This study aimed to investigate the efficacy of combining CA19-9 and tumor size to predict survival in pancreatic cancer. Tumor size alone did not predict survival group in our study, though CA19-9 was associated with survival. When CA19-9 is analyzed within tumor size groups, it strongly predicts survival in patients with small (≤ 2 cm) and large tumors (> 4 cm). This relationship persisted after adjusting for several important clinical variables including age, tumor location, neoadjuvant status, total bilirubin level and the resection’s marginal status.
CA19-9 has been extensively investigated in pancreatic cancer and previous reports have analyzed CA19-9 with regards to tumor stage, resectability, response to chemotherapy, and risk of recurrence after surgery.[6–12] These studies, while adjusting for tumor size, assumed a linear relationship between CA19-9 and survival or CA19-9 and tumor size; however this presumption does not appropriately account for variable tumor biology, as both small aggressive tumors and large indolent tumors may cause high CA19-9 levels. By combining tumor size with this biochemical marker, we may be able to improve the previous prognostic capabilities for PDAC in the adjuvant setting.
To link our findings between CA19-9, tumor size, and survival with underlying cellular mechanisms, a brief discussion of several processes is helpful. During neoplastic transformation, there is a change in the cellular surface glycans due to the incomplete synthesis of complex carbohydrates.[17–19] One such alteration occurs due to transcription downregulation of the genes encoding for the alpha 2–6 sialyltransferase enzyme. This results in the increased expression of sialyl Lewis A (CA19-9) molecule, a simple and non-functional glycan, at the expense of the disialyl Lewis A molecule, a ligand molecule essential for immune recognition and homeostasis.[20] This shift in cellular expression of the carbohydrate antigen signature molecule leads to loss of cell-to-cell recognition between mucosal epithelial cells and lymphoid cells, helping the pancreatic ductal cancer cells to evade immune recognition.[20, 21] The CA19-9 molecule is also related to malignant tumors’ ability to survive in a hypoxic environment. As neoplastic tissue adapts to hypoxia, there is increased expression of genes regulated by the Hypoxia-Inducible Factor-1.[22] Among these upregulated genes are those responsible for the glycosylation of CA19-9, leading to its increased synthesis.[23, 24] Thus, CA19-9 is a surrogate marker for degree of tumor aggression, with regards to either the tumor's ability to evade immune recognition or to survive in a hypoxic environment. Therefore, to assess the tumor aggression based on the pathophysiology detailed above, it would be appropriate to adjust CA19-9 levels for tumor size.
Clinically, small tumors with dramatically elevated CA19-9 levels may indicate an aggressive tumor unlikely to achieve cure even with an R0 resection due to the inevitable circulating tumor cells.[25] Alternatively, large tumors with potentially more neoplastic cells yet secreting relatively lower levels of CA19-9 indicate a less aggressive disease process that is more susceptible to adjuvant treatment and prolonged survival. In similar vein, high CA19-9 associated with a small tumor could indicate that more tissue is involved in the neoplastic process, either local or distant, than what is radiologically visible. Rieser et al. reported that when following PDAC patients, a post-resection increase in CA19-9 preceded the radiological evidence of cancer recurrence by a median duration of six months.[26] This may additionally mean that small tumors with elevated CA19-9 are more likely to be pathologically upstaged during resection than an otherwise larger tumor with low CA19-9 levels. For example, Olecki et al. found that 80% of clinical stage I PDAC were pathologically upstaged during resection and high CA19-9 was a significant predictor of pathological upstaging.[27] Thus, it is clinically reasonable to use the combination of both size and CA19-9 for prognostication.
The relationship between CA19-9, tumor size, and survival group was not statistically significant for intermediate-size tumors. While the scope of the analysis does not allow for more than speculation, we nevertheless provide a hypothesis that might explain this result. We hypothesized that for this group, this relationship is distorted by high heterogeneity in other variables that directly influence one of CA19-9, tumor size, or survival. For example, the tumor mass-effect, surgical approach, or complexity of dissection is likely to have high heterogeneity in intermediate-size rather than either end of the size spectrum. Similarly, the lead time to diagnosis and subsequently lymph nodes involvement or occult distance metastasis is likely more heterogenous in this group than small tumors that are more uniformly early detected or large tumors that are more uniformly detected late. Finally, smaller number of patients also likely limits the effect size more in a middle group rather than the extremes.
This study is limited by its retrospective nature of patients who were offered a therapeutic resection and the fact that it includes a highly selective cohort of patients from a single center. A particular limitation that pertains to the tertiary nature of our institution is the limited information regarding the perioperative chemotherapy for those whom their local oncologist managed. This precluded the assertion of receiving adjuvant chemotherapy for many patients, who were excluded, and limited the available details of the neoadjuvant treatment regimen for those who received it. Stratification of this selective cohort based on tumor size requires dividing the small cohort into further groups, limiting the number of variables that can be included in the multivariable analysis. FOLFIRINOX and preoperative CA19-9 response to neoadjuvant treatment are examples of such variables. Additionally, some of the CA19-9 levels in this study were obtained in patients with elevated bilirubin; however, introducing total bilirubin as a covariate in the multivariable analysis did not change the results. We were also not able to utilize imaging for all the tumors to determine the size; however, we found a good correlation between tumor size on pathology and imaging reports. Finally, this analysis does not apply to patients’ Lewis blood group of Lea-b-, or the non-secretors.[15, 16] Despite these limitations, the results of this study suggest an important benefit in analyzing CA19-9 levels in the context of tumor size rather than attempting to identify a single clinically meaningful “cutoff value.”