Molecular analysis of precursor lesions in familial pancreatic cancer

Abstract

Background: With less than a 5% survival rate pancreatic adenocarcinoma (PDAC) is almost uniformly lethal. In order to make a significant impact on survival of patients with this malignancy, it is necessary to diagnose the disease early, when curative surgery is still possible. Detailed knowledge of the natural history of the disease and molecular events leading to its progression is therefore critical.

Methods and findings: We have analysed the precursor lesions, PanINs, from prophylactic pancreatectomy specimens of patients from four different kindreds with high risk of familial pancreatic cancer who were treated for histologically proven PanIN-2/3. Thus, the material was procured before pancreatic cancer has developed, rather than from PanINs in a tissue field that already contains cancer. Genome-wide transcriptional profiling using such unique specimens was performed. Bulk frozen sections displaying the most extensive but not microdissected PanIN-2/3 lesions were used in order to obtain the holistic view of both the precursor lesions and their microenvironment. A panel of 76 commonly dysregulated genes that underlie neoplastic progression from normal pancreas to PanINs and PDAC were identified. In addition to shared genes some differences between the PanINs of individual families as well as between the PanINs and PDACs were also seen. This was particularly pronounced in the stromal and immune responses.

Conclusions: Our comprehensive analysis of precursor lesions without the invasive component provides the definitive molecular proof that PanIN lesions beget cancer from a molecular standpoint. We demonstrate the need for accumulation of transcriptomic changes during the progression of PanIN to PDAC, both in the epithelium and in the surrounding stroma. An identified 76-gene signature of PDAC progression presents a rich candidate pool for the development of early diagnostic and/or surveillance markers as well as potential novel preventive/therapeutic targets for both familial and sporadic pancreatic adenocarcinoma.

Figure 1. The pedigrees of pancreatic cancer families analysed.

(A–C) non-X families and (X) Family X.

Figure 2. Histology of PanIN lesions.

The top panel shows the histology of three members of non-X-families (A1, B1 and C2). Images at the top show PanIN-1 and -2 lesions (magnification ×100); and images at the bottom show PanIN-3 lesions from family B and C; magnification ×200). The lower panel shows the histology of three different members of Family X: X1, X5 and X6 at the top show their gross appearance (magnification ×20); images at the bottom show PanIN-1 from X1 sample (magnification ×100); and PanIN-3 lesions from X5 and X6 (magnification ×200). * indicates adjacent histologically normal appearing tissue.

Figure 3. Hierarchical clustering of all the PanIN and PDAC lesions.

(A) Venn diagram displays the numbers of common and unique probes in PDAC progression; (B) Heatmap of 93 commonly dysregulated probes (76 genes) is shown on the right. Each column represents a type of comparison and each row represents a gene probe. The level of up- and down-regulation is represented by the intensity of the red and green colour, respectively.

Figure 4. Functional modules and canonical pathways.

(A) The significant functional modules commonly altered during the transition from normal pancreas to PanIN and PDAC are shown. (B) Differences in most significantly affected canonical pathways between PanINs and PDAC samples are presented. The horizontal lines parallel to the x-axis in both images indicate a P = 0.05 threshold.

Figure 5. Localisation and expression of BCL6 and HMGB1.

(A) Representative images of BCL6 positive cells (brown staining) in the stroma in the vicinity of PanIN lesions are shown in the top two panels (both magnified ×100); two bottom images show inflammatory infiltrate with BCL6 immunoreactive cells in two PDAC cases (magnification ×100 and ×50, respectively). (B) HMGB1 nuclear expression (brown staining) was seen in all pancreatic compartments, including stromal immune infiltrate: top panels show PanIN-1 (left) and -2 (right) (magnification ×50, insert and second panel x100); bottom panels show PanIN-3 (left) and PDAC (right) (magnification ×100 and ×50, respectively).

Figure 6. Confirmation of gene expression profiling.

QRT-PCR analysis validated the differential expression for AGR2, S100P, FOS and EGR1 in primary PanIN lesions: PanA1, B1-3 and C1,2 represent non-X families, while PanX1 and 3 belong to Family X samples; PDAC1-7 represent seven different PDAC specimens.

Figure 7. Expression of TFF1 in familial PanIN lesions.

Panel (A) shows PanIN-1 with no TFF1 immunoreactivity, (B) and (C) PanIN-2 and (D) PanIN-3 lesion in the centre with strong TFF1 expression (all magnified x100).