Prediction of Postoperative Early Recurrence and Prognosis in Pancreatic Cancer Patients

Pancreatic cancer is one of the most lethal malignancies due to its rapid growth, aggressive invasiveness, and early systemic metastasis.¹ Despite recent development in diagnosis and management, the 5-year survival rate of pancreatic cancer patients is only 4%.² Most cases of pancreatic cancer are already advanced by the time of diagnosis; more than 50% are metastatic.³ Even after potentially curative resection, more than 50% of patients develop local or distant metastasis despite adjuvant therapy.⁴ The most common site of distant metastasis is the liver, and calculations based on estimates of tumor doubling times indicate that occult liver metastases may exist in many patients diagnosed as having localized tumor at the time of surgical resection.⁵ Intensive neoadjuvant chemotherapies for both borderline and fully resectable pancreatic cancers have recently been adopted to reduce postoperative recurrence and achieve a more favorable prognosis.^6–9 However, it remains unclear which patient subpopulation would benefit from intensive preoperative therapy. Prediction of early recurrence after surgery is critical for selecting the optimal therapeutic strategy for this formidable disease.

To this end, serum carbohydrate antigen (CA) 19-9 has been investigated in pancreatic cancer, and its roles in diagnosis and therapeutic management have been increasingly elucidated.^10–12 In patients who underwent surgical resection of pancreatic cancer, serum CA 19-9 level has been reported to correlate with tumor resectability, postoperative recurrence, and prognosis.^13,14 Furthermore, several studies reported that changes in serum CA 19-9 levels over the course of chemotherapy or chemoradiotherapy are indicative of the responses of advanced pancreatic tumors to treatment.^15–17 However, serum CA 19-9 can be affected by obstructive jaundice and Lewis blood group antigen. Specifically, biliary obstruction has been reported to increase serum CA 19-9 levels,¹⁸ whereas patients belonging to the Lewis phenotype Le^a−b−, which occurs in 5%–10% of the population, are not able to produce CA 19-9.¹⁹ Therefore, exclusion of patients with obstructive jaundice and Lewis phenotype Le^a−b− is essential to properly investigate the efficacy of serum CA 19-9 level as a prognostic or metastatic predictor in pancreatic cancer patients.

Many previous studies have demonstrated that inflammatory status is associated with prognosis in patients with various malignant tumors.²⁰ Several recent studies demonstrated that pancreatic cancer patients with high CRP levels had poor prognoses.^21,22 However, the cutoff values of for CRP level varied by study, and its significance as a predictor of early recurrence after surgical resection is therefore unclear. Additionally, previous large cohort studies revealed several other prognostic clinicopathologic factors such as lymph node metastasis, pathologic T stage, histologic grade, tumor size, and adjuvant chemotherapy.^23,24 Among these factors, tumor size is potentially evaluable before surgery.

The primary aim of this study was to investigate the feasibility of employing CA 19-9 level, in combination with CRP level and tumor size, for better stratification of pancreatic cancer patients based on their prognosis and the risk of early postoperative recurrence.

Materials and Methods

Results

Association between early recurrence and clinicopathologic factors

As shown in Table 1, larger tumor size, higher CA 19-9 level, higher CRP level, more advanced T factor, and lymph node metastasis were significantly associated with early recurrence after surgery.

Table 1 Association between recurrence within 1 year after surgery and clinicopathologic factors in patients with pancreatic cancer (n = 125)

View Fullscreen

Relationship between preoperative serum CA 19-9 level, CRP level, tumor size, and early recurrence

We analyzed the serum CA 19-9 level while testing several cutoff values to predict early recurrence after surgery (Table 2). A cutoff value of 100 U/mL was the most reliable for predicting early recurrence (sensitivity: 55.6%, specificity: 87.3%, and accuracy: 73.6%). The rate of early recurrence in patients with serum CA 19-9 levels >100 U/mL (76.9%) was significantly higher than in those with serum CA 19-9 levels ≤100 U/mL (27.9%). Next, we analyzed tumor size and CRP level using several cutoff values as predictors of early recurrence after surgery (Table 2). The accuracies of CRP level and tumor size were lower than that of serum CA 19-9 level (60%-plus range). Cut-off values with highest accuracy were 1.0 mg/dL and 3.0 cm for CRP and tumor size, respectively. We assessed the diagnostic significance of tumor size in 81 patients who underwent all 3 imaging modalities (CT, MRI, and EUS). When categorizing tumor sizes into ≤3 cm versus and >3 cm, 72 patients (88.9%) were accurately diagnosed using this parameter.

Table 2 Association between each prognostic predictor and recurrence within 1 year after surgery in patients with pancreatic cancer (n = 125)

View Fullscreen

Prediction of early recurrence and prognosis using serum CA 19-9 level, tumor size, and CRP level

We performed combination analysis using the 3 risk factors (serum CA 19-9 level, CRP level, and tumor size) to achieve better stratification based on the risk of early recurrence after surgery (Fig. 1A). We divided all patients into 3 groups according to their number of positive risk factors. Rates of early recurrence after surgery in patients with 0, 1, and 2 or 3 risk factors were 22.0%, 45.2%, and 91.7%, respectively. Using this method, patients could clearly be stratified by the risk of early recurrence after surgery. Next, we performed survival analysis based on these 3 risk factors (Fig. 1B). There was a significant difference in OS between each patient group, indicating that patients could be readily stratified according to their postoperative prognoses. Median OS in patients with 0, 1, and 2 or 3 risk factors were 50.3 months, 27.3 months, and 9.4 months, respectively.

View Figure

• Download as Powerpoint
• Download Figure

Discussion

Pancreatic cancer produces distant metastases at an early stage; the liver is the most common site of such metastases.⁵ In vivo lineage tracing revealed that entry of pancreatic cancer cells into the bloodstream and seeding into the liver occurs before detectable primary tumor formation.²⁶ In the present study, 54 of the 125 pancreatic cancer patients (43.2%) had postoperative recurrence within 1 year; hepatic recurrence occurred most frequently. Better patient stratification based on the risk of early recurrence and prognosis is critical for devising optimal therapeutic strategies for pancreatic cancer patients. CA 19-9 has become the most important tumor marker for pancreatic cancer since first reported by Koprowski et al in 1979.²⁷ Patients who developed early recurrence after surgery showed higher serum CA 19-9 levels, larger tumor sizes, higher CRP levels, and positive nodal statuses. With regard to prediction of early recurrence, the serum CA 19-9 cut-off value of 100 U/mL demonstrated 55.6% sensitivity, 87.3% specificity, and 73.6% accuracy. Patients with serum CA 19-9 levels >100 U/mL showed significantly higher early recurrence rates compared with those with serum CA 19-9 levels ≤100 U/mL.

Many previous studies reported the clinical utility of serum CA 19-9 in predicting recurrence and prognosis in pancreatic cancer patients after adopting various cutoff values.^10,15,16,28 Optimal cutoff values may be different depending on the type of tumor (resectable, borderline resectable, or unresectable pancreatic cancer) or the purpose (screening, resectability, recurrence, or response to chemotherapies). As mentioned already, the value of CA 19-9 can also be affected by obstructive jaundice and Lewis antigen expression. In this study, the ratio of patients excluded because of CA 19-9 levels <5.0 U/mL (8.8%) is similar to that of the reported percentage of Lewis phenotype Le^a−b− patients (5%–10%).¹⁹

We incorporated 2 additional risk factors, CRP level and tumor size, to achieve better stratification of patients based on the risk of early recurrence after surgery. CRP level and tumor size were previously reported to be prognostic factors in pancreatic cancer patients;^21–24,29 this was reaffirmed in our study. With regard to prediction of early recurrence after surgery, cutoff values of 1.0 mg/dL for CRP and 3.0 cm for tumor size showed highest accuracy. By using the 3 risk factors, serum CA 19-9 level (cutoff value: 100 U/mL), CRP level (cutoff value: 1.0 mg/dL), and tumor size (cutoff value: 3.0 cm), we divided the patients into 3 groups based on the number of the containing risk factors. This classification could stratify the pancreatic cancer patients more clearly based on not only early postoperative recurrence but also the OS. Microenvironments with dense, desmoplastic stroma rich in inflammatory cells, fibroblasts, and extracellular matrix proteins are traits of pancreatic cancer.³⁰ The inflammatory response to tumor cells, reflected by an elevated serum CRP level, may produce proinflammatory cytokines, angiogenic and lymphogenic factors, and chemokines that lead to tumor growth, angiogenesis, and metastasis.²¹

There were several limitations in this study. First, because of its retrospective nature, the treatment strategy was not uniform. Patients not subjected to adjuvant therapy tended to experience more frequent early recurrence. Second, it is difficult to judge the tumor size accurately before surgery. However, by using EUS in addition to CT and MRI, whether the tumor was larger or smaller than 3 cm was judged accurately in a majority of patients (88.9%). Despite these limitations, the present study demonstrated the ability to stratify patients based on the risk of early recurrence after surgery. Such reliable stratification may be useful to determine the treatment strategy, including intensive neoadjuvant therapies, for pancreatic cancer patients. Patients with a high risk of early recurrence ought to be administered intensive systemic chemotherapy before surgery, even those with resectable tumors.

In conclusion, our findings demonstrated that serum CA 19-9 level, together with CRP level and tumor size, improved the ability to stratify patients based not only on early recurrence, but prognosis as well.