Raised CA19–9 and CEA have prognostic relevance in gallbladder carcinoma

Patients with suspected GBC managed from October 2013 to April 2017 were reviewed from prospectively maintained database.

Informed patient consent was obtained for this research.

Patients were assessed clinically; biochemical including liver function tests and ultrasonography of the abdomen at 3 months interval during first year and 6 months interval in later years.. A CECT scan of the abdomen was done at 1 year in patients who remained well. It was done earlier if necessitated by clinical or ultrasound findings.

Patients were treated by multimodality approach with gemcitabine-based chemotherapy in both the adjuvant and palliative setting. Any node positive tumor or T1b disease and above patients were subjected to adjuvant chemotherapy. In Palliative setting maximum possible supportive care was given. .

Clinical and pathological details were collected, including age, gender, serum CEA and CA19–9 levels, Resectable or unresectable or metastatic disease status, resection margin status, AJCC staging of the disease, location of metastasis, survival time, survival status, adjuvant or palliative chemotherapy.

CA19–9 and CEA are the most commonly used tumor markers in GBC [9, 12]. The role of these markers in predicting metastasis or unresectability needs attention and evaluation. Finding a serum levels of these markers will help to increase the accuracy of predicting the prognosis of these patients. In a study by Wang et al., value of CA19–9 (but not CEA) increased gradually with progression of clinical stages [9]. The current study demonstrated that CA19–9 value increases as the tumor burden increases (p < 0.001) but the same was not true for CEA levels (p = 0.20). Thus, CA19–9 may be useful in predicting the tumor burden in these patients. In our study the cutoff value of CA19–9 > 72 and CEA > 5 had high specificity but low sensitivity in detection of metastatic disease with a AUROC of 0.67 and 0.62 for CA19–9 and CEA respectively. Liu et al. [13] found cut-off levels CA19–9 of 98.9 IU/ml, an independent predictor of resectability in GBC patients who underwent attempted resection with a sensitivity of 76.3% and specificity of 70.8%. They showed CA19–9 levels may predict R0 resection. Shukla PJ et al. [14] in their analysis of 335 patients, which also included 80 patients with jaundice, found that when CA19–9 levels exceeded 90 IU/ml, 95% patients had unresectable disease. However, they did not calculate the cut-off levels in their study. Although high levels of CA19–9 (> 70–100 IU/ml) have been associated with a high probability of unresectable disease, but one should not deny these patients a chance of surgical resection based on the level of tumor markers alone as surgery is the mainstay for curative treatment in these patients. Literature says PET-CT changes the surgical management in 17–23% of patients with GBC (Petrowsky et al. 2006, Corvera 2008 et al) [ 15, 16]. Therefore we suggest the use of PET-CT in work up alogirthm of patients when theCA19–9 is> 72 and CEA > 5 to detect occult metastasis preoperatively, potentially avoiding surgical exploration, particularly when routine imaging does not show any sign of unresectability.

Wen et al. [11] used a time dependent ROC curve and AUROC for prognostic accuracy of the tumor markers. They found AUROC at 5 years for CA19–9 was 0.77, CEA was 0.76 and for combined markers was 0.79 and concluded that the combination of an elevated CEA and CA19–9 was an independent predictor of poor prognosis in GBC patients undergoing resection.

We found a statistically significant difference in the survival of patients with CA19–9 > 37 vs ≤37 IU/ml and CEA > 4 vs ≤4 ng/ml in univariate analysis along with other tumour characteristics. However both these factors were not significant in determining overall survival on multivariate analysis. The overall 5-year survival of patients with CEA ≤4 and > 4 was 42.8 and 12.5% respectively. Similarly, patients with CA19–9 ≤ 37 had a 5-year survival of 40%, whereas none of the patients with CA19–9 > 37 survived for more than 5 years despite curative surgical resection. These observations suggest that preoperative CA19–9 and CEA levels have a utility in prognostication of patients undergoing curative surgical resection. Agarwal et al. [17] noted a 4-year survival of 78%in GBC patients undergoing extended cholecystectomy (n = 33),when CA19–9 levels were less than 20 IU/ml vs 33% when CA19–9 levels were more than 20 IU/ml. They noted that median survival was not reached when CA19–9 levels were less than 20 IU/ml vs 12 months when CA19–9 levels were more than 20 IU/ml [17].

The elevated preoperative level of CA19–9 has been associated with a high risk of recurrence after curative resection of biliary tract carcinoma in a few published studies including the current study. However, the optimal cut off value of CA19–9 remains controversial. Chung et al. found an optimal cutoff value of preoperative CA19–9 to be 55 IU/ml to predict recurrence in their study [18]. However, Chung et al. included a mix population of all biliary tract cancers in their study (only 20% of patients had GBC) and approximately 40% of patients among them had jaundice. Liu et al. [8] reported a mean value of > 250 IU/ml as cut-off for CA19–9 in predicting recurrence. The higher value reported by Liu et al. could be because they calculated the mean value and not used the ROC curve, or could also be attributed to the differences in study population. Chung et al. [18] found that the recurrence occurred in 61% patients with CA19–9 > 55 compared to25% whenCA19–9 was < 55 IU/ml. In our study, recurrence occurred in 63% patients with CA19–9 > 37 compared to 35% in patients with CA19–9 < 37 reaffirming that high CA 19–9 has significantly higher recurrence after curative surgery. The high preoperative CA19–9 level could also be considered as a criterion for considering patient for adjuvant chemotherapy after curative surgical resection.

In our present study we found CEA > 4, CA19–9 > 37, presence of T3/T4 tumors, node positive disease and presence of perineural invasion were predictors of recurrence. But on multivariate analysis, only CA19–9 > 37 was found to be statistically significant in predicting recurrence after surgery.

CA19–9 was found to be a better tumor marker than CEA in our study, as its median value increases with tumor burden. It has a higher area under ROC for prediction of metastatic disease and is a significant factor in predicting recurrence. Recent studies done to understand the molecular mechanisms of GBC show that the eukaryotic initiation factors (eIFs) monitor protein translational processes via PI3K/AKT/mTOR pathway play a pivotal role in cell growth, proliferation, apoptosis and malignant transformation. Overexpression of eIF6s and Insulin growth factor-2 mRNA binding protein, (IMP2) have been significantly correlated with advanced tumor and shorter survival of gallbladder cancer patients. Thus, besides the tumor marker CEA and CA 19–9, eIF6 and IMP2 may be used as a prognostic biomarker for overall survival in GBC patients and might be used as a potential therapeutic approach in future [19, 20]. Chronic inflammation is risk factor for gall bladder carcinoma. IL-6, a pleiotrophic cytokine is involved in such inflammation and if chronically stimulated will have detrimental effects. However the IL-6 R alpha (interleukin-6 receptor alpha) expression in GBc patients has good prognosis [21].

The strength of this study are that we included complete spectrum (both resectable and unresectable) of GBC patients to derive at the significance level of the tumour marker. We excluded patients of GBC with SOJ as jaundice by itself can erroneously lead to raised CA19–9, confounding the interpretation of data. Our study is limited by its retrospective analysis of data, which is potentially exposed to selection bias. Secondly, Lewis antigen (non secretory group for CA19–9) was not taken in to account and this could have decreased the accuracy of CA19–9 in predicting advanced disease. There was a limited data on postoperative levels of Ca 19.9 and CEA. Hence the difference between preoperative levels and post operative levels could not be calculated.