Fecal DNA for Noninvasive Diagnosis of Colorectal Cancer in Immunochemical Fecal Occult Blood Test–Positive Individuals

The clinical outcome and quality of life of colorectal cancer patients have markedly improved over the last 20 years. This achievement is due partly to progress made in medical and surgical treatments, and to advances in the early noninvasive diagnosis of small tumors and preneoplastic lesions at high risk of malignant transformation. Cancers of the colon, like those of the breast and uterine cervix, are good targets for early diagnosis because they are often preceded by preneoplastic lesions that typically have a long natural history. In fact, the overall 5-year survival for colorectal cancer patients is around 64%, rising to almost 90% if tumors are diagnosed early (1, 2).

The fecal occult blood test (FOBT) is the most common diagnostic procedure for the early detection of colorectal cancer, and although it reduces mortality by 13% to 33% depending on the test interval and the specific methodologies used (3-6), it does present important limitations, i.e., high number of colorectal cancers missed, suboptimal sensitivity in detecting precancerous lesions such as polyps and adenomas, and difficulty in distinguishing between low- and high-risk lesions in the latter subgroup. These problems have still not been resolved, despite notable improvements in FOBT sensitivity and specificity and new screening strategies (6-8).

The analysis of DNA alterations present in tumor cells exfoliated into feces could represent an interesting alternative to or support for noninvasive approaches currently used. Although the majority of studies published on this subject offer somewhat preliminary data, the biological rationale for such a diagnostic procedure seems very promising (9-12). Indeed, although bleeding is an intermittent and largely unspecific event, cellular exfoliation is a continuous process, which could thus improve test sensitivity. Normal colon epithelium renews itself each day, and this process is coupled with a continuous exfoliation of cells in the intestinal lumen (13, 14). Moreover, a number of studies have shown that cellular exfoliation is more pronounced in colorectal cancer patients than in healthy individuals. Although colon cancers typically account for <1% of the total colon surface area, tumor cell DNA actually makes up between 14% and 24% of the total DNA recovered in feces (15, 16). The first evidence that fecal DNA quantity and quality might be good indicators of the presence of neoplastic cells emerged more than 10 years ago (15). Since then studies have shown that the simple analysis of fecal DNA fragments greater than 200 bp is capable of identifying patients with tumors or precancerous lesions (16-18). The discrepancy between cancer patients and healthy individuals could reflect the different degradation pathways followed by cells once they pass into the lumen. Normal cells not yet undergoing apoptosis immediately activate this pathway, initiating DNA cleavage and producing small molecules of around 150 to 200 bp in length (17). Conversely, neoplastic cells are able to survive much longer due to the inhibition of apoptotic and cell death processes. Once in the lumen, tumor cells are almost certainly destroyed by bile salts, leading to the break-up of cell membranes. In this instance, however, DNA is less damaged and in many cases is composed of fragments of more than 150 to 200 bp in length.

In our laboratory, the method currently used to analyze fecal DNA integrity was modified to ensure a quantitative evaluation [fluorescence long DNA (FL-DNA)], obtaining specificity and sensitivity of up to 80% (19). After completing pilot and confirmatory case-control studies (19, 20), our next step, in accordance with international guidelines (21, 22), was to validate the marker in a large series of candidates for colorectal screening before being able to transfer the application to clinical practice.

As the FOBT test has been shown to produce a high number of false-positive results, we also aimed to assess whether the FL-DNA test could improve specificity in a large series of consecutively enrolled immunochemical iFOBT (iFOBT)-positive individuals by reducing false-positive results.

Study subjects were recruited from a colorectal cancer population screening program run by the Oncology Prevention Unit of Morgagni-Pierantoni Hospital (Forlì, Italy). The screening program, activated in March 2005, invites men and women aged between 50 and 69 years to undergo a simple test every two years for the detection of fecal occult blood. People testing positive are invited to undergo colonoscopy to verify the nature of the occult blood.

A total of 560 FOBT-positive individuals were enrolled in the study, and all were submitted to endoscopic examination to confirm diagnosis; 26 were diagnosed with adenocarcinomas, 264 with high-risk adenomas, and 54 with low-risk adenomas. More than one third of the series (n = 216) did not show any malignant or premalignant lesion at endoscopic examination. The type of lesion was histologically confirmed, and, in cancer patients, pathologic stage was defined according to Dukes' classification. Preneoplastic lesions were classified as low- or high-risk according to National Comprehensive Cancer Network guidelines (23). Specifically, patients were considered at high-risk when they had high-risk dysplasia, >3 adenomatous villous or tubulovillous polyps, at least one of which with a diameter of ≥1 cm, or an in situ carcinoma, whereas those who presented with <3 tubular polyps with a diameter <1 cm were considered at low-risk. Lesions were divided into subgroups according to localization: proximal colon, from cecum to transverse colon; distal colon, from descending colon to sigmoid colon; rectum; and mixed, lesions affecting both proximal and distal colon.

The study protocol was reviewed and approved by the local ethics committee. Written informed consent was obtained from all individuals agreeing to take part in the study.

The fecal occult blood test used is based on an immunological analysis (Alfa Wassermann) and was done only once, in accordance with Regional Screening Program guidelines. Individuals were provided with instructions for collecting the fecal material at home, and samples had to be brought to the analysis laboratory within 24 hours.

In accordance with regional guidelines, test positivity was defined as a hemoglobin value ≥100 ng/mL. Hemoglobin values are determined using an immunochemical technique (OC Micro Sensor). Polystyrene latex particles coated with human hemoglobin are mixed with the fecal sample. Antibodies react with the latex agglutination antigen, resulting in a change in the turbidity of the sample. As there is a direct proportionality between antigen concentration and increased turbidity, the concentration of hemoglobin can be photometrically calculated. There is no pretest dietary preparation.

The stool samples were the same as those used for the hemoglobin analysis. Specifically, iFOBT-positive feces (about 10 mg) remaining on the dipstick after occult blood test analysis were used for the FL-DNA test. Samples were immediately processed for DNA extraction or stored at −20°C for a maximum of 2 months on the basis of results from preliminary experiments on DNA stability.

After a preliminary study in which we assessed the reliability of the analysis carried out on smaller quantities (5-10 mg) of stool compared with the 200 mg previously used by our group (19, 20), stool samples were thawed at room temperature and homogenized with 10 mL of TE-9 buffer [0.5 mol/L Tris-HCl (pH 9), 20 mmol/L EDTA, and 10 mmol/L NaCl]. After centrifugation at 5,000 g for 15 minutes to remove all particulate matter, the supernatant was transferred to a tube containing 150 μL of 7.5 mol/L ammonium acetate (M-Medical) and 930 μL of 100% ethanol (Carlo Erba). DNA was recovered by centrifugation at 5,000 g for 15 minutes at room temperature, resuspended in 1.6 mL of ASL buffer, and purified by QIAamp DNA Stool Kit (Qiagen).

The fluorescence intensity of each sample-specific PCR product was determined with fluorescent-labeled primers, as described previously (20). Briefly, exons 5 to 8 of p53 and four fragments of Adenomatous polyposis coli (APC) exon 15 were amplified in a final volume of 25 μL containing 2 μL of stool DNA, 0.4 μmol/L of each primer, 200 μmol/L of deoxynucleotides (Takara Bio Inc), 1× reaction buffer with 3.5 mmol/L MgCl₂ (Qiagen), and 1 U of Taq polymerase (Qiagen). The reaction mixture was subjected to 32 cycles: 60 seconds at 94°C, 60 seconds at 60°C for p53 exons, and 58°C for APC fragments, followed by incubation at 72°C for 60 seconds. Primers used were end-labeled with fluorochromes provided by Applied Biosystems.

The p53 exons were amplified simultaneously in a single reaction mixture, whereas the four APC fragments were amplified in two different mixes (mix 1, fragments 3 and 4; mix 2, fragments 1 and 2). Electrophoresis was carried out using a 3100 Avant Genetic Analyzer (Applied Biosystems) equipped with GeneScan Analysis 3.7. The final FL-DNA value was obtained by analyzing the fluorescence intensity of each sample-specific PCR product against a reference standard curve (1, 2, 5, 10, and 20 ng) of genomic DNA, and expressed as nanograms. Three different standard curves were prepared: p53 exons 5 to 8, APC fragments 1 to 2, and APC fragments 3 to 4. All samples were run in duplicate and only intersample variations of <15% were accepted. In all other cases (15% of the series), the determination was done on a third sample, with only <10% variations accepted for the entire series. Samples showed no variations >10% at this third evaluation. All samples were amplified with a reaction mix containing 25 attograms of a plasmid with a control sequence to verify the presence of Taq inhibitors.

The relationship between FL-DNA and iFOBT values and clinical-histopathologic characteristics was analyzed using nonparametric ranking statistics (median test), whereas Spearman's coefficient (rs) was used to investigate the relationship between FL-DNA and FOBT. To estimate post-test probability, i.e., the probability of disease in a subject after the diagnostic test results are known, we first estimated the pretest probability and determined the likelihood ratio. The pre-test probability is the chance of having the disease prior to testing and this is usually related to the disease prevalence. The likelihood ratio is the ratio of the probability of the specific test result in people who do have the disease to the probability in people who do not. The results were divided into three classes according to different cutoff values (0-9, 10-29, and ≥30 ng) to determine the FL-DNA likelihood ratio, which was calculated by dividing the percentage of patients with colorectal cancer by the percentage without disease in each class. Finally, post-test probability was calculated by multiplying the likelihood ratio of the diagnostic test result by the pre-test probability. All P values were two-sided and values ≤0.05 were considered statistically significant. All analyses were carried out with SAS Statistical package version 9.1 (SAS Institute Inc.).

In the overall series, iFOBT hemoglobin values ranged from 101 to 5,826 ng/mL, with a median value of 289 ng/mL. The lowest levels were observed in individuals with no lesions (median value, 222 ng/mL) and in low-risk adenoma patients (246 ng/mL). A somewhat higher but not statistically different median value (327 ng/mL) was recorded in patients with high-risk adenomas. Considered together, the median iFOBT value for these three subgroups was about five times lower and statistically different (P = 0.0001) from that observed for cancer patients (1,511 ng/mL; Table 1).

Similar results were observed for FL-DNA values. In particular, median values were comparable for individuals with no lesions (13 ng) or with low- and high-risk adenomas (12 ng), and were >3-fold higher (38 ng; P < 0.0001) in cancer patients (Table 1). Correlation coefficient analysis carried out on the overall series showed that FL-DNA and FOBT values were largely independent variables (rs = 0.036, P = 0.42). The relationship between iFOBT and FL-DNA values within the different clinical and pathologic subgroups was analyzed separately in adenoma and cancer patients. In adenoma patients, FOBT median values were significantly higher in those with pedunculated rather than sessile adenomas (P = 0.003), as well as in patients with larger (>2 cm) rather than smaller lesions (P = 0.0001; Table 2). The FL-DNA median value was higher in older than in younger patients with adenoma (P = 0.04). Moreover, in contrast to what was observed for iFOBT, the DNA median value was higher in patients with sessile rather than pedunculated lesions (P = 0.04) and in patients with mixed lesion localization compared with those with adenomas in the proximal or distal colon or rectum (P = 0.02; Table 2).

In cancer patients, FOBT values were not related to either clinical (age, gender) or pathologic (stage, tumor site) characteristics, whereas median FL-DNA levels were about 3-fold higher in females than in males (P = 0.07), and significantly (P = 0.02) and progressively higher in patients with tumors sited in the proximal or distal colon or rectum (Table 3).

Finally, we evaluated whether the combination of iFOBT and FL-DNA could improve our ability to predict the presence of a tumor. Taking into account the diagnostic relevance of fecal hemoglobin and DNA as independent variables, we tested whether and to what extent the FL-DNA assay could improve iFOBT diagnostic accuracy. The nomogram provides a graphical tool for estimating how much the result of a test changes the probability that a patient has the disease (Fig. 1). A line is drawn connecting the pretest probability and the point on the middle vertical line corresponding to the likelihood ratio for the test result, represented by a range of test results (boxes). This line is extended to intersect with the right-hand vertical line, which gives the post-test probability. This point is the new estimate of probability that the patient has the disease. These data are also summarized in Table 4. In this first stage of the study the analysis was limited to iFOBT-positive individuals. We started from three main subgroups at low (101-200 ng/mL), intermediate (>200-1,000 ng/mL), and high (>1,000 ng/mL) iFOBT levels and used different retrospectively defined DNA cutoffs. In the low-iFOBT subgroup, the probability of there being a tumor was 0, and FL-DNA did not add any useful information. Conversely, in the two classes at intermediate and high FOBT levels, the probability of cancer was modulated by FL-DNA values. Specifically, in the intermediate iFOBT subgroup with its 4.6% overall probability of having cancer, breakdown analysis as a function of FL-DNA highlighted a subgroup with <9 ng of FL-DNA that had a 0.9% probability of having cancer and, on the other hand, a subgroup with ≥30 ng of DNA with a 13% probability of having neoplastic disease. A similar modulation was observed for the high FOBT level subgroup in which the probability of having cancer increased up to 30% as FL-DNA levels increased, with a maximum cancer prevalence of 32%. Finally, it is worth mentioning that, of the 15 individuals with fecal DNA levels >100 ng, 5 had cancer and 5 high-risk adenomas, independently of FOBT values, which ranged from 109 to 2,282 ng/mL.

There is no standard screening program for colorectal cancer, each country implementing its own, with consequent limitations and advantages (24). However, the overall benefits of colorectal cancer screening in reducing cancer mortality are widely acknowledged. This important result is ascribable to early cancer detection, which improves prognosis, but also to the removal of nonmalignant precursor lesions, which prevents cancer development. Although the FOBT is the standard noninvasive approach for early colorectal cancer diagnosis, it fails to detect a large number of cancers (6, 25). Moreover, because FOBT is based on the analysis of occult bleeding, which is not necessarily associated with cancer, numerous false-positive results ensue, leading to unnecessary invasive endoscopic examinations and further psychological stress for patients.

A new and promising strategy for the noninvasive diagnosis of colorectal cancer is based on the analysis of DNA from exfoliated cells excreted in stool (9, 10, 12). Results comparing stool genomic DNA analysis with the nonrehydrated Hemoccult II FOBT test in a large prospective screening cohort showed higher sensitivity in detecting all cancers, and comparable specificity (26).

In preliminary case-control studies, we showed the potential of fecal DNA levels in detecting colorectal cancer (19, 20). In the present study we compared the diagnostic accuracy of fecal DNA and fecal occult blood in a large series of iFOBT-positive individuals enrolled in a screening program and for whom histologic diagnosis through endoscopic examination was available.

Furthermore, we modified our previously used methodology (19, 20) to carry out iFOBT and FL-DNA analysis on the same small sample of stool.

First, our results confirmed that iFOBT and fecal DNA levels were significantly higher in cancer patients than in patients with low- or high-risk adenomas and in healthy individuals. Second, in patients with neoplastic and preneoplastic lesions, analysis on the overall series showed that fecal DNA and FOBT values were variously related to the different lesion and patient characteristics and were largely independent variables, as highlighted by the low correlation coefficient. Specifically, in adenoma patients, iFOBT values were significantly and directly related to the dimension and macroscopic pedunculated appearance of the adenoma, suggesting that structural characteristics cause a more extensive extrusion of the lesion surface into the lumen and probably a higher risk of abrasion with consequent bleeding. Conversely, fecal FL-DNA values were significantly higher in patients with sessile rather than with pedunculated lesions and in those with mixed rather than with single lesion localizations, indicating a different relationship between the entire lesion surface and exfoliation of proliferating cells into the colorectal lumen. These data also suggest that there may be a correlation between DNA integrity and/or occult blood levels and some molecular alterations related to adenoma subtypes characterized by exfoliation or bleeding, i.e., proliferation or apoptosis markers. The identification of biological parameters correlated with these markers could enable us to better define the role and importance of the double approach (iFOBT and FL-DNA) for the early diagnosis of colorectal cancer.

Interesting results were also obtained by evaluating high- and low-risk adenoma subgroups together. In fact, FOBT values were slightly higher in high-risk adenomas, whereas FL-DNA values were equally distributed in the two subgroups. It must be pointed out that adenoma risk classification was based on pathologic parameters. Conversely, the two markers were probably correlated with other biological characteristics that would be interesting to analyze using a molecular approach.

The novelty of the present study is the sequential use of the diagnostic iFOBT and FL-DNA methods. The double assay showed that, in the subgroups with medium and high iFOBT values, fecal DNA provided more accurate diagnostic information and identified small subgroups with a very different probability of having a tumor.

In conclusion, our results suggest that it could be useful to consider a combined approach based on different markers capable not only of predicting the presence of neoplastic lesions but also of determining a patient's risk of harboring a preneoplastic or neoplastic lesion. Such information could help clinicians in planning diagnostic tests. However, a number of key issues remain to be clarified to improve the diagnostic accuracy of the tests under investigation, the most important being the frequency with which the tests should be carried out, and the number of stool samples that need to be analyzed at specific time points for each individual. It must also be remembered that, although DNA analysis has high sensitivity to detect colorectal cancer, it is, like FOBT, unable to accurately identify all high-risk adenomas, which are important targets for early detection. Moreover, our study could not compare the accuracy of either approach as it was limited to patients with iFOBT-positive results, and such data cannot be applied to a screening setting. New methods of increasing diagnostic accuracy for early lesions are thus being evaluated in our laboratory. One such approach is the detection of very small differences in DNA integrity level using real-time PCR. Another is the analysis of epigenetic markers associated with adenoma-cancer transition and with the first phases of tumor transformation.

No potential conflicts of interest were disclosed.

We thank Gráinne Tierney for editing the manuscript.

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.