A total of 218 patients diagnosed with BTCs during January 2015 to July 2020 in our hospital were collected. Sixty-one patients with LNM diagnosed at the first examination were identified as the study cohort, including 15 patients with GBC, 35 patients with iCCA and 11 patients with eCCA. There are 31 patients with PALN in study cohort (Table1). 45 patients with postoperative recurrence were collected, including 12 GBC, 18 iCCA and 15 eCCA. Among them, 21 patients with intrahepatic recurrence only, 7 patients with distant organ recurrence (peritoneum, bone, muscle) and 19 patients with lymph node recurrence (Table2). 19 patients with postoperative lymph node recurrence were selected as validation cohort (Table1). There were 14 patients with PALN recurrence in the validation cohort.
Distribution of positive lymph nodes in study cohort and validation cohort
A total of 251 lymph nodes were identified in study cohort. These lymph nodes mainly concentrated in region 16a2 (13%) and 16b1 (17%), followed by region 12 (12%) and 8 (9%). Instead, regions 9, 10, 11, 13, 14 , 16a1, 16b2 , 17 and 112 showed only slight risk of lymph node involvement. And other regions, including 7,111, mediastinal region and left supraclavicular region , showed moderate risk of lymph node invasion. According to different tumor locations, subgroup analysis showed that region 16b1 was the most common area of LNM in both iCCA and GBC, with the metastasis rates of 17% and 18%, respectively, followed by 16a2 (11%), 111 (11%) and mediastinal region (11%) in iCCA, left supraclavicular region (17%) and 12 (15%) in GBC. In eCCA, the highest involvement region is 16a2 (28%), followed by region 12(24%) and 16b1(17%). However, the high metastasis rate of region 16b1 was similar among tumors at different sites, and the metastasis rate of region 16a2 was higher in eCCA than iCCA and GBC. As shown in figure1A. In study cohort, there were 92 PALNs, the specific distribution is shown in Additional file2. Over half (60%) of PALNs located in the LPA, 40% are located in RPA and 60% are located in APA.
A total of 93 positive lymph nodes were diagnosed in validation cohort. As shown in figure1B. Among them, the highest involvement region was 16b1(33%), followed by 16a2 (18%). And the involvement rates of regions 111 and 7 were 13% and 11%, respectively. The involvement rates of region 16b1 in GBC, iCCA and eCCA were 29%, 37% and 32%, respectively, while the involvement rates of region 16a2 were 7%, 26%, and 8% respectively. There were 56 PALNs recurrence in validation cohort, the distribution of PALNs is similar to that in the study cohort, about 70% of the lymph nodes are located in LPA, 30% in RPA and 59% in APA(see Additional file2).
Prognostic impact of PALN recurrence in BTC
The median OS of all relapsed patients was 18.9 months (IQR9.8-32.95). Patients with distant recurrence after surgery were divided into three subgroups: with PALN recurrence, with intrahepatic recurrence and with other distant organ recurrence. Survival analysis was carried out for any two groups. Patients with PALN recurrence had a better prognosis (Hazard Ratio(HR)=0.219, 95% confidence interval(CI) 0.067-0.711, P=0.006) when compared with patients with distant organ recurrence(Figure2A). And there was no significant difference in survival between the patients with PALN recurrence and patients with intrahepatic recurrence only(HR=1.183, 95% CI 0.539-2.594, P = 0.675)(see Additional file3). However, the OS in patients with intrahepatic recurrence only was significantly better than that in patients with distant organ recurrence(HR=0.367, 95% CI0.142-0.948, P = 0.031)( see Additional file3). In addition, we divided all postoperative patients with lymph node recurrence into two groups. In all patients with lymph node reecurrence, the prognosis in patients with PALN was not significantly worse than that of patients without PALN(HR=3.271; 95%CI 0.704-15.185; P=0.111)(Figure2B).
Delineation of para-aortic CTV
A total of 92 PALNs in study cohort were projected onto axial CT images of the standard patient. As shown in Additional file4. In the para aortic region, 16a2 and 16b1 have extremely high involvement rates in both cohorts. Therefore, we define the upper boundary of 16a2, that is, the upper edge of celiac trunk, as the superior border of para aortic CTV, and the lower boundary of 16b1, that is, the upper edge of inferior mesenteric artery, as the inferior border of para aortic CTV, which is equivalent to from the middle of T12 vascular body to the upper edge of L3 vascular body. In the horizontal direction, the mean distances of all lymph nodes in 16a2 and 16b1 from the volumetric centre of the nodes to the proximal edge of the aorta were 9 mm(range 4-24) for APA, 7mm(range 3–14) for LPA and 12mm (range 5–29) for RPA. The 90th percentile distances from the center of lymph nodes to the aorta were 14mm for APA, 11mm for LPA and 22mm for RPA. The 95th percentile distances were 18mm, 12mm and 24mm (Table3). The expansion margins of the aorta at 9 mm in the front, 7 mm on the left and 12 mm on the right got 67%(n = 51/76) coverage of the lymph node centre. The expansion margins of the aorta at 14mm in the front, 11 mm on the left and 22 mm on the right got 93% (n = 71/76) coverage. The expansion margins of the aorta at 18mm in the front, 12 mm on the left and 24 mm on the right got 96% (n = 73/76) coverage. Margins were cropped at bowel, muscle, and bone(Figure3).
Validation of para-aortic CTV coverage
In validation cohort, a total of 56 PALNs were projected onto the standardized axial CT images (Figure4A, 4B). Within the range from the upper edge of celiac trunk to the upper edge of inferior mesenteric artery, the average distance from the volume center of lymph nodes on the left side of the aorta to the edge of the aorta is 10mm(range 5–20). Similarly, the average distance on the right side also is 10mm(range 6–23). The average distance in front of the aorta was 9 mm(range5–23)(Table3). The CTV expansion margins calculated from the study cohort was applied to the validation cohort and 96% (n = 47 / 49) coverage of lymph nodes was obtained(Figure 4C).