Detection of Colorectal Cancer by a Quantitative Fluorescence Determination of DNA Amplification in Stool

Abstract

DNA amplification of exfoliated cells in stool represents an inexpensive and rapid test, but has only 50% to 60% sensitivity. A new quantitative method, called fluorescence long DNA, was developed and validated in our laboratory on stool obtained from 86 patients with primary colorectal cancer and from 62 healthy individuals. It consists of the amplification of stool DNA with fluorescence primers and the quantification of the amplification using a standard curve. Results are arbitrarily expressed in nanograms. The potential of the new method compared to the conventional approach was analyzed in a subgroup of 94 individuals (56 patients and 38 healthy volunteers). In the present series, DNA amplification analysis showed a specificity of 97% and a sensitivity of only 50%. Conversely, fluorescence DNA evaluation, using the best cutoff of 25 ng, showed a sensitivity of about 76% and a specificity of 93%. Similar sensitivity was observed regardless of Dukes stage, tumor location, and size, thus also permitting the detection of early-stage tumors. The present study seems to indicate that quantitative fluorescence DNA determination in stool successfully identifies colorectal cancer patients with a sensitivity comparable, if not superior, to that of multiple gene analysis but at a lower cost and in a shorter time.

Introduction

In recent years, a great deal of information has been accumulated on the molecular alterations that take place during the development of tumors, such as gene mutations or genomic rearrangements, highlighting the possibility of detecting tumor alterations in biologic fluids and, consequently, indicating the use of these markers as a valid noninvasive diagnostic approach.

A tumor that has been widely investigated with this approach is colorectal cancer, which is one of the most common forms of cancer worldwide, with a clinical outcome varying considerably according to the type of lesion and stage of disease at diagnosis [1–3]. An early diagnosis is fundamental to reduce morbidity and mortality because a high percentage of patients diagnosed in the early stages of disease comprises long-term survivors [4]. Moreover, the possibility of detecting premalignant lesions makes this tumor an ideal target for screening programs. However, although several screening methods are available, a high percentage of individuals does not participate in colorectal cancer screening programs. There are many reasons for this low compliance, such as a lack of knowledge of the benefits of available screening methods, especially colonoscopy, as well as the unpleasant and troublesome procedures [5].

Gene mutations in stool, especially K-ras [6–12] and, to a lesser extent, p53 [13], APC gene [14,15], and microsatellite instability [16], have been repeatedly investigated. Results have shown the presence of these molecular alterations in stool in only a fraction of patients, due to the relatively low frequency of single marker alterations in colorectal cancer. Multiple mutations have been analyzed in parallel on the same stool sample, and this approach has led to improved test sensitivity, but is expensive, time-consuming, and cannot easily be applied to screening programs [17–21].

The diagnostic potential of DNA amplification of exfoliated cells in stool has recently been considered. Preliminary evidence [19–21] has shown that the semiquantitative evaluation of DNA amplification (long DNA, or L-DNA) of some DNA fragments longer than 200 bp detects more than 50% of colorectal cancers, with a very high specificity.

In the present study, we aimed to discuss the results obtained from this inexpensive and rapid approach, both quantitative and objective, to increase its accuracy and thus permit a better discrimination between affected and nonaffected individuals. For this purpose, we assessed the diagnostic potential of a new DNA amplification method (fluorescence long DNA, or FL-DNA) on a series of patients and healthy donors.

Patients and Methods

Case Series

Stool samples from 86 patients with primary colorectal cancer were collected in the Gastroenterology Unit and Department of Surgery I, Morgagni Hospital (Forlì, Italy) and in the Departments of Oncology and General Surgery, Infermi Hospital (Rimini, Italy). Stool samples were collected from 62 individuals who proved negative for cancer or benign lesions after colonoscopy, and from laboratory personnel.

Stool samples were obtained at least 3 days after the administration of laxative treatments in preparation for colonoscopy to allow for the recovery of normal bowel functionality. The fecal specimens were immediately frozen and stored at -70°C for a maximum of 2 months.

Cancer diagnosis was histologically confirmed and pathological stage was defined according to Dukes classification: 8 tumors were classified as stage A, 30 as stage B, 37 as stage C, and 9 as stage D. Moreover, 19 cancers were located in ascending colon, 30 in descending colon, 2 in transverse colon, and 35 in the rectal tract. Staging information was not available for only two cases.

Of the 86 patients, 42 were male and 44 were female, and median age was 72 years (range 36–90 years). Of the 62 controls, 29 were male and 33 were female, and median age was 51 years (range 21–87 years).

DNA Purification

Approximately 4 g of stool was thawed at room temperature. DNA was extracted after a 15-minute homogenization with 16 ml of TE-9 buffer pH 9 (0.5 M Tris-HCl, 20 mM EDTA, and 10 mM NaCl) by ULTRA-Turrax T25 (Janke and Kunkel GmbH and Co. KG IKA-Labortechnik, Staufen, Germany). After centrifugation at 5000g for 15 minutes, the supernatant was transferred to a tube containing 5 ml of 7.5 M ammonium acetate (M-Medical, Florence, Italy) and 30 ml of 100% ethanol (Carlo Erba, Milan, Italy). DNA was recovered by centrifugation at 5000g for 15 minutes at room temperature. Stool samples were suspended in 1.6 ml of ASL buffer and DNA was extracted using the QIAamp DNA Stool Kit (QIAGEN, Hilden, Germany).

L-DNA Analysis

p53 exons 5 to 8 and fragments 1 to 4 of APC exon 15 were amplified in a final volume of 25 µl containing 2 µl of DNA from stool, 0.4 µM of each primer, 200 µM deoxynucleotide triphosphates, 1 x reaction buffer with 3.5 mM MgCl₂, and 1 U of Taq polymerase (QIAGEN). The reaction mixture was subjected to 38 polymerase chain reaction (PCR) cycles: 60 seconds at 94°C, 60 seconds at 58°C, and 60 seconds at 72°C. Primer sequences without fluorescence-labeled 5′ ends have been described previously. Gel electrophoresis was performed by running 5 µl of PCR reaction in 2% agarose gel. In parallel, 5 µl of K-ras PCR product and a control plasmid were used to verify the presence of Taq inhibitors in fecal DNA samples. The relative amplification intensities of p53 exons 5 to 8 and fragments 1 to 4 of APC were analyzed independently by two operators and classified as high, medium, low, or not detectable. An interobserver concordance was observed in all cases.

This semiquantitative L-DNA analysis was performed in parallel with the quantitative FL-DNA approach in 94 individuals (56 patients and 38 healthy volunteers) recruited during the first period of the study.

FL-DNA Analysis

Amplifications of exons 5 to 8 of p53 and fragments 1 to 4 of APC exon 15 were carried out on 2 µl of DNA from stool in a total volume of 25 µl containing 0.4 µM of each primer, 200 µM deoxynucleotide triphosphates, 1 x reaction buffer with 3.5 mM MgCl₂, and 1 U of Taq polymerase (QIAGEN). The reaction mixture was subjected to 32 cycles: 60 seconds at 94°C and then 60 seconds at 60°C for p53 exons, and 58°C for APC fragments, followed by incubation at 72°C for 60 seconds.

The p53 exons were amplified simultaneously in a single reaction mixture and the four APC fragments were amplified in two different mixes (mix 1: fragments 1 and 2; mix 2: fragments 3 and 4). For this purpose, primers used for L-DNA analysis and those previously described [21] were end-labeled with fluorochromes provided by Applied Biosystems (Foster City, CA).

Electrophoresis was carried out using a 3100 Avant Genetic Analyzer (Applied Biosystems) equipped with GeneScan Analysis 3.7.

FL-DNA was performed by analyzing the fluorescence intensity of each sample-specific PCR product. The quantification of each sample was calculated by reference to a standard curve (1, 2, 5, 10, and 20 ng) of genomic DNA and expressed as nanograms. To verify the presence or absence of Taq inhibitors, amplification was performed on all samples with a mix containing 2 µl of DNA extracted from stool and 25 ag of a plasmid with a control sequence. All determinations were performed in duplicate and repeated in about 20% of samples in which the variation was >20%.

Statistical Analysis

FL-DNA concentrations were considered as a continuous variable. The most accurate cutoff values to discriminate between healthy donors and patients were calculated using the receiver operating characteristic (ROC) curve. In the ROC curves, sensitivity (true positive rate) was plotted against 1-specificity (false-positive rate) for all classification points.

Sensitivity, specificity, and relative 95% confidence interval (95% CI) were calculated for the most discriminant cutoff values.

Results

DNA amplification levels were evaluated by the semiquantitative L-DNA method on stool samples from a series of 56 patients with primary colorectal cancer and 38 healthy individuals. DNA levels were expressed as the number of high amplifications (i.e., as a discrete variable). A different distribution was observed in affected and unaffected individuals (Table 1). In particular, 79% (30 of 38) of healthy donors showed no amplification, whereas at least one amplification was observed in about 70% (39 of 56) of patients. The analysis of the accuracy of this approach using different cutoffs (Table 1) showed a very high specificity ranging from 79% to 100%. Conversely, sensitivity was very poor and did not exceed 70% at any cutoff value.

Table 1.

Sensitivity and Specificity of L-DNA Analysis.

High Amplifications	Healthy Donors	Patients	Sensitivity (%)	Specificity (%)
At least 1	8/38	39/56	70	79
≥2	1/38	28/56	50	97
≥3	0/38	20/56	36	100

Amplification levels of fecal DNA were analyzed by the quantitative FL-DNA method in the overall series and expressed as a continuous variable (Figure 1). Only one stool sample from patients and three from healthy donors were not evaluable due to the presence of Taq inhibitors.

Figure 1.

p53 analysis for FL-DNA quantification. (A) Amplification of scalar concentrations of genomic DNA using the same primers and conditions of stool sample determination. (B) The area values under the electropherogram peaks are plotted in a calibration curve. (C) p53 electropherograms of six stool samples. The amount of amplified DNA from individual samples is quantified on the basis of the calibration curve.

Fluorescence signals ranged from 0 to 283 ng (median, 47 ng) in patient stool and from 0 to 87 ng (median, 4 ng) in healthy donor stool. No differences in median values were observed with respect to age of patients and the size, site, and stage of tumor.

When the results from the two approaches were compared, a direct relation was observed, but with a wide variability of FL-DNA levels within the subgroups defined according to the number of L-DNA high amplifications. Moreover, fluorescence by FL-DNA method was detected in 33 of 47 individuals who did not show any high amplification by L-DNA assay. These results are clearly indicative of a higher sensitivity of the fluorescence method than of the conventional approach.

The ROC curve analysis of FL-DNA levels (Figure 2) shows a good diagnostic accuracy of this approach. In particular, very high specificity ranging from 83% to 95% and high sensitivity ranging from 82% to 72% were observed for the most discriminant cutoffs of 15, 20, 25, and 30 ng of DNA (Table 2). When the cutoff of 25 ng, which provides the best overall accuracy, was analyzed in relation to the different tumor characteristics, sensitivity remained high in patients with small tumors (70%) compared to large tumors (82%) and was similar for the different Dukes stage tumors (Table 3). More importantly, a similar sensitivity was observed in detecting tumors localized in ascending and descending colon tracts.

Figure 2.

ROC curve of FL-DNA analysis for the overall series of stool samples from patients and healthy donors.

Table 2.

Sensitivity and Specificity of FL-DNA Analysis.

DNA Levels Cutoff (ng)	Healthy Donors		Patients		Sensitivity (%)	95% CI	Specificity (%)	95% CI
	Positive	Negative	Positive	Negative
15	10	49	70	15	82	(74–90)	83	(73–93)
20	7	52	70	15	82	(74–90)	88	(80–96)
25	4	55	65	20	76	(67–85)	93	(86–100)
30	3	56	61	24	72	(62–82)	95	(89–100)

Table 3.

Sensitivity^* as a Function of Different Characteristics in Colorectal Cancer.

Category	Number of Patients	Positive	Negative	Sensitivity (%)
Size (cm)
0.1–4.0	40	28	12	70
≥4.1	38	31	7	82
Dukes stage
A	8	7	1	88
B	29	25	4	86
C	37	25	12	68
D	9	8	1	89
Location
Ascending	18	13	5	72
Transverse	2	2	0	100
Descending	30	22	8	73
Rectum	35	28	7	80

^*Cutoff value is 25 ng.

Discussion

The possibility of performing population-based screening for colorectal cancer as well as for all tumor types is dependent on several factors such as complexity, time and cost, accessibility, and acceptability of screening methods.

Although several screening methods are currently available and have proven effective in reducing colorectal cancer mortality [22–27], a large-scale screening program comparable to those used for breast or prostate cancer does not exist. There are many reasons for this—the main one being the uncertainty of the best strategy to adopt [fecal occult blood test (FOBT), FOBT plus sigmoidoscopy, and so on]. Moreover, the real cost benefits of each method, considering the large number of endoscopic or radiologic procedures required for large-scale colorectal cancer screening, have not been determined [28–31]. Another important reason is that, for certain individuals, some of these techniques are not easily accepted, further reducing the compliance of the screening program itself [32].

A diagnostic approach that is less invasive, more accurate, and optimized in terms of time and cost is undoubtedly warranted. An important prospect is the analysis of molecular alterations detectable in human DNA extracted from stool. Many authors have investigated this area by analyzing a single molecular target or a combination of different molecular targets [6–21]. An interesting target that has recently been evaluated is the level of DNA amplification (L-DNA), which appears to be related to the presence and number of tumor cells in stool specimens. Only three studies to date have evaluated this marker in combination with some specific gene alterations using a semiquantitative method of analysis [19–21]. Their results show a good specificity and a relatively low sensitivity of this approach—the latter possibly due to the lack of an objective and quantitative evaluation, which compromises accurate discrimination between affected and nonaffected individuals, or to the insufficient sensitivity of the method.

In an attempt to improve the diagnostic accuracy of DNA amplification in exfoliated cells from stool, we set up and used an approach based on the evaluation of DNA amplification by a fluorescence method (FL-DNA). The results showed that this approach has a sensitivity comparable, if not superior, to that of multiple gene analysis, but is less expensive and less time-consuming. This sensitivity also made it possible to detect small, low-grade, and early-stage tumors. Moreover, unlike the fecal analysis of BAT26 instability or K-ras alterations, the determination of DNA amplification is able to detect tumors in all colon sites [16,33].

These unique features make this molecular marker an interesting tool for colorectal tumor diagnosis as it is characterized by all the benefits of other molecular analyses, such as noninvasiveness, simplicity, high compliance, reasonable costs, and time-efficient procedures.

Furthermore, this method could be improved and simplified by using alternative quantification systems such as chemoluminescence, spectrofluorimetry, real-time PCR, and so on, with the aim of developing kits that can be more easily utilized in all laboratories.

A more exhaustive study, including adenomas and benign polyps, is needed to verify the real sensitivity and specificity of this method. However, these original preliminary results would seem to indicate the validity of this test and its potential usefulness in screening programs or in monitoring members of families at risk for colorectal cancer.

Abbreviations

L-DNA

long DNA

FL-DNA

fluorescence long DNA

APC

adenomatous polyposis coli

ROC

receiver operating characteristic

FOBT

fecal occult blood test