C-reactive protein is a prognostic biomarker in pancreatic ductal adenocarcinoma patients

Abstract

Aim: The 5-year survival rate of pancreatic ductal adenocarcinoma (PDAC) is approximately 11% and has only improved marginally over the last three decades. For operable PDAC, resection and adjuvant FOLFIRINOX chemotherapy is standard of care. There is growing interest in perioperative regimens to improve outcomes. The non-randomized Phase II study "Gemcitabine and Abraxane for resectable Pancreatic cancer" (GAP) demonstrated the feasibility of perioperative gemcitabine/abraxane. Long-term survival in PDAC requires an effective immune response; hence, we undertook this translational study of the GAP trial cohort to identify immune-oncology biomarkers for clinical use.

Methods: We combined Nanostring nCounter technology with immunohistochemistry to investigate the correlation between gene expression and overall patient survival. Findings were investigated in samples from the International Cancer Genome Consortium (ICGC, n = 88) and the Australian Pancreatic Genome Initiative (APGI, n = 227).

Results: We confirmed that human equilibrative nucleoside transporter 1 (hENT1) expression was not a prognostic marker in PDAC but patients with high levels of hENT1 were more likely to live longer than 24 months post-surgery. Additionally, CD274 (PD-L1) and two novel biomarkers of survival, cathepsin W (CTSW) and C-reactive protein (CRP), were identified in the GAP cohort (n = 19). CRP expression was confirmed in data from the ICGC. Although PD-L1 and CTSW proteins were not significant across all three cohorts, results show that low CRP mRNA and protein expression are associated with longer overall survival in all three patient groups.

Conclusion: PDAC patients with long survival have higher hENT1 expression levels. Furthermore, CRP expression is a biomarker of poor prognosis following perioperative chemotherapy and resection in PDAC patients and thus may be useful for identifying patients who could benefit from more aggressive adjuvant strategies.

Keywords: C‐reactive protein; biomarker; pancreatic adenocarcinoma.

FIGURE 1

Study design. Biomarker discovery was performed in an archival tissue collection from patients enrolled in the Gemcitabine and Abraxane for resectable Pancreatic cancer (GAP) trial. RNA and protein expression were confirmed in two validation cohorts from the International Cancer Genome Consortium (ICGC) and Australian Pancreatic Genome Initiative (APGI). Biomarkers were correlated with patient survival outcomes to determine clinical utility.

FIGURE 2

Human equilibrative nucleoside transporter 1 (hENT1), marker of response to Gemcitabine. (A) hENT1 staining in tumor from a patient with short overall survival (OS) compared to tumor from a patient with long OS. Original magnification 20×. (B) Dot plot comparing the spread of the immunohistochemistry (IHC) H‐scores calculated from staining in patients with short OS (<24 months) to long OS (>24 months) (p = 0.0316). The dotted line indicates median hENT1 expression. (C) Kaplan–Meier (log‐rank) survival analysis, curves split by median expression of hENT1 for OS (p = 0.1621) (D) Kaplan–Meier survival analysis, curves split by median expression of hENT1 for progression‐free survival (PFS) (p = 0.1086).

FIGURE 3

Differentially expressed mRNAs. The Nanostring panel determined differentially expressed genes between patients with short (<24 months) and long (>24 months) overall survival (OS). Two‐tailed unpaired t‐test analysis of (A) CD274 (p = 0.0168). (B) Cathepsin W (CTSW) (p = 0.0053) (C) C‐reactive protein (CRP) (p = 0.0645). Kaplan–Meier (log‐rank) analysis of OS with groups split around the median expression of each marker. (D) CD274 (median 49.5, *p = 0.0200). (E) CTSW (median 41.13, p = 0.0572). (F) CRP (median 220.26, *p = 0.0314). Survival analysis in an independent International Cancer Genome Consortium (ICGC) RNA sequencing dataset. Kaplan–Meier curves were stratified according to gene expression. (G) PD‐L1 (**p = 0.0057). (H) CTSW (p = 0.1628, not significant). (I) CRP (**p = 0.0099).

FIGURE 4

Immunohistochemistry (IHC) analysis of PD‐L1, cathepsin W (CTSW), and C‐reactive protein (CRP) in tumor microarray (TMA) samples from the Australian Pancreatic Genome Initiative (APGI). (A) Staining is shown in one patient with short overall survival (OS) (2.5 months) compared to one patient with long OS (115.5 months). Original magnification 10× and 20× as indicated, nuclei are counterstained with hematoxylin. (B) Kaplan–Meier (log‐rank) analysis of OS with groups split around the median expression of each marker. PD‐L1 (median 5.71, p = 0.5159), CTSW (median 3.98, p = 0.5736), CRP (median 35.72, *p = 0.0439).