Jonathan J. Harrington

Fecal calprotectin levels predict colorectal inflammation among patients with chronic diarrhea referred for colonoscopy

Fecal calprotectin levels predict colorectal inflammation among patients with chronic diarrhea referred for colonoscopy

Aberrant methylation of secreted frizzled-related protein genes in esophageal adenocarcinoma and Barrett's esophagus.

Hypermethylation of secreted frizzled‐related proteins (SFRP) genes frequently occurs with several cancers but has not been studied in esophageal adenocarcinoma or its precursor—Barretts esophagus. To explore the role of SFRP methylation in the neoplastic progression of Barretts esophagus and to evaluate methylated SFRP genes as biomarkers for Barretts esophagus and cancer, methylation of SFRP genes was determined in esophageal adenocarcinomas, Barretts esophagus and normal epithelia using methylation‐specific PCR. Protein expression of SFRP genes was then assessed in these tissues by immunohistochemistry. The mRNA expression of SFRP genes was quantified by real‐time reverse‐transcription PCR in esophageal adenocarcinoma cell lines with and without demethylation by 5‐aza‐2′deoxycytidine and inhibition of deacetylation by trichostatin A treatment. Hypermethylation of SFRP1, 2, 4 and 5 was detected in 93%, 83%, 73% and 85% of 40 cancers; 81%, 89%, 78% and 73% of 37 Barretts epithelia; 25%, 64%, 32% and 21% of 28 adjacent normal epithelia from Barretts patients; and 10%, 67%, 0% and 13% of 30 normal esophagogastric epithelia from healthy individuals, respectively (p < 0.001 for SFRP1, 4 and 5; p < 0.05 for SFRP2). Protein expression of SFRP1, 2 and 4 was downregulated in 87%, 67% and 90% of cancers, and expression correlated inversely with grade and stage of cancers and with grade of dysplasia. Expression of SFRP2 and SFRP4 proteins was lower in cancers with corresponding gene methylation (p < 0.05). Demethylation treatment effectively re‐expressed SFRP mRNA in cancer cell lines. Thus, hypermethylation of SFRP genes is a common early event in the evolution of esophageal adenocarcinoma, and methylation of SFRP1, 4 and 5 might serve as biomarkers for Barretts neoplasia. Aberrant promoter methylation appears to functionally silence SFRP gene expression in esophageal adenocarcinoma.

Highly Methylated Genes in Colorectal Neoplasia: Implications for Screening

Discriminant markers are required for accurate cancer screening. We evaluated genes frequently methylated in colorectal neoplasia to identify the most discriminant ones. Four genes specifically methylated in colorectal cancer [bone morphogenetic protein 3 (BMP3), EYA2, aristaless-like homeobox-4 (ALX4), and vimentin] were selected from 41 candidate genes and evaluated on 74 cancers, 62 adenomas, and 70 normal epithelia. Methylation status was analyzed qualitatively and quantitatively and confirmed by bisulfite genomic sequencing. Effect of methylation on gene expression was evaluated in five colon cancer cell lines. K-ras and BRAF mutations were detected by sequencing. Methylation of BMP3, EYA2, ALX4, or vimentin was detected respectively in 66%, 66%, 68%, and 72% of cancers; 74%, 48%, 89%, and 84% of adenomas; and 7%, 5%, 11%, and 11% of normal epithelia (P < 0.01, cancer or adenoma versus normal). Based on area under the curve analyses, discrimination was not significantly improved by combining markers. Comethylation was frequent (two genes or more in 72% of cancers and 84% of adenomas), associated with proximal neoplasm site (P < 0.001), and linked with both BRAF and K-ras mutations (P < 0.01). Cell line experiments supported silencing of expression by methylation in all four study genes. This study shows BMP3, EYA2, ALX4, and vimentin genes are methylated in most colorectal neoplasms but rarely in normal epithelia. Comethylation of these genes is common, and pursuit of complementary markers for methylation-negative neoplasms is a rational strategy to optimize screening sensitivity. (Cancer Epidemiol Biomarkers Prev 2007;16(12):2686–96)

High Detection Rates of Colorectal Neoplasia by Stool DNA Testing With a Novel Digital Melt Curve Assay

BACKGROUND & AIMS Current stool DNA tests identify about half of individuals with colorectal cancers and miss most individuals with advanced adenomas. We developed a digital melt curve (DMC) assay to quantify low-abundance mutations in stool samples for detection of colorectal neoplasms and compared this test with other approaches. METHODS We combined a melt curve assay with digital polymerase chain reaction and validated the quantitative range. We then evaluated its ability to detect neoplasms in 2 clinical studies. In study I, stool samples from patients with colorectal tumors with known mutations (KRAS, APC, BRAF, TP53) were assayed. In study II, archived stool samples from patients with advanced adenomas containing known KRAS mutations were assayed, along with controls. Results were compared with those from the stool DNA test PreGenPlus (Exact Sciences, Marlborough, MA), Hemoccult, and HemoccultSensa (both Beckman-Coulter, Fullerton, CA). RESULTS The DMC assay detected samples in which only 0.1% of target genes were mutated. In study I, the DMC assay detected known mutations in 28 (90%) of 31 tumor samples and 6 (75%) of 8 advanced adenoma samples. In study II, the DMC assay detected 16 (59%) of 27 advanced adenoma samples that contained KRAS mutations, compared with 7% with the Hemoccult, 15% with the HemoccultSensa, and 26% with the PreGenPlus assays (P < .05 for each, compared with the DMC assay); specificities did not differ significantly. CONCLUSIONS The DMC assay has a high level of sensitivity in detecting individuals with colon neoplasms and is better than current stool screening methods in detecting those with advanced adenomas. Further studies are indicated.

A Sensitive Method to Quantify Human Long DNA in Stool: Relevance to Colorectal Cancer Screening

Human long DNA in stool may reflect nonapoptotic exfoliation and has been used as a colorectal cancer (CRC) marker. Targeting human-specific Alu repeats represents a logical but untested approach. A real-time Alu PCR assay was developed for quantifying long human DNA in stool and evaluated in this study. The accuracy and reproducibility of this assay and the stability of long DNA during room temperature fecal storage were assessed using selected patient stools and stools added to human DNA. Thereafter, long DNA levels were determined in blinded fashion from 18 CRC patients and 20 colonoscopically normal controls. Reproducibility of real-time Alu PCR for quantifying fecal long DNA was high (r2 = 0.99; P < 0.01). Long DNA levels in nonbuffered stools stored at room temperature fell a median of 75% by 1 day and 81% by 3 days. An EDTA buffer preserved DNA integrity during such storage. Human long DNA was quantifiable in all stools but was significantly higher in stools from CRC patients than from normal controls (P < 0.05). At a specificity of 100%, the sensitivity of long DNA for CRC was 44%. Results indicate that real-time Alu PCR is a simple method to sensitively quantify long human DNA in stool. This study shows that not all CRCs are associated with increased fecal levels of long DNA. Long DNA degrades with fecal storage, and measures to stabilize this analyte must be considered for optimal use of this marker. (Cancer Epidemiol Biomarkers Prev 2006;15(6):1115–9)

Frequent Methylation of Eyes Absent 4 Gene in Barrett's Esophagus and Esophageal Adenocarcinoma

Most esophageal adenocarcinomas arise within Barretts esophagus but the cause of this increasingly prevalent condition remains unknown. Early detection improves survival and discriminant screening markers for Barretts esophagus and cancer are needed. This study was designed to explore the natural history of eyes absent 4 (EYA4) gene methylation in the neoplastic progression of Barretts esophagus and to evaluate methylated EYA4 as a candidate marker. Aberrant promoter methylation of EYA4 was studied by methylation-specific PCR using bisulfite-treated DNA from esophageal adenocarcinomas, Barretts esophagus, and normal epithelia, and then confirmed by sequencing. Eight cancer cell lines were treated with the demethylation agent 5-aza-2′-deoxycytidine, and EYA4 mRNA expression with and without treatment was quantified by real-time reverse-transcription PCR. EYA4 hypermethylation was detected in 83% (33 of 40) of esophageal adenocarcinomas and 77% (27 of 35) of Barretts tissues, but only in 3% (2 of 58) of normal esophageal and gastric mucosa samples (P < 0.001). The unmethylated cancer cell lines had much higher EYA4 mRNA expression than the methylated cancer cell lines. Demethylation caused by 5-aza-2′-deoxycytidine increased the mRNA expression level by a median of 3.2-fold in methylated cells, but its effect on unmethylated cells was negligible. Results indicate that aberrant promoter methylation of EYA4 is very common during tumorigenesis in Barretts esophagus, occurs in early metaplasia, seems to be an important mechanism of down-regulating EYA4 expression, and represents an intriguing candidate marker for Barretts metaplasia and esophageal cancer.

Prospective evaluation of fecal calprotectin as a screening biomarker for colorectal neoplasia.

OBJECTIVES:Stool testing is a well established method of screening for colorectal neoplasia. Emerging data suggest that novel biomarkers may offer performance advantages over fecal occult blood. In this large, prospective study, we assessed fecal calprotectin (a leukocyte-derived protein) as a screening biomarker for colorectal neoplasia. Fecal calprotectin was directly compared to fecal hemoglobin (Hb) and colonoscopy as the existing criterion standards for stool screening and structural evaluation, respectively.METHODS:Subjects included colonoscopy patients with a personal history of colorectal neoplasia, family history of colorectal cancer, or iron deficiency anemia. Stool specimens were collected before purgation, processed appropriately, and quantitatively analyzed for calprotectin (Nycomed Pharma, Oslo, Norway) and for Hb (Mayo Medical Laboratories, Rochester, MN) by masked technicians. Colonoscopies were performed by experienced endoscopists without prior knowledge of the fecal assay results.RESULTS:Among 412 subjects, 97 (24%) subjects had one or more colorectal neoplasms (including three with adenocarcinomas). Fecal calprotectin levels did not differ significantly between subjects with versus subjects without colorectal neoplasms (p= 0.33). Neither tumor number (p= 0.85) nor tumor size (p= 0.86) significantly influenced the observed fecal calprotectin concentrations. Estimates of the sensitivity, specificity, and positive and negative predictive values of fecal calprotectin for any colorectal neoplasms were 37%, 63%, 23%, and 76%, respectively. Comparable performance estimates for fecal Hb were 3%, 97%, 27%, and 77%, respectively.CONCLUSIONS:In this cohort of colonoscopy patients at above average risk, fecal calprotectin was a poor screening biomarker for colorectal neoplasia. Further investigation of tumor-derived, rather than blood-based, biomarkers may be a more rewarding approach to stool screening for colorectal neoplasia.

Quantification of Methylated Markers with a Multiplex Methylation-Specific Technology

BACKGROUND Aberrantly methylated genes represent important markers for cancer diagnosis. We describe a multiplex detection approach to efficiently quantify these markers for clinical applications such as colorectal cancer screening. METHODS Quantitative allele-specific real-time target and signal amplification (QuARTS) combines a polymerase-based target amplification with an invasive cleavage-based signal amplification. The fluorescence signal is detected in a fashion similar to real-time PCR. We measured the dynamic range and analytical sensitivity of multiplex QuARTS reactions with titrated plasmid DNA. We used the QuARTS technology to quantify methylated BMP3, NDRG4, VIM, and TFPI2 genes on 91 DNA samples extracted from colorectal tissues, including 37 cancers, 25 adenomas, and 29 healthy epithelia. The assays were designed in triplex format that incorporated ACTB as a reference gene. Percent methylation was calculated by dividing methylated strands over ACTB strands and multiplying by 100. RESULTS The QuARTS method linearly detected methylated or unmethylated VIM gene down to 10 copies. No cross-reactivity was observed when methylated assays were used to amplify 10(5) copies of unmethylated gene and vice versa. The multiplex assay detected methylated genes spiked in unmethylated genes at a 0.01% ratio and vice versa. At a diagnostic specificity cutoff of 95%, methylated BMP3, NDRG4, VIM, and TFPI2 detected 84%, 92%, 86%, and 92% of colorectal cancers and 68%, 76%, 76%, and 88% of adenomas, respectively. CONCLUSIONS The QuARTS technology provides a promising approach for quantifying methylated markers. The markers assayed highly discriminated colorectal neoplasia from healthy epithelia.

Detection of Occult Upper Gastrointestinal Tract Bleeding: Performance Differences in Fecal Occult Blood Tests

OBJECTIVE To compare rates of detection of occult upper gastrointestinal (GI) tract bleeding by guaiac (Hemoccult II [HO]), immunochemical (HemeSelect [HS]), and heme-porphyrin (HemoQuant [HQT]) fecal occult blood tests. PATIENTS, SUBJECTS, AND METHODS: In a cross-sectional study to detect native occult upper GI tract bleeding, single stools were collected from 56 patients with iron deficiency and a proven hemorrhagic GI tract lesion. In a longitudinal study to detect simulated occult upper GI tract bleeding, 3 stool samples were serially collected from 10 clinically normal subjects after ingestion of 5 and 15 mL of autologous blood. All stool samples were subjected to blinded fecal occult blood determinations with use of the 3 tests. RESULTS In the cross-sectional study, the HQT test detected 88% (37/42) of hemorrhagic upper GI tract lesions compared with 26% (11/42) detected by the HO test (P<.001) and 2% (1/42) by the HS test (P<.001). In the longitudinal study, all preingestion fecal occult blood test results were negative. After ingestion of 5 mL of blood, the HQT result became positive in 60% (6/10), and the HO and HS results remained negative (P=.03). After ingestion of 15 mL of blood, the HQT result became positive in all 10 cases, the HO result was positive in 6 (P=.12 vs HQT), and the HS result was positive in none (P=.002 vs HQT); all 3 stool samples collected after the 15-mL ingestion were positive in each of the 10 subjects by the HQT test but in only 1 subject by the HO test (P=.003). CONCLUSION The HQT test detects occult upper GI tract blood loss significantly more frequently than the HO or HS test.

Altered DNA Mismatch Repair Expression in Synchronous and Metachronous Colorectal Cancers

BACKGROUND & AIMS It is not clear what proportion of synchronous or metachronous colorectal cancers (CRCs) are associated with DNA mismatch repair (MMR) alterations or unsuspected Lynch syndrome. On the basis of tissue analyses, the aims were to evaluate DNA MMR expression in metachronous, synchronous, and isolated sporadic CRCs and to assess within-patient concordance of MMR expression in metachronous and synchronous CRCs. METHODS Tissue was evaluated from 34 patients with metachronous CRC, 34 matched solitary CRC patients, and 40 patients with synchronous CRCs. Subjects with known hereditary CRC were excluded. Immunohistochemical staining for MLH1 and MSH2 was performed on all tissues. RESULTS Absent MLH1 or MSH2 staining of the initial metachronous tumor was observed in 27% compared with 21% of control CRCs, P = .58. The odds of metachronicity with absent immunostaining were 1.33 (95% confidence interval, 0.46-3.84). Loss of MMR expression was observed in at least one cancer in 30% of patients with synchronous CRC. MMR expression loss was discordant in 70% of metachronous CRCs and 50% of synchronous CRCs. Lynch syndrome was subsequently diagnosed in 2 patients with synchronous CRCs, both with concordant tumor MMR loss but in no patient with metachronous CRC. CONCLUSIONS Among those without known Lynch syndrome, development of multiple primary CRCs appears to be due largely to somatic events and often occurs via different molecular pathways within the same patient. Altered MMR expression in sporadic CRC has low predictive value for metachronicity. MMR expression analysis in cases of multiple primary CRC might identify a small number of patients with unsuspected Lynch syndrome.