hypotheses related to programmed death-1 therapy[27]. These findings are substantial, but external validation with calibration across independent populations and comparison with simpler clinicopathological models are needed before CA72-4 can be considered clinically actionable in esophageal squamous cell carcinoma.

Gastroesophageal junction adenocarcinoma requires separate consideration as its biology and biomarker behavior may overlap more closely with gastric cancer than with esophageal squamous cell carcinoma. Future studies should therefore report tumor location and histology separately. Analyses that combine gastric cancer, Siewert II adenocarcinoma and esophageal squamous cell carcinoma risk obscuring disease-specific biomarker performance and may overstate the generalizability of CA72-4 across upper gastrointestinal cancers.

Colorectal, appendiceal and pseudomyxoma peritonei

CEA remains the dominant conventional serum marker in colorectal cancer. CA72-4 has been evaluated as an adjunctive diagnostic marker, but its low sensitivity limits its value as a screening or stand-alone diagnostic test[28]. More recent studies place CA72-4 more appropriately within multimarker and molecular diagnostic models rather than as an independent colorectal cancer biomarker. In a 2026 case-control study, fecal SDC2 methylation alone achieved a sensitivity of 86.2% and specificity of 96.7% for colorectal cancer. The addition of serum CEA and CA72-4 increased sensitivity to 96.9%, but reduced specificity to 78.3%[29]. Similarly, in a 636-participant ctDNA methylation study, a 21-marker methylation model achieved sensitivity 87.82%, specificity 91.88% and accuracy 89.85% for combined detection of advanced adenomas and colorectal cancer, whereas CA72-4 alone achieved an accuracy of only 55.66%[30]. For stage stratification, the methylation model achieved an accuracy of 83.42%, compared with 46.17% for CA72-4[30]. CA72-4 has also been incorporated into radiomics-based precision models. In colon cancer, age, pericentric lymph-node metastasis and CA72-4 were identified as significant clinical predictors of mismatch repair deficiency or microsatellite instability status. A combined clinicopathological-radiomics nomogram achieved an AUC of 0.94 in the training set and 0.91 in the validation set[31]. These data support the view that CA72-4 may have value as a covariate in multimodal colorectal models rather than a primary colorectal biomarker.

CA72-4 may have a more coherent biological rationale in mucinous and peritoneal disease phenotypes. Appendiceal mucinous neoplasms and pseudomyxoma peritonei require integrated interpretation of tumor markers, imaging and operative findings. In a 2025 radiomics study of appendiceal pseudomyxoma peritonei, CA72-4 was included among clinical variables, although CA19-9 and radiomic features were more prominent in the final interpretability analysis[32]. At present, CA72-4 is best regarded as a secondary component of multimodal assessment in appendiceal and peritoneal mucinous disease rather than a decisive marker for diagnosis, staging or management.

Pancreatic cyst fluid and pancreaticobiliary disease

CA72-4 and TAG-72 have a limited historical role in pancreatic cyst-fluid analysis. Alles et al[33] measured CA72-4 in aspirates from pancreatic cystic lesions and reported utility for distinguishing mucinous cystic tumors from benign neoplasms and pseudocysts. A prior synthesis also noted that cyst-fluid CA72-4 above 40 U/mL had high negative predictive value for malignant intraductal papillary mucinous neoplasm in a small study[1]. These findings should be interpreted against contemporary pancreatic cyst practice, which relies mainly on cross-sectional imaging, EUS morphology, cytology when adequate fluid is available, cyst-fluid CEA or glucose for mucinous differentiation, and molecular testing in selected cases[34,35]. These tools address clinically relevant decisions such as mucinous lineage, high-grade dysplasia, invasive carcinoma and need for surveillance or resection. Against this background, CA72-4 is therefore best framed as an older cyst-fluid adjunct with biological interest, rather than a routine marker that currently changes pancreatic cyst management.

CA72-4 also has no established routine role in gallbladder cancer. This limitation is consistent with the broader gallbladder malignancy literature, where reliable diagnostic biochemical markers remain lacking and diagnosis often depends on imaging followed by histology after resection[36]. CA72-4 may appear in exploratory biomarker panels, but current evidence does not support its stand-alone diagnostic, staging or surveillance use in gallbladder malignancy.

Malignant effusions, ascites and cytology

Fluid-based CA72-4 testing may be more clinically defensible than population serum screening because pre-test probability of malignancy is higher. In this setting, the clinical question is usually whether a pleural or peritoneal fluid collection is malignant, whether cytology has missed disease, or whether a marker panel can strengthen diagnostic confidence. A 2025 meta-analysis of eight studies including 828 malignant and 963 benign pleural effusions reported a pooled sensitivity of 0.47, specificity of 0.98 and summary AUC 0.77 for pleural-fluid CA72-4, although publication bias and threshold-selection concerns were noted[37]. Clinically, this profile supports pleural-fluid CA72-4 as a rule-in adjunct when positive, but a negative result cannot exclude malignancy.

Ascitic-fluid tumor markers have also been studied. Jain et al[38] evaluated serum and ascitic levels of CEA, CA19-9 and CA72-4 in ascites and reported that CA72-4 helped identify peritoneal carcinomatosis in their cohort. Gastric cancer-associated malignant ascites may represent a more coherent setting for fluid CA72-4 than nonspecific ascites. In a study comparing 138 gastric cancer ascites samples with 64 cirrhotic ascites samples, ascitic CA72-4 had a sensitivity of 79.0% and specificity of 100.0%, while combined CEA and CA72-4 achieved a sensitivity 86.2% and specificity 100.0%[39]. These findings are consistent with the broader peritoneal-disease signal seen in gastric cancer. TAG-72/B72.3 also remain relevant in effusion cytology. A recent review of diagnostic markers in effusion cytology identified TAG-72/B72.3 among robust markers used to differentiate mesothelioma from metastatic carcinoma, alongside markers such as claudin-4, calretinin and HEG1[40]. Older immunohistochemical work similarly identified B72.3 among useful negative markers for differentiating epithelial peritoneal mesothelioma from papillary serous carcinoma involving the peritoneum[41]. Together, these data support targeted fluid-based testing in high pre-test probability settings, integrated with cytology, imaging and clinical findings.

Benign elevations and false positives

False-positive results are a major limitation of CA72-4 testing. Similar to other tumor markers, CA72-4 may be elevated in benign disease, inflammatory states, organ dysfunction or because of analytical interference. A 2024 review of tumor-marker concentrations in the absence of neoplasia included CA72-4 among commonly used markers that may be affected by non-malignant conditions[42]. Disease-specific gastric data reinforce this caution. A 2025 study comparing chronic non-atrophic gastritis, gastric ulcer and gastric cancer found higher CEA, CA19-9 and CA72-4 levels in gastric cancer than in chronic gastritis or gastric ulcer, with additional differences in pepsinogen and gastrin-17 patterns. Combined detection achieved an AUC of 0.826[43]. These findings support combined serological interpretation, but also illustrate why CA72-4 should be interpreted in relation to benign gastric pathology rather than treated as a cancer-specific signal.

Tissue studies further reinforce this caution. Thor et al[44] examined 117 paraffin-embedded colonic specimens from 56 patients with ulcerative colitis, together with 11 inflammatory controls, and found increasing TAG-72/B72.3 cellular reactivity across reactive atypia, low-grade dysplasia, high-grade dysplasia, carcinoma and advanced/invasive malignant lesions. Mean cellular reactivity was 23.7% in reactive atypia, 26.5% in low-grade dysplasia, 36.7% in high-grade dysplasia, 70.0% in carcinoma and 84.3% in advanced or invasive malignant lesions. Active ulcerative colitis also showed significantly more TAG-72-immunoreactive cells than quiescent disease[44]. These data suggest that inflammation and premalignant epithelial change can produce TAG-72 expression without invasive carcinoma, limiting its specificity for cancer detection in inflammatory gastrointestinal disease. Clinically, a mildly elevated isolated CA72-4 result in an asymptomatic low-risk patient should therefore prompt confirmation, assay review and clinical assessment before extensive investigation.

Practical clinical interpretation

Several factors help explain the inconsistent performance of CA72-4 across studies. First, assay platforms and cutoffs vary, limiting direct comparison between cohorts. Second, many studies combine biologically distinct populations, including early and advanced disease, intestinal and diffuse histology, gastric and gastroesophageal junction tumors, or serum and fluid samples. Third, case-control designs may overestimate diagnostic performance compared with real-world screening or surveillance populations. Fourth, benign gastric, inflammatory and ulcerative conditions may increase CA72-4 or TAG-72 expression, reducing specificity in low-pretest-probability settings[42-44]. Fifth, multimarker models are often reported without external validation, calibration or decision-curve analysis. These limitations explain why CA72-4 may appear useful in selected datasets but remain difficult to translate into routine management.

Clinicians should interpret CA72-4 through four practical questions. First, why was the test ordered? A result obtained for monitoring known gastric cancer has a different meaning from one obtained during nonspecific health screening. Second, what is the sample type? Serum, pleural fluid, ascites, pancreatic cyst fluid and tissue staining require different thresholds and different clinical reasoning. Third, what is the pre-test probability? Symptoms, known malignancy, abnormal imaging, anemia, weight loss, ascites, pleural effusion or suspected peritoneal disease increase the clinical significance of a positive result. Fourth, what management decision follows from the result? A marker that does not change endoscopy, imaging, staging laparoscopy, cytology review, treatment selection or surveillance interval has limited clinical value.

A reasonable pathway for unexpected serum CA72-4 elevation begins with confirmation. Repeat testing, review of the assay platform and an assessment for symptoms or abnormal imaging should precede extensive investigations. Persistent, rising or markedly elevated CA72-4 in a patient with gastric symptoms, previous gastric cancer, ascites, pleural effusion or peritoneal concern should prompt targeted evaluation. In contrast, mild isolated elevation in an asymptomatic low-risk person should not trigger a blind whole-body investigation. CA72-4 should not be used for population screening. Overall, the clinical utility of CA72-4 and TAG-72 is best understood as context-dependent and adjunctive rather than a stand-alone marker (Figure 3). Its application cluster around gastric cancer, peritoneal disease assessment, malignant pleural effusion work-up, cytology and immunohistochemistry, and selected multimarker panels. Across these settings, CA72-4 is most useful when it refines a defined clinical question, and should therefore be interpreted alongside clinical findings, imaging, endoscopy, cytology, histology and other tumor markers, rather than used in isolation.

Figure 3 Clinical utility of tumor-associated glycoprotein 72 and carbohydrate antigen 72-4 across clinical contexts. CA: Carbohydrate antigen; TAG-72: Tumor-associated glycoprotein 72; CEA: Carcinoembryonic antigen.

THE FUTURE

CA72-4, precision medicine and personalized risk assessment

Precision medicine should test whether a biomarker improves a specific clinical decision. For CA72-4, plausible applications include selection for staging laparoscopy, interpretation of cytology-negative ascites or pleural effusion, recurrence assessment after treatment, and integration into multimarker models for peritoneal disease risk. Future studies on CA72-4 should therefore report clinical utility rather than diagnostic accuracy alone. Useful endpoints would be to include improved calibration, net benefit, incremental AUC, selection for staging laparoscopy, repeat cytology yield, biopsy-site selection, detection of actionable recurrence, avoidance of non-therapeutic laparotomy or change in treatment intent. Sensitivity, specificity and AUC are insufficient if the result does not alter a decision. This is especially important for gastric cancer, cytology-negative ascites, malignant pleural effusion and post-treatment surveillance, where the value of a marker depends on whether it changes the next clinical step. Future CA72-4 studies should report whether testing improves selection for staging laparoscopy, repeat cytology yield, biopsy-site selection, detection of actionable recurrence, treatment-intent change, avoidance of non-therapeutic laparotomy, calibration and decision-curve net benefit.

Gastric cancer provides the clearest opportunity for this approach. Tian et al[45] reported that a combined model using serum miR-181, miR-652 and CA72-4 achieved an AUC of 0.917 for early gastric cancer, with sensitivity of 92.5% and specificity of 86.8%. miR-181 and miR-652 also correlated positively with CA72-4 levels. These findings support the development of combined molecular and conventional marker models rather than isolated CA72-4 testing. Peritoneal metastasis is another clinically important setting as it changes treatment intent, yet conventional imaging may underestimate occult peritoneal disease. In a Singapore series of patients with gastric malignancy, staging laparoscopy upstaged 48% of selected patients and avoided non-therapeutic laparotomy in those with laparoscopically detected metastasis[22]. CA72-4 research should therefore move beyond reporting marker positivity and test whether the marker improves selection for laparoscopy, peritoneal cytology, biopsy, or treatment-intent change.

Fluid-based CA72-4 also fits a personalized-risk framework. Yang et al[39] reported that ascitic CA72-4 had 79.0% sensitivity and 100.0% specificity for distinguishing gastric cancer-associated malignant ascites from cirrhotic ascites, while combined CEA and CA72-4 achieved 86.2% sensitivity and 100.0% specificity. These findings require validation, but they illustrate a clinically coherent use case: CA72-4 may be more useful in a high-pretest-probability fluid compartment than as a low-pretest-probability serum screening marker.

The colorectal literature highlights the competitive pressure on conventional markers. In a 636-participant ctDNA methylation study, the methylation model achieved sensitivity of 87.82%, specificity of 91.88% and accuracy of 89.85% for combined detection of advanced adenomas and colorectal cancer. By comparison, CA72-4 accuracy was 55.66% for detection and 46.17% for stage modeling[30]. Recent colorectal biomarker reviews increasingly position conventional markers such as CEA, CA19-9 and TAG-72 alongside ctDNA, circulating tumor cells, circulating microRNAs, methylation assays and other molecular biomarkers[46]. This narrows the question for CA72-4: It must demonstrate incremental value within disease-specific and sample-specific models, rather than rely on association alone. Recent work has also argued that a subset of hepatocellular carcinoma may be understood through microbial ecology and microbiome-conditioned tumor biology[47]. Future studies should therefore connect the marker expression to the tumor phenotype, mucin glycosylation, peritoneal dissemination, fluid-compartment biology and therapeutic vulnerability, rather than treat CA72-4 as a generic cancer signal.

TAG-72 as an imaging and therapeutic target

TAG-72 has long been investigated not only as a source of circulating antigen, but also as a target for tumor localization and therapy. Radiolabeled B72.3 and CC49 constructs established TAG-72 as a target for tumor localization, with subsequent antibody engineering aimed at improving tumor penetration, blood clearance and immunogenicity[7]. Clinical translation has been explored most clearly in colorectal cancer. In a phase I pre-targeted radioimmunotherapy study, nine patients with TAG-72-positive metastatic colorectal cancer received a genetically engineered CC49 fusion protein, followed by a synthetic clearing agent and radiolabeled DOTA-biotin[48]. More than 95% of circulating CC49 fusion protein was cleared within 6 hours after the clearing agent. Radiolabeled DOTA-biotin localized rapidly to tumor sites, and the mean tumor-to-marrow radiation dose ratio was 139:1[48]. No infusion-related, renal, hepatic or hematological toxicities were reported[48]. These data support the feasibility of TAG-72-directed localization, although they also illustrate that technical targeting success has not yet translated into routine oncological use.

Modern CC49 engineering has attempted to overcome the developability barriers associated with TAG-72 directed antibodies. Lin et al[49] identified aggregation-prone motifs in the CC49 light-chain variable domain and generated engineered CC49 IgG mutants with improved thermal stability, retained binding performance, nearly 15-fold greater solubility, and 97% purity, compared with 70% for the parent molecule at 0.3 mg/mL. Cellular therapy has also renewed interest in TAG-72. A phase I study of TAG-72-specific CAR-T cells in metastatic colorectal cancer reported feasibility and relative safety, but limited persistence and efficacy[50]. Reviews of ovarian cancer immunotherapy similarly list TAG-72 among potential CAR-T and CAR-NK targets, while emphasizing the persistent barriers of solid-tumor treatment, including antigen heterogeneity, poor trafficking, an immunosuppressive tumor microenvironment and toxicity risk[51]. Serum CA72-4 elevation does not prove tissue TAG-72 accessibility. TAG-72 directed cellular therapy should therefore be regarded as an emerging exploratory strategy rather than a near-term routine treatment. Targetability requires tissue expression, antigen density, membrane accessibility, tumor penetration, limited off-tumor expression, manageable soluble antigen sink and proof of therapeutic index.

The future of TAG-72 targeting may therefore lie in selected regional or compartment-based strategies rather than broad systemic treatment. Peritoneal disease is a plausible setting because regional delivery, high tumor burden and accessible fluid compartments may improve therapeutic exposure. However, this remains a research hypothesis rather than an established clinical application. Engle et al[52] showed in mice that CA19-9 can promote pancreatitis and pancreatic cancer, demonstrating that a glycan tumor marker can have functional biology rather than merely reflect tumor burden[52]. Comparable mechanistic work is needed for TAG-72-positive gastrointestinal and peritoneal disease. Future models should test whether TAG-72-positive clones show greater survival in suspension, mesothelial attachment, immune escape or sensitivity to antibody-directed therapy.

Future research

CA72-4 deserves reappraisal as clinicians continue to encounter it in practice and recent studies continue to include it in diagnostic, prognostic and multimarker models. CA72-4 is most informative in tumors with mucin-glycoprotein biology, serosal spread, ascites, cytology-positive disease or occult peritoneal metastasis. This links gastric cancer, ascites, pseudomyxoma/peritoneal mucinous disease, pleural effusion and TAG-72 targeting into one biological theme. This is especially relevant in gastric cancer, where peritoneal disease may be underestimated by conventional imaging and where its detection can change treatment intent. A high or rising CA72-4 in a patient with gastric cancer should therefore prompt careful assessment for peritoneal disease, particularly when supported by CA125 elevation, symptoms, imaging findings, ascites or cytology.

The next phase of CA72-4 research should focus on clinical decisions rather than marker positivity. In newly diagnosed gastric or gastroesophageal junction adenocarcinoma, studies should test whether CA72-4 improves prediction of occult peritoneal metastasis or positive cytology beyond computed tomography findings, tumor stage, histology, ascites and CA125. A useful prospective design would collect serum CA72-4, CEA, CA19-9, CA125, computed tomography features, endoscopic findings, histology, radiomics, peritoneal cytology and staging-laparoscopy findings, with occult peritoneal metastasis or positive cytology as primary outcomes. Such a design would test whether CA72-4 improves personalized staging rather than merely correlating with advanced disease.

Fluid-based CA72-4 also warrants more focused study. Pleural-fluid data suggest high specificity, but low sensitivity limits its use as an exclusionary test. Ascitic-fluid studies may be more informative if they focus on defined cancer types, cytology-negative cases and decision pathways. The clinically relevant question is whether CA72-4 helps avoid diagnostic delay, select patients for staging laparoscopy, support repeat cytology, guide biopsy sites or initiate appropriate oncological staging. Surveillance studies should also evaluate same-platform CA72-4 kinetics after treatment, using recurrence, treatment-intent change, non-therapeutic laparotomy avoided and survival as outcomes. These endpoints would move the field from marker association to clinical consequence.

In early gastric and colorectal cancer, molecular diagnostics, ctDNA methylation, endoscopic artificial intelligence, radiomics and validated risk scores are likely to outperform conventional carbohydrate markers in early gastric and colorectal cancer, particularly for population-level screening. CA72-4 may still contribute as an accessible variable within multimarker and multimodal risk panels, but its inclusion should be justified by demonstrable incremental value beyond simpler clinical variables and newer molecular tests. Future studies should therefore compare models with and without CA72-4 and assess not only discrimination, but also calibration, external validity and decision-curve net benefit.

CA72-4 currently has a more established clinical role as a serum and fluid biomarker, particularly in selected gastric cancer, peritoneal disease and malignant effusion settings. TAG-72 may broaden future applications through cytology, immunohistochemistry, molecular imaging and cellular therapy, where the antigen is used more directly than in serum marker measurement. Although regional delivery and patient selection may improve the therapeutic index in peritoneal disease, solid-tumor targeting still faces major barriers, including antigen heterogeneity, poor trafficking, immunosuppressive microenvironment, toxicity and lack of proven clinical benefit. The field should avoid confusing biological detectability with clinical utility. Researchers should define disease-specific thresholds, report assay platforms, separate serum, fluid and tissue applications, and test management consequences. Clinicians should use CA72-4 only to refine a defined diagnostic, staging or surveillance question rather than as a stand-alone test. This approach preserves the useful aspects of an older marker while protecting patients from low-value investigation.

CONCLUSION

CA72-4 has its strongest evidence base in gastric cancer, particularly in prognosis, recurrence assessment and evaluation of peritoneal dissemination. Combined serum marker panels may improve sensitivity, but they do not replace endoscopy, imaging, staging laparoscopy or cytology. Fluid-based CA72-4 has a stronger rationale in selected high-pretest-probability settings such as in malignant pleural effusion and ascites, where high specificity can support a rule-in interpretation. TAG-72 broadens the field beyond serum testing into cytology, immunohistochemistry, molecular imaging and therapeutic targeting. Future work should define where CA72-4 changes clinical decisions, rather than merely where it is detectable or statistically associated with disease.