Evaluating a preoperative protocol that includes magnetic resonance imaging for lymph node metastasis in the Cholangiocarcinoma Screening and Care Program (CASCAP) in Thailand

Cholangiocarcinoma (CCA) is a primary hepatic malignancy that arises along the intrahepatic and extrahepatic bile ducts. It is the second most common liver cancer globally [1] and the most common liver cancer in the resource-limited setting of the northeastern provinces of Thailand [2], where its high incidence is due to risk factors unique to this tropical disease, i.e., the consumption of raw or undercooked cyprinoid fish, which are the intermediate hosts for the food-borne trematode Opisthorchis viverrini (Ov), one of the few carcinogenic pathogens [3]. When detected early, CCA is amenable to liver resection, which substantially improves survival [4, 5], with the median survival time of liver resection patients at 23 months compared with less than 8 months in inoperable patients [5]. However, due to the asymptomatic or non-specific nature of the symptoms associated with CCA during its early stages, this cancer is often detected at an advanced stage, with a concomitantly poor prognosis and dismal survival rate [5]. The Cholangiocarcinoma Screening and Care Program (CASCAP) was instituted in Northeast Thailand with the objective of conducting community-based ultrasound (US) screening programs for the early diagnosis of CCA, making liver resection a viable post-diagnostic option for the treatment of CCA [6].

As the objective of CASCAP is to increase survival from CCA by aggressively screening for this cancer by mobile US units in Ov-endemic areas, a critical component of this endeavor was to establish a preoperative evaluation protocol for liver resection for CCA, which would include factors such as (1) tumor infiltration beyond the second order bile duct branches, (2) tumor invasion of major vessels such as main hepatic artery or portal vein, and (3) the extent of lymph node metastasis (LNM) [7‐15]. Due to its location in the hepatoduodenal ligament, CCA quickly metastasizes outside the liver via the lymph node system or perineural invasion [16, 17], even during the early stages of this bile duct cancer. While several methods are available for detecting LNM, CASCAP attempted magnetic resonance imaging (MRI) to scan for LNM as part of a preoperative evaluation protocol for CCA due to its speed and “accuracy” in other cancer contexts [18]. However, the accuracy of MRI for depicting LNM for CCA has to be studied for each individual cancer, and it has been poorly studied of CCA [19‐21]. The assessment of LNM by MRI for CCA is seldom performed due to limited access to tertiary care centers in this resource-poor setting. Herein, we investigated the diagnostic performance of MRI for the detection of LNM, as a part of the preoperative evaluation protocol for CCA, compared to the “gold standard” of pathological confirmation of individuals suspected of CCA by lymph node resection.

A total of 433 individuals suspected of CCA by US were sent for MRI scanning, with a follow-up confirmation of 255 cases of CCA (58.9%). There were 133 (52.2%) individuals with intrahepatic CCA, 112 (43.9%) with perihilar CCA, and 10 (3.92%) with distal CCA. In the CCA group confirmed by MRI scans, 248 (97.3%) subjects also had radiographic lymph node findings, with 125 (50.4%) showing LNM. Of these 255 MRI-confirmed CCA subjects, 130 received surgical treatments with pathologically confirmed CCA. The 51 underwent lymph node dissection, with only 21 (41.2%) having confirmed LNM based on the conventional pathological diagnosis (Fig. 1).

Of the 51 individuals who underwent lymph node dissections, the mean age was found to be 61.5 years (SD 10.3), with 33 out of 51 (64.7%) being males. Of these individuals, 35.3% had intrahepatic, 60.8% perihilar, and 3.9% distal CCA (Table 1).

The MRI scan accurately detected the presence or absence of LNM in 29 out of 51 CCA cases (56.9%, 95% CI 42.2–70.7), resulting in a sensitivity of 57.1% (95% CI 34.0–78.2) and a specificity of 56.7% (95% CI 37.4–74.5), with a PPV and a NPV of 48.0% (95% CI 27.8–68.7) and 65.4% (95% CI 44.3–82.8), respectively. The positive likelihood ratio was 1.32 (95% CI 0.76–2.29), and the negative likelihood ratio was 0.76 (95% CI 0.42–1.36) (Tables 2 and 3).

For intrahepatic CCA, the MRI scan had a sensitivity of 50.0% (95% CI 15.7–84.3) and a specificity of 88.9% (95% CI 51.8–99.7), with a PPV and a NPV of 80.0% (95% CI 28.4–99.5) and 66.7% (95% CI 34.9–90.1), respectively. The positive likelihood ratio was 4.50 (95% CI 0.63–32.4), and the negative likelihood ratio was 0.56 (95% CI 0.27–1.17) (Table 3). In the case of perihilar CCA, the MRI scan had a sensitivity of 61.5% (95% CI 31.6–86.1) and a specificity of 38.9% (95% CI 17.3–64.3), with a PPV and NPV of 42.1% (95% CI 20.3–66.5) and 58.3% (95% CI 27.7–84.8), respectively. The positive likelihood ratio was 1.01 (95% CI 0.57–1.77), and the negative likelihood ratio was 0.99 (95% CI 0.40–2.43) (Table 3).

The age-adjusted incidence rates for CCA in Northeastern Thailand are 89.2 and 35.5 per 100,000 for males and females, respectively, which are among the highest in the world. The high incidence of this relatively rare cancer in Northeast Thailand is due to the dietary habits of the inhabitants of the region, who consume raw or undercooked cyprinoid fish, the intermediate host of the food-borne pathogen Opisthorchis viverrini. CCA tends to present late, often going undetected until an advanced stage, when overall survival is less than 12 months [4, 25‐29]. Currently, the most successful treatment for CCA is radical liver resection at an early stage of the disease and in the absence of LNM. With its location in the hepatoduodenal ligament, CCA tumors tend to metastasize outside the liver either via the lymph node system or via perineural invasion [16, 17]. Health disparities in this region of Thailand exacerbate this poor prognosis, as there are limited early cancer screenings and minimal post-diagnostic treatment. CASCAP was instituted in Northeastern Thailand to resolve these health disparities associated with CCA by conducting community-based US screening to detect early CCA.

The development of a preoperative protocol for CCA is central to the strategy of CASCAP. Currently, the preoperative protocol for CCA includes an (1) extension of the cancer to the hepatic artery or portal vein and (2) extension of lymph node metastasis (LNM) [7‐15]. CCA cases that underwent liver resection in the presence of pathologically confirmed LNM were reported to have a zero 5-year survival rate [30‐33], whereas CCA cases that underwent liver resection in the absence of LNM were reported to have a 35–72% 5-year survival rate [33‐36]. However, the specific outcome of liver MRI on the surgical management of CCA has yet to be systematically evaluated. While several studies have estimated the diagnostic accuracy of MRI as part of the preoperative evaluation of bile duct involvement and vascular involvement [19, 20, 37], our study is the first to have determined the accuracy of MRI in the preoperative evaluation of LNM in CCA.

In the current study, MRIs were observed to have limited utility for preoperative evaluation for CCA compared to the “gold standard” of pathological confirmation of LNM. An especially concerning finding was the high false positive rate of MRIs for CCA LNM, which would exclude individuals otherwise eligible for surgery. We quantified the sensitivity of MRI scans in determining LNM for CCA at 57.1%, which was far lower than other preoperative resectability criteria such as the bile duct and vascular involvement [19, 20, 37]. A factor that may limit the accuracy of MRI scans in the assessment of CCA LNM is the reliance on nodal size in the axial short axis because of the inability to detect microscopic metastasis in normal-size nodes and to distinguish benign enlargement from malignant lymph nodes [12]. Our findings are similar to a recent study that found the survival rate of CCA with preoperative lymph node enlargement improved after hepatectomy [38]. We also agree with a study that recommended routine lymphadenectomy in all cases of hepatectomy [36]. An important limitation of this study is that MRI scanning was carried out at nine different tertiary care centers, using different machines, techniques, and radiologists. This indicates the possibility of variation in the data collection. Second, the lymph node pathology, which is the gold standard of this study, was determined in only some surgical subjects: not all MRI scans confirmed CCA diagnosis. The site and number of lymph nodes from MRI and pathological diagnosis also might be different. This might have an effect on our results. Finally, the time between an MRI scan and surgical operation might have affected the progression of the disease, as it may affect the period during which LNM may occur.

Liver resection has long been shown to increase survival from CCA, though it is highly dependent on LNM. Herein, we utilized the unique early screening program and a large sample size (CASCAP) in Northeastern Thailand to determine the quality of data for MRI scanning to determine LNM in the preoperative evaluation for CCA. We found that MRI scanning for LNM had limited sensitivity, with a low positive predictive value, and a high number of false positives, which would define eligible patients as inoperable. The evaluation of the nodal metastasis based on this radiological imaging modality has significant clinical implications for the surgical treatment of CCA. As such, we strongly recommend a “reconsideration” of the inclusion of MRI results for LNM in the preoperative evaluation for CCA, especially when it is used as the single or principle criterion for whether or not an individual will have curative liver resection for CCA.

MRI has limited sensitivity and positive predictive value for the diagnosis of LNM in CCA. Therefore, surgeons should be aware of the number of false positives in determining the inoperability of CCA subjects.