Pancreatic Cancer Insights: Optimization of the Diagnostic Capacity of Tumor Biomarkers

Jose Antonio Delgado, PhD; Maria Antonieta Ballesteros, PhD; María Magdalena Parera, PhD; Josep Miquel Bauça, PhD


Lab Med. 2021;52(6):550-557. 

In This Article

Abstract and Introduction


Objective: Pancreatic cancer (PC) is one of the deadliest malignancies. The aim of this study was to determine the usefulness of the carbohydrate antigen 19.9 (CA19.9)/carcinoembryonic antigen (CEA) ratio as a diagnostic tool.

Methods: This was a retrospective observational study (2015–2019), including laboratory requests with increased CA19.9 and CEA but no previous neoplasia. Receiver operating characteristic (ROC) curve analyses were performed for the CA19.9/CEA ratio and for CA19.9 and CEA alone for the detection of PC, and cutoff values for all strategies were selected separately and in combination.

Results: A total of 373 individuals were included. The area under the curve (AUC) for CA19.9/CEA was 0.872, whereas the AUC for CA19.9 was 0.847 and for CEA was 0.554. Cutoff values with the greatest diagnostic power were CA19.9/CEA >40, CA19.9 >1130 U/mL, and CEA > 14.5 U/mL. The combination of CA19.9/CEA > 40 with CA19.9 > 550 U/mL maximized the diagnostic accuracy for PC.

Conclusion: Our results highlight the relevance of the measurement of serum CA19.9 and CEA in the detection of PC.


The main functions of pancreatic cells include the synthesis and secretion of digestive enzymes, bicarbonate, and different types of hormones. Pancreatic cancer (PC) is one of the malignancies with the highest mortality rate, despite its low prevalence.[1,2] Most PCs are adenocarcinomas, although low prevalent neoplasia has also been described, such as neuroendocrine tumors (producing insulin or glucagon) or acinar carcinomas (producing digestive enzymes). It is estimated that only 20% of pancreas carcinomas are surgically treatable upon diagnosis and among these, only half are truly resectable.[3] There are large differences in incidence between regions worldwide, highlighting the fact that lifestyle variables (smoking habit, alcohol, diet) and environmental factors could influence such variation.[4] In Europe, PC has an approximate incidence of 78,000 diagnoses/year (8,169 new diagnoses in Spain in 2019), representing the third cause of cancer deaths in Spain in 2018, with a total of 7,132,[5] and it is expected to become the second cause of death from cancer in the next decade, if the incidence does not change.[6] One of the main causes of such a poor prognosis for PC is its detection at advanced stages, especially because of the lack of specific symptoms, the absence of sensitive and specific tumor biomarkers, and the difficulties in performing imaging studies in early stages.[2]

The 2 main risk factors for the development of PC are age, with a maximum number of diagnoses between the seventh and eighth decades of life, and smoking habit, with a clear dose-response relationship.[7] Obesity and physical inactivity have also been associated with the development of PC, alongside with saturated fat–rich diets and a strong alcohol habit. Likewise, diabetes mellitus has been reported both as cause and as consequence in the first stages of PC.[8]

Icterus is one of the characteristic clinical signs of PC, along with blood glucose decompensations and abdominal pain originating in the stomach zone and radiating to the back. Patients often refer to weight loss in previous months and a loss of appetite for no apparent reason. Most symptoms are rather unspecific, which hinders the diagnosis of PC.

No population-based screening strategies exist to date, so diagnosis is established by means of imaging studies and, ultimately, by means of a tissue biopsy (gold standard). Unlike other pathologies, simple, economic, and low-invasive methods for the early detection of PC are still lacking.

Tumor markers (TM) are different molecules (eg, carbohydrates, proteins, peptides) produced or induced by the tumor or by tissues in response to the presence of the tumor, and they can be detected and quantified in serum, biological fluids, and tissues.[9] Carbohydrate antigen 19.9 (CA19.9), also known as sialyl Lewis A, is the recommended TM in case of suspicion of malignancy of the gastrointestinal tract, with special relevance for a pancreatic orientation. Nevertheless, different sources of increased CA19.9 values exist in the absence of malignancy, including several benign pathologies (biliopancreatic, pancreatitis, cholangitis, choledocolithiasis, and kidney disease).[10,11] This fact stimulated the publication of multiple studies in which cutoff values for CA19.9 were examined, aiming to increase the capacity of this biomarker to differentiate between benign pathologies and neoplasia.[12,13] In addition, it has long been accepted that patients who are Lewis-negative (approximately 5% of the population) do not express this antigen, leading to false-negative CA19.9 values. However, Luo, Fan, et al[14] recently reported that up to 27.4% of patients who were Lewis-negative with PC had CA19.9 >37 U/mL. Pairing this together with the fact that only 65% of patients with resectable PC have elevated levels of CA19.9 in serum,[15] different authors have advised against the use of CA19.9 in the differential diagnosis of PC, although great discrepancies exist.[16–18]

Another TM used in cancer screening is carcinoembryonic antigen (CEA), a glycoprotein produced during fetal development and found in concentrations <5 ng/mL in the blood of healthy adults. This marker was the first human antigen associated with the presence of colon cancer in 1965[19] and has been widely researched as a diagnostic marker for other types of cancer.[20–24] Nevertheless, as is the situation with other tumor markers, CEA is not free from false-positive results. Some studies have suggested that values between 5 ng/mL and 10 ng/mL have a high probability of representing false-positive results as a consequence of comorbid situations, such as diabetes, smoking habit, or colorectal polyps.[25,26] Along these lines, some authors have reported false-positive values for CEA up to 20 ng/mL in patients with kidney or liver disease.[27]

The usefulness of CEA in the diagnosis of PC is still under controversy, as happens with CA19.9. Extensive research has been done on the diagnostic value of CA19.9 and CEA for PC, with inconsistent conclusions. Xing et al[28] reported that CA19.9 has a higher diagnostic value than CEA, except for specificity, which opens the door to the need for a combination of both biomarkers (CEA and CA19.9) to improve their diagnostic value and to differentiate their levels within PC stages.

The main advantages of serum TMs are the ease and readiness of their quantification (immunoassays are available in almost all clinical laboratories worldwide), their noninvasiveness (they are measured in blood or, at low frequency, in urine), and their low cost. Therefore, their implementation is healthcare is widely extended.

As mentioned above, given the nonspecific clinical symptomatology of PC, a differential diagnosis with other digestive neoplasia is usually necessary. Considering that 85% of PCs are ductal adenocarcinomas,[29] some authors have included additional TM in their diagnostic algorithms, such as CEA) secretory mucin 5AC, or carbohydrate antigen 125.[30,31]

The aim of this study was to determine the usefulness of CA19.9 in combination with CEA (CA19.9/CEA ratio) as a diagnostic tool for the differentiation of PC from benign pathology (cirrhosis, acute or chronic pancreatitis, cholangitis, choledocolithiasis) or other types of cancers in our hospital.