| Characteristic | Overall, N = 771 | Non-PNI, N = 521 | PNI, N = 251 | p-value2 |
|---|---|---|---|---|
| Age at surgery | 66 (57, 73) | 66 (57, 73) | 66 (57, 71) | 0.8 |
| Sex | 0.3 | |||
| F | 34 (44%) | 25 (48%) | 9 (36%) | |
| M | 43 (56%) | 27 (52%) | 16 (64%) | |
| Tumor type | 0.007 | |||
| 0 | 4 (5.2%) | 3 (5.8%) | 1 (4.0%) | |
| 1 | 56 (73%) | 40 (77%) | 16 (64%) | |
| 2 | 11 (14%) | 3 (5.8%) | 8 (32%) | |
| 3 | 6 (7.8%) | 6 (12%) | 0 (0%) | |
| Tumor grade | 0.9 | |||
| 1 | 17 (23%) | 12 (24%) | 5 (22%) | |
| 2 | 43 (59%) | 30 (60%) | 13 (57%) | |
| 3 | 13 (18%) | 8 (16%) | 5 (22%) | |
| Unknown | 4 | 2 | 2 | |
| Lymphovascular invasion | 17 (23%) | 10 (19%) | 7 (32%) | 0.2 |
| Unknown | 3 | 0 | 3 | |
| Metastatic disease | 14 (19%) | 9 (18%) | 5 (20%) | >0.9 |
| Unknown | 3 | 3 | 0 | |
| Liver disease | 18 (23%) | 10 (19%) | 8 (32%) | 0.2 |
| Tumor size (cm) | 4.30 (3.10, 6.50) | 4.90 (3.43, 6.68) | 3.80 (2.45, 5.10) | 0.032 |
| Unknown | 12 | 6 | 6 | |
| Tumor location | 0.2 | |||
| Central/Hilum | 16 (21%) | 8 (15%) | 8 (35%) | |
| Left | 25 (33%) | 18 (35%) | 7 (30%) | |
| Right | 34 (45%) | 26 (50%) | 8 (35%) | |
| Unknown | 2 | 0 | 2 | |
| Small duct | <0.001 | |||
| 0 | 14 (24%) | 3 (8.1%) | 11 (52%) | |
| 1 | 43 (74%) | 34 (92%) | 9 (43%) | |
| 2 | 1 (1.7%) | 0 (0%) | 1 (4.8%) | |
| Unknown | 19 | 15 | 4 | |
| Tumor enhancement > 2/3 - Steve | 11 (23%) | 6 (19%) | 5 (29%) | 0.5 |
| Unknown | 29 | 21 | 8 | |
| Tumor enhancement > 2/3 - Meg | 7 (39%) | 4 (31%) | 3 (60%) | 0.3 |
| Unknown | 59 | 39 | 20 | |
| Tumor enhancement > 2/3 - Leslie | 2 (15%) | 2 (22%) | 0 (0%) | >0.9 |
| Unknown | 64 | 43 | 21 | |
| Soft tissue stranding - Steve | 21 (28%) | 9 (18%) | 12 (52%) | 0.002 |
| Unknown | 3 | 1 | 2 | |
| Soft tissue stranding - Meg | 24 (33%) | 17 (34%) | 7 (30%) | 0.8 |
| Unknown | 4 | 2 | 2 | |
| Soft tissue stranding - Leslie | 18 (26%) | 8 (17%) | 10 (43%) | 0.017 |
| Unknown | 7 | 5 | 2 | |
| Perivascular stranding - Steve | 18 (24%) | 10 (20%) | 8 (35%) | 0.2 |
| Unknown | 3 | 1 | 2 | |
| Perivascular stranding - Meg | 34 (47%) | 22 (44%) | 12 (52%) | 0.5 |
| Unknown | 4 | 2 | 2 | |
| Perivascular stranding - Leslie | 16 (23%) | 7 (15%) | 9 (39%) | 0.023 |
| Unknown | 7 | 5 | 2 | |
| Peritumoral stranding - Steve | 8 (11%) | 4 (7.8%) | 4 (17%) | 0.2 |
| Unknown | 3 | 1 | 2 | |
| Peritumoral stranding - Meg | 35 (48%) | 22 (44%) | 13 (57%) | 0.3 |
| Unknown | 4 | 2 | 2 | |
| Peritumoral stranding - Leslie | 27 (39%) | 15 (32%) | 12 (52%) | 0.10 |
| Unknown | 7 | 5 | 2 | |
| Deceased | 59 (77%) | 39 (75%) | 20 (80%) | 0.6 |
| 1 Median (IQR); n (%) | ||||
| 2 Wilcoxon rank sum test; Pearson’s Chi-squared test; Fisher’s exact test | ||||
Correlation of image findings with cholangiocarcinoma perineural invasion
Patient characteristics
Patient characteristics, including imaging results, are summarized by perineural invasion status in Table 1 below.
Imaging findings vs perineural invasion (PNI)
Because each category is rated by three readers, we assign a score 0, 1, 2, or 3 by the number of readers rating positive. We then assess the association between the scores with the status of perineural invasion (PNI). For each category, an odds ratio (OR) is calculated from a (univariate) logistic regression of PNI vs score. Table 2 below summarizes the results.
| Category | Score | No-PNI (N = 52) | PNI (N = 25) | OR | P |
|---|---|---|---|---|---|
| Tumor enhancement > 2/3 | 0 | 5 (71.4%) | 0 (0%) | 1.25 | 0.808 |
| 1 | 0 (0%) | 1 (100%) | |||
| 2 | 1 (14.3%) | 0 (0%) | |||
| 3 | 1 (14.3%) | 0 (0%) | |||
| Soft tissue stranding | 0 | 27 (57.4%) | 6 (26.1%) | 1.87 | 0.02 |
| 1 | 12 (25.5%) | 8 (34.8%) | |||
| 2 | 5 (10.6%) | 6 (26.1%) | |||
| 3 | 3 (6.4%) | 3 (13%) | |||
| Perivascular stranding | 0 | 24 (51.1%) | 7 (30.4%) | 1.56 | 0.076 |
| 1 | 13 (27.7%) | 7 (30.4%) | |||
| 2 | 6 (12.8%) | 5 (21.7%) | |||
| 3 | 4 (8.5%) | 4 (17.4%) | |||
| Peritumoral stranding | 0 | 25 (53.2%) | 7 (30.4%) | 1.6 | 0.067 |
| 1 | 9 (19.1%) | 6 (26.1%) | |||
| 2 | 11 (23.4%) | 7 (30.4%) | |||
| 3 | 2 (4.3%) | 3 (13%) |
The receiver operating characteristic (ROC) curve for PNI against each score is plotted in Figure 1 below, with the area under the curve (AUC) and its 95% confidence interval labelled on the graph.
Between-reader agreement
Kappa coefficients for imaging findings between readers are summarized in Table 3 below. The results show that the agreement between readers is moderate to substantial for all imaging findings.
| Steve v Meg | Meg v Leslie | Steve v Leslie | All three | |
|---|---|---|---|---|
| Tumor enhancement > 2/3 | 0.429, p=0.074 | 0.526, p=0.073 | 0.625, p=0.02 | 0.556, p=0.006 |
| Soft tissue stranding | 0.157, p=0.176 | 0.349, p=0.003 | 0.302, p=0.011 | 0.277, p<0.001 |
| Peritumoral stranding | 0.15, p=0.035 | 0.505, p<0.001 | 0.231, p=0.007 | 0.271, p<0.001 |
| Perivascular stranding | 0.205, p=0.049 | 0.381, p<0.001 | 0.405, p<0.001 | 0.315, p<0.001 |
Pathology vs PNI
To look at the association between tumor grade and PNI, we plot the prevalence of PNI for each grade level below, with OR (95% CI) calculated from a logistic regression model for PNI against tumor grade. There is a fairly strong positive association, but the association is not statistically significant.
Survival analysis
Proportion of patients who died by PNI status is plotted below.
The Kaplan–Meier curves for survival time from surgery to death are plotted below by PNI status and compared by the log-rank test.
