Association of Immunologic Markers With Survival in Upfront Resectable Pancreatic Cancer

Abstract

This cohort study assesses the association of immunologic variables and survival in patients with resectable pancreatic ductal adenocarcinoma.

Pancreatic ductal adenocarcinoma (PDAC) is typically infiltrated by immune cells that influence the response to conventional therapies.^1,2 In addition, growing tumors cause systemic immune dysregulation, including expansion of myeloid populations.³ Finally, recovery from surgery results in a systemic wound healing response that can perturb peripheral immune populations.⁴ We hypothesized that peripheral expansion of myeloid or T cells would reflect similar increases in the tumor microenvironment and that increased intratumoral T-cell levels and robust postoperative immune recovery would correlate with improved outcome.

Methods

We conducted a prospective cohort study of resectable PDAC to determine which immunologic variables are associated with improved prognosis. We restricted our analysis to adult patients (18-100 years of age) who underwent upfront resection for PDAC from June 14, 2010, to November 24, 2014. We excluded patients who received neoadjuvant therapy, had other histologic subtypes, or had palliative surgery only. Peripheral populations were quantified in blood drawn before incision on the day of surgery and on the first postoperative visit with quantitative flow cytometry. We evaluated levels of CD3⁺, CD8⁺, CD8⁺/CD25⁺, CD4⁺, CD4⁺/CD25⁺, monocytes, CD14⁺ monocytes, and total granulocytes. Tumor-infiltrating immune cells were quantified by immunohistochemistry and quantitative digital image analysis for CD3⁺, CD68⁺, and CD8⁺ cells, as well as percentage of cells staining positive for programmed cell death ligand 1. Data were analyzed from November 6, 2011, through August 16, 2017. The primary outcome was overall survival. Secondary outcomes were correlation between peripheral circulating blood populations and tumor immune infiltrates. Significance is indicated by P < .05. This study was approved by the institutional review board of Providence Portland Cancer Center, Portland, Oregon, and all patients provided written informed consent.

Results

We recruited 63 patients (36 men [57%] and 27 women [43%]; median [interquartile range] age, 67 years [49-87 years]). Demographic characteristics and survival of these patients are outlined in Table 1. Clinicopathologic variables were evaluated for association with survival outcomes (Table 2). Median (interquartile range) overall survival was 19.7 months (12.8-35.8 months) from the time of surgery. Preoperative analysis of peripheral blood populations was complete for 50 patients. Analysis by Cox proportional hazards regression did not reveal any association with survival (Table 2), although higher preoperative monocyte levels trended toward decreased overall survival. Regression analysis did not reveal any association between peripheral blood populations and the presence of intratumoral CD3⁺, CD8⁺, or CD68⁺. Immunohistochemical analysis of tumor infiltrate was completed for 53 patients and revealed a positive association between CD3⁺ infiltrate and overall survival (hazard ratio, 0.27; 95% CI, 0.07-0.96; P = .04) but not with survival for CD68⁺ or CD8⁺ cells.

Table 1. Demographic and Immunohistochemical Features.

Characteristic (N = 63)	Data
Age, median (range), y	67 (49-87)
Sex, No. (%)
Male	36 (57)
Female	27 (43)
Tumor size, median (range), cm	3.0 (0.3-9.5)
Grade, No. (%)
Well differentiated	3 (5)
Moderately differentiated	24 (38)
Poorly differentiated	34 (54)
Perineural invasion present	39 (62)
Lymphovascular invasion present	28 (44)
Margin, No. (%)
R0	43 (68)
R1 (≤1 mm)	20 (32)
Nodal metastases, median (range), No.	2 (0-11)
T stage, No. (%)
1	12 (19)
2	38 (60)
3	13 (21)
N stage, No. (%)
0	15 (24)
1	29 (46)
2	19 (30)
TNM stage, No. (%)a
1a	6 (10)
1b	6 (10)
2a	3 (5)
2b	29 (46)
3	19 (30)
Adjuvant chemotherapy administered, No. (%)	44 (70)
Adjuvant radiation therapy administered, No. (%)	16 (25)
Vital status, No. (%)
Deceased	51 (81)
Alive	12 (19)
Overall survival, median (IQR), mo	19.7 (12.8-35.8)
Kaplan-Meier survival estimates, % (95% CI)
2 y	41.3 (29.1-53.0)
5 y	15.6 (7.1- 27.2)
Intratumoral stainingb
CD8+, median (range), cells/mm2	0.23 (0.05-0.69)
CD3+, median (range), cells/mm2	0.47 (0.14-1.30)
CD68+, median (range), cells/mm2	0.43 (0.14-1.33)
PD-L1 positive, median (range), %	0.1 (0.01-1.47)

Abbreviations: IQR, interquartile range; PD-L1, programmed cell death ligand 1.

^aIndicates the American Joint Commission on Cancer Cancer Staging Manual, 8th edition.

^bData were available for 53 patients.

Table 2. Cox Proportional Hazards Regression Analysis and Kaplan-Meier Survival Estimates for Populations With High and Low CD3⁺ and Peripheral Monocyte Levels.

Univariable Modeling Variable	HR (95% CI)	P Value
Clinicopathologic Variables
Sex (reference, male)	0.89 (0.51-1.55)	.68
Tumor size, cm	1.07 (0.94-1.22)	.32
Grade	1.23 (0.76-2.00)	.39
Perineural invasion	1.62 (0.91-2.90)	.10
Lymphovascular invasion	2.26 (1.27-4.02)	.005a
Positive margin	1.40 (0.79-2.48)	.24
No. of positive lymph nodes, per lymph node	1.14 (1.03-1.27)	.009a
T stage	1.28 (0.88-1.87)	.20
N stage	1.65 (1.13-2.40)	.009a
TNM stageb	1.48 (1.15-1.91)	.002a
Adjuvant chemotherapy	0.50 (0.28-0.89)	.02a
Adjuvant radiotherapy	0.58 (0.30-1.14)	.12
Intratumoral staining
CD8+, cells/mm2	0.31 (0.02-4.71)	.40
CD3+, cells/mm2	0.27 (0.08-0.96)	.04a
CD68+, cells/mm2	0.58 (0.15-2.21)	.43
PD-L1 staining, % positive cells	0.74 (0.36-4.21)	.74
Peripheral Circulating Cells, Log Cells/mL
Preoperative peripheral blood analysisc
CD3+	0.92 (0.49-1.71)	.79
CD8+	1.11 (0.69-1.78)	.66
CD8+/CD25+	1.10 (0.86-1.41)	.44
CD4+	0.92 (0.51-1.63)	.76
CD4+/CD25+	1.24 (0.85-1.81)	.26
Monocytes	1.68 (0.82-3.44)	.15
CD14+ monocytes	1.74 (0.84-3.59)	.13
Granulocytes	1.08 (0.64-1.84)	.76
Postoperative peripheral blood analysisd
CD3+	0.52 (0.22-1.26)	.15
CD8+	0.91 (0.54-1.52)	.71
CD8+/CD25+	0.89 (0.67-1.18)	.41
CD4+	0.45 (0.19-1.03)	.06
CD4+/CD25+	0.63 (0.36-1.10)	.10
Monocytes	0.34 (0.18-0.65)	.001a
CD14+ monocytes	0.36 (0.20-0.64)	.001a
Granulocytes	0.69 (0.46-1.04)	.08
Multivariable Modelinge (n = 33)
Postoperative monocytes, log cells/mL	0.44 (0.21-0.94)	.03
CD3+, cells/mm2	0.18 (0.03-0.94)	.04
TNM stageb	1.52 (1.03-2.22)	.03
Lymphovascular invasion	2.63 (1.08-6.37)	.03

Abbreviations: HR, hazard ratio; PD-L1, programmed cell death ligand 1.

^aIndicates significant difference between populations.

^bIndicates the American Joint Commission on Cancer Cancer Staging Manual, 8th edition.

^cIncludes 51 patients.

^dIncludes 41 patients.

^eIncludes 33 patients.

Postoperative peripheral blood analysis was complete for a total of 41 patients. Of these, only increasing monocyte count was associated with improved outcome (Table 2), with improved survival found with increasing CD4⁺ cell levels. Patients were also divided into groups with high (n = 26) and low (n = 27) levels of CD3⁺ using median expression as a cutoff. Median survival was 20.5 and 18.8 months, respectively, and the 5-year survival estimate was 18.0 months (95% CI, 6.0-35.1 months) and 5.56 months (95% CI, 0.5-2.6 months), respectively. A similar analysis was performed for those with high (n = 22) and low (n = 22) levels of postoperative circulating monocytes. Median survival was 25.3 and 21.6 months, respectively; 5-year estimate of survival, 21.3 months (95% CI, 4.7-45.9 months) and 9.5 months (95% CI, 0.9-31.1 months), respectively.

Multivariate Cox proportional hazards regression was performed using clinicopathologic and immunohistochemical information. Initially, all variables significant on the univariate analysis were included, and subsequently nonsignificant variables were removed in a stepwise fashion until the final model was generated. American Joint Committee on Cancer stage, CD3⁺ tumor infiltrate, lymphovascular invasion, and postoperative circulating monocytes remained significant (Table 2).

Discussion

Only a subgroup of patients with pancreatic cancer present with malignant neoplasms that are eligible for surgical resection, and these are skewed toward early-stage disease. Within our study population and within the immune cell types studied, preoperative analysis of peripheral lymphocytes was not associated with survival and was not reflective of the intratumoral immune infiltrate. However, higher levels of tumor-infiltrating CD3⁺ T cells and higher postoperative circulating levels of monocytes were associated with an improved prognosis. These data suggest that, although peripheral blood immunocytes do not reflect the immune environment of the tumor, consideration of the systemic immune response during recovery from resection may be of value. Our data suggest a possible role for the patient’s immune response on the outcome of pancreatic tumor resection and that interventions that change the tumor immune environment and the systemic response in the postresection period could influence patient outcomes.