The presence of tumour-associated lymphocytes confers a good prognosis in pancreatic ductal adenocarcinoma: an immunohistochemical study of tissue microarrays

Patients who underwent surgery with curative intent for pancreatic ductal adenocarcinoma between 8/3/96 and 21/6/11 were included in this study. All operations were performed in the Department of Hepatopancreaticobiliary Surgery, Nottingham University Hospitals NHS Trust. In this centre, 30-50 patients are operated on per year for periampullary cancer. In this study, we included only pancreatic ductal adenocarcinomas. Therefore, tumours of the ampulla, neuroendocrine tumours, cholangiocarcinomas and any other histological tumour types were excluded. Types of operations performed were Whipple’s procedure, total pancreatectomy, and distal pancreatectomy for tumours in the body, head, neck and tail of pancreas. Median follow up time was 42 months (6 to 163 months). Ethical approval was obtained from the National Regional Ethics Service Committee East Midlands - Nottingham 1 for the use of anonymised archival specimens and the requirement for patient/relative consent was waived by the Ethics Committee. The study was conducted according to REMARK criteria [19].

Tissue microarrays (TMA) were prepared using triplicate 0.6 mm tissue cores of tumour, identified by a specialist pathologist (AMZ), placed into a single recipient paraffin block, using a semi-automated instrument and targeted cores. Sections of TMA 4 μm thick were mounted on poly-L-lysine coated slides.

Investigation of CD3, CD8 and CD20 was conducted using a tissue microarray. Immunohistochemical staining was performed using Bond Max automated staining machines (Leica Microsystems, Wetzlar, Germany). In our laboratory the antigen retrieval step is performed on our automated staining machines. The temperature at which the Leica Bond Max automated stainers retrieve is 100°C for the appropriate time (i.e.20 minutes) followed by 12 minutes at ambient temperature. CD3 (NCL-L-CD3-565, Leica Microsystems) was stained optimally at a dilution of 1/100 with antigen retrieval performed using Epitope Retrieval solution 2 (ER2) for 20 minutes. CD8 (NCL-L-CD8-295, Leica Microsystems) was stained optimally at a dilution of 1/50 with antigen retrieval performed using Epitope Retrieval solution 1 (ER1) for 30 minutes. CD20 (M0755, Dako) was stained optimally at a dilution of 1/400 with antigen retrieval performed using ER1 for 20 minutes. ER1 (AR9961, Leica Microsystems) is a ready-to-use citrate based pH 6.0 solution; ER2 (AR9640, Leica Microsystems) is a ready-to-use EDTA based pH 9.0 solution. Sections were cut, dried at room temperature for 20 minutes, then incubated at 60°C for 20 minutes prior to loading onto the Bond stainers (Leica Biosystems, Milton Keynes, UK). The Bond staining protocol comprises several steps. The first step is dewaxing using Leica Dewax solution (AR9222) for 30 seconds at 72°C, followed by the antigen retrieval step and application of the endogenous peroxidase block for 5 minutes at room temperature (RT). There follows primary antibody incubation for 15 minutes at RT, post primary incubation for 8 minutes at RT, polymer incubation for 8 minutes at RT, DAB for 10 minutes at RT, DAB enhancer (AR9432) for 5 minutes at RT and haematoxylin counterstain for 5 minutes at RT. Wash steps were performed using Leica Bond Wash solution (AR9590) between each step. Peroxidase blocks, post primary, polymer, DAB and Haematoxylin were all supplied in Leica Bond Refine Detection kit (DS9800). Sections were then removed from the Bond staining machine, dehydrated in IMS (Genta Medical, Rudgate, UK), cleared in Xylene (Genta Medical) and permanently mounted under glass coverslips using Pertex Histolab (Algol Diagnostics, Espoo, Finland). The sections are washed in Leica Bond wash buffer to rehydrate after the dewaxing step.

Evaluation of the number of CD3 and CD8 positive lymphocytes was performed as follows: the absolute number of each immunohistochemically detectable T-cell subgroup was counted manually by two independent researchers, one a pathologist. The localisation of each cell was taken into account – T lymphocytes lying among epithelial carcinoma cells were regarded as 'intratumoural’ while those within the tumour stroma were regarded as 'stromal’. The proportion of tumour and stroma was estimated visually by both researchers independently. From these figures, the densities of lymphocytes per mm² were calculated using the TMA core area as reference [20]. The mean area of TMA cores was 0.355 mm². All cases were scored without prior knowledge of pathological stage of tumour or patient outcome.

As CD20 staining was primarily seen in the stroma, only stromal B lymphocytes stained with CD20 were counted. The proportion of stroma was estimated by both researchers independently. These figures were used to calculate the density of CD20 positive B lymphocytes in the stroma using the TMA core area as reference. Therefore, CD20 results are not presented as 'intratumoural’ or 'stromal’ but simply as 'CD20’ which refers to CD20 positive TALs seen in stromal tissue.

For inter-observer concordance, a subset of stained cores was scored by both observers and Kappa statistics were calculated to assess inter-observer variability. A subset of complete sections from patients in whom TMA sections had been evaluated was also scored by both observers. The mean number of TALs for each group resulting from the three tumour-TMA cores was attributed to the corresponding patient. The relationship between categorised protein expression and clinicopathological variables was assessed using Pearson Chi Square (χ ²) test of association. Survival curves were plotted according to the Kaplan-Meier method and significance determined using the log-rank test. Multivariate survival analysis was performed by Cox Proportional Hazards regression model. All differences were deemed statistically significant at the level of p < 0.05. Statistical analysis was performed using SPSS 19.0 software (IBM Corporation, NY, USA).

X-tile software (Yale School of Medicine, CT, USA) was used to stratify patients into 'high’ and 'low’ for each of the lymphocytes stained. The use of X-tile has been described previously [21]. X-tile plots provide a single, global assessment of every possible way of dividing a population into low, medium, and high-level marker expression. X-tile data are presented in a triangular grid where each point represents a different cut-point. The intensity of the colour of each cut-off point represents the strength of the association. The X-tile software allows the user to move a cursor across the grid and provides a histogram of the resulting population subsets along with an associated Kaplan-Meier curve. This histogram can be used to determine the optimal cut-off point which shows up as the brightest pixel on the X-tile plot. The use of X-tile software allows results to be produced on a 'test’ cohort and tested on a 'validation’ cohort. Therefore, all survival results have been tested in a simulated independent cohort and found to be valid.

Stratification cut-points were determined using X-Tile software for survival analysis (Table 1) and receiver operating curves (ROC) for clinicopathological features and were determined prior to statistical analyses [21].

CD3 staining was seen in both the tumour tissue and stroma of pancreatic adenocarcinomas (Figure 1a-d) and each was scored separately. Similarly, CD8 staining was seen in tumour and stroma (Figure 2a-d) and each was scored separately. CD20 staining was seen primarily in the stroma (Figure 3a). Staining was also seen in lymphoid follicles (Figure 3b). In poorly differentiated pancreatic adenocarcinomas, there was little positive CD20 staining in either the tumour or stroma (Figure 3c).

There was no association between sex of patients and survival (p = 0.113) (Figure 4a). There was also no significant association between tumour grade and survival (p = 0.103) (Figure 4b). Kaplan Meier plots of survival in pancreatic ductal adenocarcinoma in the presence of TALs alone, or in combination, are shown in Figure 5 (a-d) and Figure 6 (a-d). The presence of intratumoural CD3 correlated with improved survival (p = 0.028). When intratumoural CD3 and stromal CD3 were present, survival was improved (p = 0.043). The presence of CD20 positive lymphocytes correlated with improved survival (p = 0.029). When intratumoural CD3 and CD20 were present and stromal CD8 and CD20 were present, survival appeared improved (p = 0.001 and p = 0.013 respectively).

Cut-offs values for each of the clinicopathological features were determined using ROC curves. A number of prognostic factors including presence of venous invasion, perineural invasion, tumour size, grade of differentiation and lymph node status (positive or negative for tumour) were tested to determine whether CD3, CD8 and CD20 positive lymphocytes in the tumour or stroma were related to them.

In pancreatic ductal adenocarcinoma, the presence of intratumoural CD3 was significantly associated with grade of tumour differentiation (p = 0.049). Stromal CD8 also correlated with grade of tumour differentiation (p = 0.015). The presence of stromal CD3 and CD8 in combination, and stromal and intratumoural CD3 in combination also correlated with grade of tumour differentiation (p = 0.049 and p = 0.010 respectively). The only other positive correlation was between the presence of intratumoural CD8 and CD20 in combination with perineural invasion (p = 0.048). Table 3 shows all the correlations tested with their results.