Patients
We investigated 115 US and 105 NACRT patients. Among the baseline clinicopathological characteristics of US and NACRT patients, there was a significant difference between the age, regional lymph node metastasis, and resection margin status. For adjuvant therapy, significantly greater patients received HAI after NACRT than US patients (75.2 % vs. 51.3 %, P = 0.0002; Table 1). Moreover, the overall survival (OS) of NACRT was better than US patients (37.1 M vs. 27.3 M, P = 0.018). OS was better after HAI than systemic therapy in both NACRT (57.8 M vs. 23.0 M, P = 0.0001) and US patients (41.2 M vs. 19.4 M, P = 0.0003). In a univariate and multivariate analysis of all patients, HAI administration (hazard ratio (HR) = 0.514, 95% confidence interval (CI) 0.297-0.892, P = 0.018) and distant metastasis were found to be independent prognostic factors (Additional file 1: Table S1).
Prognostic impact of CD200 in PDAC in association with pre-operative treatment
First, we evaluated CD200 expression in whole tissue specimens by IHC. The staining of CD200 was consistently intense in a small, round to spindle cell found in the peripheral of cancer lesions (Fig. 1a). Patients positive of this cell were defined as CD200+ PDAC. The CD200 positivity rate was significantly higher in NACRT (n = 67, 66.4%), compared with US patients (n =37, 32.2%) (P < 0.001; Fig. 1b). Furthermore, CD200 positive cells detected in half or more of the inspected fields were significantly greater in NACRT (n = 31, 46.3%) than US (n = 3, 8.1%) (P < 0.0001). In US patients, there was no difference in postoperative survival between CD200+ and CD200- cases (21.9 M vs. 30.6 M, P = 0.31; Fig. 1c). On the contrary, in NACRT patients, CD200+ cases had a significantly poorer prognosis than CD200- (33.4 M vs. 57.8 M, P = 0.03; Fig. 1d). Subsequently, we investigated the clinicopathological factors associated with poor prognosis in NACRT patients. Among the baseline patient demographics, RECIST response was the only significant difference associated with CD200 expression (Table 2). A univariate and multivariate Cox regression analysis of NACRT patients showed CD200 expression to be an independent poor prognostic factor (HR = 2.51, 95% CI 1.35-4.66, P = 0.004), as well as resection status, post-NACRT CA19-9, and distant metastasis (Table 3). However, in univariate and multivariate analysis of all patients, CD200 was not a prognostic factor.
To evaluate why CD200 lacked prognostic value in US patients, we assessed the differences in patient background. Since HAI was an independent prognostic factor in all patients, the significant difference in HAI administration rate between US and NACRT patients needed consideration. Therefore, we evaluated the prognostic impact of CD200 in a subgroup of patients only which received HAI therapy. In this subgroup, there was no significant difference between the OS of US (n = 59) and NACRT (n = 79) patients (41.2 M vs. 57.8 M, P = 0.19). Moreover, CD200 was nonetheless associated with poor prognosis in NACRT patients (37.1 M vs. 65.4 M, P = 0.0064), but not in US (28.5 M vs. 41.2 M, P = 0.29) (Fig. 1d,e).
Potential immunological impact of CD200
To investigate the mechanisms of poor prognosis in CD200+ patients after NACRT, we first looked into the association between CD200 and immunological factors.
We and others have reported TILs as an important immunological factor, and also a prognostic factor in PDAC [30-33]. In this study, we evaluated CD4, CD8, CD45RO positive TILs (Fig. 2a). The CD4+, CD8+ and CD45RO+ TIL count of CD200+ versus CD200- patients were 24.84 ± 11.31 vs. 31.81 ± 14.68 (P = 0.021), 32.69 ± 19.65 vs. 44.48 ± 23.30 (P = 0.019), and 52.71 ± 26.28 vs. 68.61 ± 29.88 (P= 0.016), respectively in NACRT patients (Fig. 2b). Prognosis did not differ in groups dichotomized by median or quartile TIL counts.
Correlation of CD200 with pancreatic cancer CSC marker
Since CD200 has been reported to co-express in CSCs [19], we next considered whether CSC properties were associated with the poor prognosis in CD200+ NACRT patients. However, to the author's knowledge, there have been no reports regarding the association of CD200 with PDAC CSC markers, including CD44/CD24/epithelial specific antigen (ESA) [34], or CD133 [35]. Thus, we first investigated the correlation of CD200 with recognized PDAC CSC markers. CD44 is reported to have an essential role in tumor progression [34], as well as radiotherapy resistance [36]. CD44s, a subtype of CD44, is expressed on tumor-initiating cells [37] and is reported to have a significant role in regulating stemness in pancreatic cancer cells [38]. Therefore, as an initial investigation, we evaluated the correlation of CD200 with CD44s expression by IHC in whole tissue samples of NACRT patients.
CD44s membranous staining was observed in some cells in the peripheral of cancer lesions, similar to those with CD200 staining (Fig. 3a). CD200 and CD44s positivity significantly correlated, with CD200+ found in 65.7 % of CD44s+ , and 23.7% in CD200- (P<0.0001; Fig. 3b). Since we found the significant correlation of CD200 and CD44s expression, we hypothesized that CD200 was associated with CD44+/CD24+/ESA+ pancreatic CSCs.
Influence of irradiation to CD200 and CSC marker expression in human pancreatic cancer cells
Next, since the CD200 expression rate increased after NACRT in the clinical setting, we investigated the role of irradiation on CD200 expression in an in vitro model. Also, we sought to verify our hypothesis of CD200 and CSC marker correlation by cell-surface marker analysis of CD200, CD44, CD24, and ESA with flow cytometry in human PDAC cells.
After a single-fraction irradiation of 10Gy, the CD200 positivity rate increased from 0.5% in control to 13.5% in irradiated PANC-1 cells, and from 1.7% in control to 17.5% in irradiated Capan-2 cells (Fig. 3c). Regarding CSC markers, post-irradiation CD200+ cells showed a stronger association with CSC markers after irradiation, compared with post-irradiation CD200- or control cells (Fig. 3d). The CD44+/CD24+ double-positive rate was higher after irradiation in CD200+ cells in both PANC-1 and Capan-2 cells. Although ESA expression was high (>96 %) regardless of irradiation in both PANC-1 and Capan-2 cells, CD200+ had a higher ESA+ rate compared to CD200-, in both control and irradiated cells (data of ESA expression from control cells not shown).
Impact of CD200 on the post-operative recurrence patterns
Finally, we investigated the association between CD200 and post-operate recurrence patterns. Since we have reported the preventive effect of HAI against hepatic recurrence [27], and also, since hepatic recurrence generally has a more unsatisfactory clinical outcome compared with other recurrence types, we especially paid attention to the hepatic recurrence patterns. The total hepatic recurrence rate did not differ between US (19.1%, n = 22) and NACRT (22.9%, n = 24) (P = 0.51). Among the patients who received post-operative treatment, the hepatic recurrence rate was lower after HAI (6.1 %, n = 7 ) than systemic therapy (11.3 %, n = 13) in US (P = 0.05), but was similar (15.2 %, n = 16 vs. 5.7 %, n = 6) in NACRT (P = 0.37). The rate of other failure patterns did not differ between HAI and systemic therapy in either NACRT or US patients (Additional file 2: Table S2).
First, we investigated the recurrence patterns, according to CD200 expression. In US recurrence patients, the hepatic recurrence proportion was similar in CD200+ (n = 7/26; 26.9%) and CD200- (n = 15/53; 28.3%) (P = 1.0). In contrast, in NACRT recurrence patients, the proportion of hepatic recurrence was significantly higher in CD200+ (n = 19/45; 42.2%) than in CD200- (n = 5/27; 18.5%) (P = 0.043). Interestingly, pulmonary recurrence proportion was significantly higher in NACRT CD200- (n = 14/27; 51.9%) than in CD200+ (n = 8/45; 17.8%) (P = 0.0036). There were no other distinct recurrence patterns associated with CD200 expression.
Next, we further investigated the association between CD200 and hepatic recurrence according to the adjuvant therapy regimen in NACRT. NACRT patients were stratified into 4 groups according to CD200 expression and treatment regimen (Fig. 4a). For this analysis, 6 patients who could not receive any adjuvant therapy were excluded. As a result, in NACRT CD200- patients, hepatic recurrence proportion was lower in HAI (n = 2/18; 11.1%) than systemic therapy (n = 3/8; 37.5%). In contrast, in NACRT CD200+ patients, hepatic recurrence proportions were similar in both HAI (n = 14/34; 41.2%) and systemic therapy (n = 3/8; 37.5%). Furthermore, the post-HAI hepatic recurrence onset was significantly earlier as well as higher in CD200+ (14.1 M, 27.5%) than CD200- (26.6 M, 7.1%) (P = 0.022, Fig. 4b).
Likewise, the association between CD200 and hepatic recurrence according to adjuvant therapy regimen was investigated in US patients. As a result, the proportion of hepatic recurrence was lower after HAI compared to systemic therapy, regardless of CD200 expression (Fig. 4c). Hepatic recurrence occurred in 1 out of 9 (11.1%) of HAI/CD200+, 6 out of 30 (20.0%) of HAI/CD200-, 5 out of 16 (31.3%) of systemic/CD200+, and 8 out of 21 (38.1%) of systemic/CD200- patients with recurrence.