Purpose: MicroRNA (miRNA) alterations are likely to contribute to the development of pancreatic cancer and may serve as markers for the early detection of pancreatic neoplasia.
Experimental Design: To identify the miRNA alterations that arise during the development of pancreatic cancer, we determined the levels of 735 miRNAs in 34 pancreatic intraepithelial neoplasias (PanIN) and 15 normal pancreatic duct samples isolated by laser capture microdissection using TaqMan miRNA microarrays. Differential expression of selected miRNAs was confirmed by FISH analysis and by quantitative real-time reverse transcription PCR (qRT-PCR) analysis of selected candidate miRNAs in an independent set of PanIN and normal duct samples.
Results: We identified 107 aberrantly expressed miRNAs in different PanIN grades compared with normal pancreatic duct samples and 35 aberrantly expressed miRNAs in PanIN-3 lesions compared with normal pancreatic duct samples. These differentially expressed miRNAs included those that have been previously identified as differentially expressed in pancreatic ductal adenocarcinomas (PDAC; including miR-21, miR-200a/b/c, miR-216a/b, miR-217, miR-146a, miR-155, miR-182, miR-196b, miR-203, miR-222, miR-338-3p, miR-486-3p, etc.) as well as miRNAs not previously described as differentially expressed in these lesions (miR-125b, miR-296-5p, miR-183*, miR-603, miR-625/*, miR-708, etc.). miR-196b was the most selectively differentially expressed miRNA in PanIN-3 lesions.
Conclusions: Many miRNAs undergo aberrant expression in PanIN lesions and are likely to be important in the development of PDAC. The miRNAs, such as miR-196b, whose expression is limited to PanIN-3 lesions or pancreatic cancers could be useful as diagnostic markers. Clin Cancer Res; 18(4); 981–92. ©2011 AACR.
Pancreatic cancer is the fourth leading cause of cancer death in the United States and is characterized by advanced disease at the time of diagnosis and resistance to most therapeutic treatments. Investigating the precursor neoplasms of pancreatic cancer can help elucidate the molecular mechanisms responsible for the development of pancreatic cancer and can also identify markers that could potentially be used to help identify high-grade precursor neoplasms among patients undergoing pancreatic screening. In this study, we conducted comprehensive quantitative analysis of more than 700 microRNAs (miRNA) in pancreatic intraepithelial neoplasias (PanIN) and normal pancreatic duct samples to identify differentially expressed miRNAs at each PanIN grade. We found numerous differentially expressed miRNAs. The most specifically overexpressed miRNA in PanIN-3 lesions (carcinoma in situ) was miRNA-196. MiRNA-196 has potential use as a marker of PanIN-3 lesions.
Pancreatic cancer is the fourth leading cause of cancer-related deaths in the United States (1). In the United States in 2010, 43,140 new patients were diagnosed, and approximately 36,800 Americans died from pancreatic cancer (1). The poor prognosis and high mortality rate result, at least in part, from the generally late presentation of the disease and the lack of effective therapies (2). Although early detection is considered the best way to cure pancreatic cancer, most early-stage pancreatic cancers do not cause symptoms. As a result, there is considerable interest in pancreatic screening for individuals considered to be at significantly increased risk of developing pancreatic cancer, such as those with an inherited predisposition. Because even early-stage invasive pancreatic cancer is usually incurable, the primary goal of pancreatic screening programs has been to prevent the development of invasive pancreatic cancer by detecting and treating pancreatic precursor lesions. The most common of these precursor lesions are pancreatic intraepithelial neoplasias (PanIN; ref. 3). Because PanINs are too small to be reliably detected by pancreatic imaging tests (4, 5), there is considerable interest in identifying markers of advanced PanINs (6, 7) to improve our ability to detect advanced PanINs during pancreatic screening. Although numerous studies have described the molecular alterations of pancreatic ductal adenocarcinomas (PDAC; refs. 8–11), fewer studies have investigated the timing of such alterations in PanIN lesions. Understanding the molecular alterations of PanINs may not only identify markers for pancreatic screening but also to identify important biologic pathways.
Changes in the expression of microRNAs (miRNA) are important to the development of cancer. miRNAs are small endogenous noncoding RNAs of 14 to 24 nucleotides that negatively regulate protein expression at the posttranscriptional level by inhibiting translation and/or by targeting mRNAs for degradation (12). Furthermore, because miRNAs are stable and detectable in human plasma, they are being investigated for their use as diagnostic serum markers (13).
Alterations in the expression of miRNAs are suspected to contribute to the development and progression of pancreatic and other cancers (14–16). PDACs overexpress several miRNAs including miR-21, miR-34, miR-146a, miR-155, miR-196a-2, and miR-200a/b (13, 17–24).
Although several studies have reported miRNA alterations of pancreatic cancer (4, 17, 20, 22, 24), the role of these alterations during early pancreatic neoplastic development is not well understood. Kent and colleagues revealed that the repression of the miR-143/145 cluster by oncogenic Ras promotes pancreatic cancer development (25). But to date, few miRNAs have been examined for alterations in mouse or human PanIN lesions (26, 27).
In this study, we used the TaqMan array human miRNA cards (Sanger miRbase v16) to comprehensively profile PanIN miRNA expression relative to normal pancreatic duct cells.
Materials and Methods
Fresh pancreatic tissues were snap frozen in liquid nitrogen, embedded in Tissue-Tek optimum cutting temperature (OCT) compound medium (Sakura FineTek) and stored at −80°C. The samples were subsequently sectioned onto UV-treated PALM membrane slides (Carl Zeiss MicroImaging, Inc.) for laser capture microdissection (LCM) and stored at −80°C (5). In each case, the PanIN lesions were examined histologically and the diagnosis was confirmed by two of the authors (R.H. Hruban and S.-M. Hong), who are expert pancreatic pathologists. PanINs were graded as PanIN-1 (low grade), PanIN-2 (intermediate grade), and PanIN-3 (high grade) as previously described (28). All specimens were collected and analyzed with the approval of the Johns Hopkins Committee for Clinical Investigation (29).
Tissue microarray construction
Tissue microarrays (TMA) were constructed from the archival formalin-fixed, paraffin-embedded tissue blocks of surgically resected primary PDAC using a manual Tissue Puncher/Arrayer (Beecher Instruments) as previously described (30). A total of ninety-four 1.4-mm cores (42 PDAC, 44 PanINs, and 8 normal pancreatic ducts) were arrayed on the recipient blocks. The PanIN lesions comprised 13 PanIN-1, 15 PanIN-2, and 16 PanIN-3 lesions.
Cells and culture conditions
The human pancreatic ductal epithelial cell line (HPDE; provided by Dr. Ming-sound Tsao at the University of Toronto, Toronto, ON, Canada) was cultured in serum-free keratinocyte media supplemented with supplied growth factors according to the manufacturer's instructions. Twenty-nine pancreatic cancer cell lines, including A38-41, A38-44, AsPC-1, BxPC-3, CAPAN-1, CAPAN-2, CFPAC-1, HPAFII, Hs766T, Mia PACA-2, Pa01C, Pa02C, Pa03C, Pa07C, Pa08C, Pa09C, Pa14C, Pa16C, Pa18C, Pa20C, Pa21C, Pa28C, PANC-1, Panc486, PK8, PK9, PL11, Su86.86, and SW1990 were used in this study. All cancer cell lines were maintained in Dulbecco's Modified Eagle's Medium (4.5 mg/mL glucose; Invitrogen) supplemented with 10% fetal bovine serum and antibiotics (100 μg/mL streptomycin and 100 units/mL penicillin).
As a guide, one frozen section slide was stained with hematoxylin and eosin. Thirty-four PanIN lesions (12 PanIN-1, 11 PanIN-2, and 11 PanIN-3 lesions) and 15 samples of normal pancreatic ductal epithelial cells adjacent to PanIN lesions from patients with PDAC (n = 6) or other diagnoses [intraductal papillary mucinous neoplasms (IPMN) or serous cystadenoma, n = 9] were selectively isolated with the PALM laser microdissection platform (PALM, Carl Zeiss MicroImaging, Inc.) according to the manufacturer's protocols (ref. 31; Supplementary Fig. S1A–S1D). A separate set of PanIN lesions (6 PanIN-1 lesions, 3 PanIN-2 lesions, and 2 PanIN-3 lesions) and 9 samples of normal pancreatic ductal epithelium [from patients with either PDAC (n = 1) or benign neoplasms (serous cystadenoma, IPMN, n = 8)] were laser capture microdissected in the same fashion to validate the differential expression of candidate miRNAs. In addition, to measure the expression of candidate miRNAs in pancreatic cancer, neoplastic cells from 14 primary PDACs were also laser capture microdissected. We microdissected a mean of 20,000 PanIN and normal ductal epithelial cells to help ensure the detection of abundance for the microarray analysis and a mean of 4,000 normal, PanINs and PDAC cells for the validation analysis.
Total RNA was extracted using mirVana miRNA isolation kit (Ambion 1560) for cultured cells and RNAqueous-Micro kit (Ambion 1931) for microdissected cells, following the manufacturer's protocols (total RNA isolation procedure and RNAqueous-Micro procedure for LCM, respectively). The extracted RNA was quantified by the absorbance at 260 nm, and the purity of the extracted RNA was evaluated by the absorbance ratio at 260 or 280 nm with a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies).
miRNA expression profiling
Comprehensive miRNA expression profiling was carried out with the TaqMan Array Human miRNA Cards (Cards A and B, v2.1 and v3.0, respectively; Applied Biosystems) using the 7900HT thermocycler (Applied Biosystems). These 2 cards are designed with 750 unique assays of human miRNAs from Sanger miRbase v14, of which we identified 735 human miRNAs from Sanger miRbase v16 (Supplementary Table S1). miRNAs were amplified after specific reverse transcription and preamplification using Megaplex Assay Performance (Megaplex RT Primer Pools and Megaplex PreAmp pools, both from Applied Biosystems) according to manufacturer's instructions (Applied Biosystems) and normalized against RNU6B (U6 snRNA, an endogenous control assay designed in both cards). Relative expression was determined using the ΔΔCt method and a ≥32 Ct value was interpreted as amplification too low to quantify.
Individual miRNA expression detection
miRNAs that were candidates for being differentially expressed were analyzed using the TaqMan Small RNA Assay (Applied Biosystems), a 2-step quantitative PCR (qPCR). The 7900HT Thermocycler was used to measure the abundance of individual candidate differentially expressed miRNAs. Candidate miRNAs were amplified after specific reverse transcription and preamplification using Megaplex Assay Performance for LCM samples or after specific reverse transcription using TaqMan MicroRNA Reverse Transcription Kit (Applied Biosystems) for cultured cell samples, according to manufacturer's instructions (Applied Biosystems) and normalized against RNU6B. All candidate miRNAs' and RNU6B's primers were provided from Applied Biosystems. Each sample was run in triplicate. Relative expression was determined using the ΔΔCt method and a ≥32 Ct value was interpreted as amplification too low to quantify.
Locked nucleic acid-FISH
Locked nucleic acid-FISH (LNA-FISH) was carried out on TMA slides using LNA oligonucleotide probes against miR-196b and U6, both labeled with fluorescein at the 5′-end (Exiqon), according to the protocol described in the work of de Planell-Saguer and colleagues (32). LNA-U6 was used as a positive measure of probe specificity. Briefly, after deparaffinization, slides were prehybridized for 30 minutes at 50°C (30°C below the RNA Tm of LNA-miR-196b probe which is 80°C) in a humid chamber with prehybridization buffer, then incubated with hybridization buffer containing LNA-miR-196b probe (1:1,000) for 1 hour at 50°C in a hybridization oven. After several washes, the slides were incubated in 2-nitro-5-thiobenzoate (TNB) blocking buffer (PerkinElmer) containing goat anti-fluorescein antibody peroxidase conjugate [fluorescein isothiocyanate (FITC)/horseradish peroxidase (HRP); Rockland; 1:1,000) for 30 minutes in a humid chamber at room temperature]. Then the signals were amplified using tyramide signal amplification (TSA; PerkinElmer) for 10 minutes in a humid chamber at room temperature. After incubation in 4′,6-diamidino-2-phenylindole (DAPI) staining solution for 5 minutes at room temperature, slides were mounted with Prolong Gold anti-fade reagent (Invitrogen) and incubated overnight at 4°C and signals were visualized using a fluorescent microscope. All steps beginning with hybridization were carried out in the dark. The LNA-miR-196b-FISH results were quantified at a single-cell level by counting expression spots per cell as previously described (33).
Principal component analysis (PCA) mapping of comprehensive miRNA expression profiling was carried out by using Partek Genomics Suite software (Partek Incorporated). Differences in median expression were determined using the Mann–Whitney U test, differences in proportions of expressing samples were determined with Pearson χ2 test, and differences in LNA-FISH abnormalities between categorical variables were determined with the Student t test. Statistical significance was defined as a value of P < 0.05. To adjust for multiple comparisons, we calculated the false discovery rate (FDR) from the subset of significantly deregulated miRNAs with P < 0.05 (34). All statistical analyses were conducted using the SPSS Statistics 19.0 software.
miRNA expression differs in PanIN lesions and normal pancreatic ducts
The miRNA profiles of 34 laser capture–microdissected PanIN lesions were compared with 15 LCM samples of normal pancreatic duct using TaqMan Array Human MicroRNA Cards containing 735 human miRNA assays. The miRNA concentrations were normalized against RNU6B. We first used PCA to compare the global miRNAs expression profile of PanIN lesions with that of samples of normal pancreatic ductal epithelial cells (Supplementary Fig. S1E) using the data from TaqMan Array Human MicroRNA Cards. This analysis revealed that miRNAs effectively separated PanIN lesions (all PanINs or each group of PanINs, PanIN-1, PanIN-2, and PanIN-3) and normal pancreatic ducts, whereas PanIN lesions of different grades overlapped in their miRNA profiles (Supplementary Fig. S1E).
Numerous miRNAs are aberrantly expressed in PanIN lesions
To identify differentially expressed miRNAs among PanINs overall relative to normal pancreatic duct samples, we calculated the median fold change of each miRNA between PanIN and normal duct samples analyzed by TaqMan microarray using the ΔΔCt method (Supplementary Table S2). This analysis identified 65 significantly differentially expressed miRNAs in PanIN lesions and normal pancreatic duct samples (fold change >2 or <0.5, respectively; P < 0.05, corrected for FDR; Table 1).
To independently validate the differential expression of miRNAs identified by microarray analysis, we chose 13 candidate overexpressed miRNAs (miR-146a, miR-182, miR-193a-3p, miR-193b, miR-200a, miR-200b, miR-200c, miR-21, miR-29b, miR-425, miR-486-3p, miR-708, and miR-874) and 1 candidate underexpressed miRNA (miR-296-5p) for qPCR analysis in an independent set of 20 samples, including 9 normal pancreatic duct samples and 11 PanIN lesions isolated by LCM. We confirmed the differential expression of 13 of these 14 miRNAs (all except miR-296-5p). These results are shown in Fig. 1 and Supplementary Fig. S2 and Table S3.
miRNA alterations in different grades of PanINs
Because PanINs undergo a series of histologic and molecular changes as they progress toward invasive adenocarcinoma, we next compared TaqMan Array miRNA profiles of intermediate-to-high-grade PanIN lesions (PanIN-2/-3) and those of low-grade PanIN lesions (PanIN-1) with normal pancreatic ducts. This comparison yielded 10 significantly overexpressed and 8 significantly underexpressed miRNAs in PanIN-2/-3 samples compared with PanIN-1/normal duct samples (fold change >2 or <0.5, respectively; Supplementary Table S4), of which 4 of the overexpressed miRNAs (miR-18b, miR-21, miR-338-3p, and miR-874) and 6 of the underexpressed miRNAs in the higher grade lesions (miR-139-3p, miR-214, miR-216b, miR-296-5p, miR-370, and miR-622) reached significance after adjusting for FDR. Several of these miRNAs were among the differentially expressed miRNAs between PanINs and normal pancreata (Table 1).
Because PanIN-3 lesions are an important lesion to identify in patients undergoing pancreatic screening, we next compared miRNA profiles of PanIN-3 with those of normal pancreatic duct samples. This comparison yielded 26 significantly overexpressed and 9 significantly underexpressed miRNAs in PanIN-3 lesions compared with normal pancreatic duct samples (fold change >2 or <0.5, respectively; corrected for FDR; Table 2; Supplementary Table S5). Only one of these miRNAs (miR-196b) was selectively overexpressed in PanIN-3 lesions relative to other PanINs (Table 1). In addition, 8 miRNAs (miR-125b, miR-126, miR-127-3p, miR139-3p/5p, miR-216b, miR-218, and miR-452) were selectively underexpressed in PanIN-3 lesions relative to other PanINs (Table 2).
Silencing and induction of miRNA expression in PanIN lesions
Analysis of differences in the median expression of miRNAs may fail to identify differences in miRNA expression that occur as a result of either aberrant silencing or aberrant induction of expression in a subset of PanIN lesions. We therefore classified the miRNA expression profiles in PanIN and normal duct samples into expressing and nonexpressing and compared them to identify changes in the proportion of expressing samples among the groups. For each miRNA, we classified expressing samples as those with a quantitative real-time reverse transcription PCR (qRT-PCR) Ct value <32. This classification yielded 2 significantly upregulated miRNAs (miR-516a-3p and miR-659) and 16 significantly downregulated miRNAs (miR-107, miR129-3p, miR-181c, miR-184, miR-198, miR-216a/b, miR-217, miR-370, miR-372, miR-379, miR-433, miR-539, miR-543, miR-618, and miR-758) in PanIN lesions compared with normal pancreatic duct samples, after adjusting for FDR (Supplementary Table S6). miR-196b was significantly more likely to be expressed and miR-216b was more likely to be silenced in PanIN-3 lesions compared with normal pancreatic ductal samples (Table 2; Supplementary Table S6). All of the miRNAs identified as differentially expressed were also identified as differentially expressed in PanINs versus normal pancreatic duct samples or between PanIN-2/3 lesions versus PanIN-1/normal duct samples (Supplementary Tables S4 and S6).
We next analyzed the expression of 3 of the candidate-overexpressed miRNAs (miR-182, miR-196b, and miR-486-3p) and 3 candidate-underexpressed miRNAs (miR125b, miR216b, and miR-296-5p) in additional PanIN-3 lesions by qPCR. Combined analysis of these samples with PanIN-3 samples and normal pancreatic duct samples including samples from the original set of PanIN-3 and normal pancreatic duct samples (24 normal pancreatic ducts and 13 PanIN-3 samples in total) confirmed the differential expression of all 6 miRNAs after adjusting for FDR (Fig. 2; Supplementary Fig. S3).
Candidate miRNA expression profiles in pancreatic cancers
We next examined pancreatic cancers for miRNA expression of miRNAs that have not been described as differentially expressed in invasive pancreatic cancers previously including 2 miRNAs that were overexpressed in PanINs (miR-182 and miR-196b) and one that was underexpressed (miR-296-5p). Expression of these miRNAs was examined in 29 pancreatic cancer cell lines and in the nonneoplastic pancreatic epithelial line, HPDE. As shown in Supplementary Fig. S4, only miR-296-5p showed differential expression in pancreatic cancer cell lines relative to HPDE. To confirm that miRNA expression patterns of pancreatic cancer cell lines were the same in primary pancreatic cancer cells, we examined the expression of these 3 candidate miRNAs in 14 samples of laser capture–microdissected PDAC cells compared with 8 samples of microdissected normal pancreatic ductal cells. As shown in Supplementary Fig. S4, the expression of these 3 miRNAs was significantly different in primary PDAC samples relative to normal pancreatic ducts for all 3 miRNAs, including miR-182 (P < 0.001), miR-196b (P < 0.001), and miR-296-5p (P = 0.002), respectively.
Expression of miRNA-196b in PanIN-3 lesions by LNA-FISH
Because miRNA-196b is the most differentially expressed miRNA in PanIN-3 lesions by miRNA microarray analysis, we further examined its expression in PanIN lesions using LNA-FISH. We found that the mean expression of miR-196b in 16 PanIN-3 lesions was 3.1 spots per cell [95% confidence interval (CI), 2.0–4.3] versus 0.5 spots per cell (95% CI, 0.2–0.7) in 8 normal pancreatic duct cells (P = 0.027). MiRNA-196b levels were also significantly higher in the PanIN-3 lesions than in 13 PanIN-1 (0.4 spots per cell; 95% CI, 0.2–0.7) and 15 PanIN-2 lesions (0.6 spots per cell;95% CI, 0.1–1.0; P = 0.010, and 0.011, respectively; Fig. 3; Supplementary Fig. S5). The miRNA levels were significantly higher in 42 PDACs (7.1 spots per cell; 95% CI, 6.0–8.3) than in PanIN-3 lesions, lower grade PanINs, and normal pancreatic duct cells (Fig. 3).
Numerous miRNAs are aberrantly expressed during pancreatic carcinogenesis
We further compared miRNAs' expression in each PanIN grade with those in normal pancreatic duct cells by the timing of aberrant PanIN expression (fold change >2 or <0.5; Fig. 4; Supplementary Table S7). This comparison yielded 107 miRNAs significantly aberrantly expressed during one or more PanIN stages, including 37 miRNAs not previously identified as aberrantly expressed in pancreatic cancer (13, 17, 19, 20, 24, 26, 35–40). Almost all of these miRNAs were identified by comparing the miRNA levels of PanINs as a group with normal pancreatic ducts. One exception is miR-494, which was significantly elevated in PanINs as an overall group, (Table 1), but not in any one PanIN subgroup. Among the miRNAs not previously reported to be aberrantly expressed includes overexpressed miRNAs, miR-135b*, miR-18b, miR-183*, miR-516a-3p, miR-603, miR-625, miR-629, miR-652, miR-659, miR-708, and miR-874, and underexpressed miRNAs, miR-10a*/b*, miR136*, miR-146b-3p, miR-223*, miR-433, miR-452, miR-539, miR-543, miR-551b, miR-618, miR-744*, miR-758, and miR-885-5p, and others. In addition, some of the miRNAs we identified as differentially expressed miRNAs have been previously found to show the same differential expression in pancreatic cancers. These include miRNAs overexpressed in PanINs: miR-129-3p, miR-130b, miR-133a, miR-135b, miR-148a/b, miR-151-5p, miR-183, miR-185, miR-193a-3p, miR-200c, miR-320b, miR-330-3p, miR-331-3p/5p, miR-34c-5p, miR-378, miR-422a, miR-423-5p, and miR-425, and miRNAs underexpressed in PanINs including miR-125b, miR-126, miR-127-3p, miR-137, miR-181c, miR-184, miR-198, miR-205, miR-32, miR-370, miR-372, and miR-379 (17, 20, 22, 24, 37, 40). Furthermore, several miRNAs were identified as significantly overexpressed only in low-grade PanINs (only in PanIN-1 or PanIN-1/PanIN-2 lesions) including miR-129-3p, miR-130b, miR-133a, miR-148a/b, miR-151-5p, miR-185, miR-200c, miR-330-3p, miR-331-3p/5p, miR-34c-5p, miR-378, and miR-423-5p.
In this study, we have comprehensively profiled the expression of miRNAs in PanINs and normal pancreatic duct samples. Many of the differentially expressed miRNAs we identified in PanIN lesions have been also identified as differentially expressed in invasive pancreatic cancers (Supplementary Table S7). These differentially expressed miRNAs include overexpression of let-7f/g, miR-101, miR-103, miR-106b, miR-146a, miR-15b, miR-155, miR-18a, miR-182, miR-190, miR-193b, miR-194, miR-196b, miR-200a/b, miR-203, miR-21, miR-222, miR-29a/b/c, miR-31, miR-338-3p, miR-429, miR-486-3p, miR-93 and miR95, and underexpression of miR-107, miR-139-3p/5p, miR-216a/b, miR-217, miR-218, and miR-483-5p (13, 17, 20, 22, 37). Several of these miRNAs have been found to have oncogenic or tumor-suppressive functions. These include miR-107, miR-146a, miR-155, miR-200a/b, miR-203, miR-21, miR-216a/b, and miR-217 (13, 22, 38, 41–43).
The overexpression of let-7a, miR-200, and miR-21 we observed in PanIN lesions confirm the results of the study by Du Rieu and colleagues who found overexpression of these miRNAs in mouse PanIN lesions and found that the miR-21 upregulation preceded phenotypic changes in pancreatic duct cells (26). In contrast, Laconti and colleagues found overexpression of miR-10, miR-16, miR-21, miR-100, and miR-155 in mouse PanINs, whereas we identified overexpression of only miR-21 and miR-155 in human PanINs (27).
We also identified numerous differentially expressed miRNAs that have not been identified as differentially expressed in pancreatic cancers (Supplementary Table S7). For example, miR-183, a member of miR-182/183/96 family (44), targets EGR1 and PTEN and promotes tumor cell migration (45). We also identified several differentially expressed miRNAs that had differential expression only in low-grade PanINs (PanIN-1 or PanIN-1/-2 lesions). Levels of one such candidate, miR-200c, have been reported to influence E-cadherin expression, inhibit invasion of pancreatic cancer cells, and correlate with patient survival (40). Such a pattern of expression is perhaps not surprising given the different phenotypic and molecular profiles of low-grade and high-grade PanINs. Several miRNAs described as elevated in pancreatic cancers including miR-143/145 and miR-34a did not display significantly different expression patterns to normal pancreatic duct samples (Supplementary Tables S2 and S5). None of the miRNAs identified as overexpressed are located at chromosomal loci that are known targets of chromosomal amplification.
The identification of markers of high-grade dysplasia and invasive PDACs such as miR-196b could have implications for pancreatic screening. We and others have been screening individuals with extensive family histories of pancreatic cancer (4, 5, 46–48). Pancreatic screening by endoscopic ultrasound typically identifies subtle nonspecific changes in the pancreas, and when such pancreata are resected, PanIN lesions are often found in association with lobulocentric atrophy (49). However, these imaging alterations are nonspecific and are also detected in patients without an increased risk of pancreatic cancer as well as among older individuals and patients with chronic pancreatitis (50). One important goal of pancreatic screening is to identify high-grade lesions so that patients with significant lesions can be offered pancreatic resection prior to the development of invasive cancer. Markers are needed that are specific for high-grade PanIN-3 lesions and invasive PDAC among patients undergoing pancreatic screening. Such markers could potentially be detected in pancreatic fluid samples obtained after secretin infusion at the time of upper endoscopy. Our results indicate that miR-196b has such specific expression patterns and that miR-196b deserves further investigation to determine whether it can identify PanIN-3 among patients undergoing screening.
In summary, we have conducted a quantitative real-time PCR analysis of more than 700 miRNAs in PanIN lesions to identify aberrantly expressed miRNAs at each PanIN stage. These aberrant miRNA expression patterns may provide insights into pancreatic neoplastic development. The almost exclusive expression of miR-196b in PanIN-3 lesions and pancreatic cancers compared with lower grade PanINs and normal pancreata suggest that its detection could be useful for identifying PanIN-3 lesions in patients undergoing pancreatic screening.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
J. Yu and M. Goggins conceived and designed the experiments. J. Yu, A. Li, and S.-M. Hong carried out the experiments. J. Yu, A. Li, and M. Goggins analyzed the data. J. Yu, R.H. Hruban, and M. Goggins wrote the manuscript. J. Yu, A. Li, S.-M. Hong, R.H. Hruban, and M. Goggins agree with the manuscript's results and conclusions. J. Yu and M. Goggins wrote the draft of the manuscript.
This work was supported by NIH grants (P50-CA62924, R01-CA120432, RC2CA148376) and the Michael Rolfe Foundation.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).
- Received September 17, 2011.
- Revision received November 14, 2011.
- Accepted November 15, 2011.
- ©2011 American Association for Cancer Research.