Variables Associated with Detection of Methylated BCAT1 or IKZF1 in Blood from Patients Without Colonoscopically Evident Colorectal Cancer

Colorectal cancer is a preventable cancer, and yet it is the second leading cause of death from cancer in the developed world (1).

Screening for colorectal cancer is an important prevention strategy that has been proven to reduce mortality rates worldwide (2). The majority of global colorectal cancer screening programs use fecal immunochemical testing (FIT) to identify those in need of a diagnostic colonoscopy, but low participation rates are commonly observed (3). In Australia, participation rates in the national organized program with FIT are suboptimal at 42%, with similar screening uptake rates reported in trials in the United States (4, 5). Behavioral barriers and medical contraindications contribute to poor participation (6, 7). The barriers for screening with fecal tests might be overcome through screening using blood-based biomarkers for colorectal cancer.

Circulating cell-free DNA (cfDNA) is shed from both healthy and cancer cells as a result of cell apoptosis and necrosis, with levels of cfDNA shown to be higher in patients with late-stage cancer (8). Analysis of cfDNA for tumor-specific mutations or methylation patterns, that is, circulating tumor (ctDNA), is being evaluated for its utility in cancer detection and management. Testing for ctDNA has shown promising results for colorectal cancer detection, surveillance, and monitoring of treatment efficacy (9, 10).

The genes BCAT1 (branched‐chain amino acid transaminase 1) and IKZF1 (Ikaros family zinc‐finger 1 protein) are both methylated with high frequency in colorectal neoplastic tissues (11). Detection of methylated BCAT1 and/or IKZF1 DNA in blood was shown to be 62% sensitive and 92% specific for colorectal cancer, which was not significantly different to the performance of FIT (12). Likewise, these biomarkers are also promising for colorectal cancer recurrence surveillance (13, 14).

Methylation biomarkers such as BCAT1 and IKZF1 may also have a potential use for risk-based triaging of patients with symptoms for colonoscopy and therefore address an unmet need for many countries to manage the colonoscopy workload and focus it onto those more likely to have cancer (15). Whether used for this reason or for screening, it is important to identify confounding variables (those other than neoplasia) that are associated with detection of these markers in blood across a range of clinical settings.

The aim of this study was to identify variables that are associated with detection of methylated BCAT1 and IKZF1 DNA in blood from patients without colonoscopically evident colorectal cancer (a “false-positive” ctDNA result) and to assess the biomarker profiles and methylated DNA levels in patients without colorectal cancer compared with cases with confirmed colorectal cancer.

The study population comprised patients undergoing colonoscopy for any indication and who provided a blood sample for methylated BCAT1 and IKZF1 DNA testing prior to colonoscopy. The first part of this study was to determine whether any demographic and clinical variables were associated with detection of methylated BCAT1 or IKZF1 DNA in 1,479 patients found not to have colorectal cancer at colonoscopy. The second part of the study was a case–control study where cases (positive for either biomarker, n = 126) were compared with an age- and sex-matched group of controls (negative for both biomarkers, n = 252) for variables identified in the first part of the study, as well as examine whether prior or subsequent medical conditions that occurred within 12 months of blood testing were associated with detection of methylated BCAT1 and IKZF1 in blood. Finally, the levels and profiles of methylated DNA were compared in patients with and without colorectal cancer.

The study was approved by the Southern Adelaide Clinical Human Research Ethics Committee (ethics No. 134.045) and written informed consent was obtained from all study participants prior to blood collection.

Patients underwent colonoscopy in one of the Southern Adelaide Local Health Network public hospitals (Repatriation General Hospital, Daw Park and Flinders Medical Centre, Bedford Park, South Australia) between September 2011 and June 2014. The indications for colonoscopy included positive FIT, symptoms, anemia, and surveillance of moderate-high risk patients (personal history of adenoma/colorectal cancer or first-degree relative with colorectal cancer). There were 1,593 occasions where venous blood was collected from consenting patients on the day of but prior to colonoscopy (Fig. 1). Plasma was isolated and stored at −80°C until the cfDNA was extracted and analyzed for methylated BCAT1 and IKZF1 as described below. Patient demographics (including socioeconomic status), current comorbidities, and medications were collected at the time of colonoscopy. Findings at colonoscopy (confirmed by histopathology examination by the institutional pathologists as necessary) were recorded for all patients, with the main outcomes categorized as colorectal cancer, advanced adenoma (including adenomas with any of the following features: size ≥10 mm, high-grade dysplasia, or villous component), non-advanced adenoma (including tubular adenomas <10 mm with low-grade dysplasia), benign pathology (including hemorrhoids, diverticular disease, angiodysplasia, hyperplastic polyps), and no evidence of colorectal disease.

The biomarker case–control study was performed with a 1:2 ratio. Cases were all patients without colonoscopically evident colorectal cancer and who were positive for methylated BCAT1 and/or IKZF1 DNA in blood (n = 126). The controls (n = 252) were sampled (using a random number generator) from the group of patients without colonoscopically evident colorectal cancer and in whom neither biomarker was detected (n = 1,353), but matched within age decade and sex.

Hospital medical records of patients in the biomarker case and control groups were reviewed out to a minimum of 12 months pre- and post-blood collection to determine medical conditions, especially neoplastic pathologies that might be associated with detection of methylated BCAT1 and/or IKZF1 DNA in circulation. Information on demographics—including country of birth (1), medication use, tobacco smoking (never, current at time of colonoscopy, or previous), and alcohol usage [three categories: nil, light-moderate (defined as social/occasional or two to four standard drink/day), or heavy (defined as history of alcohol abuse, >5 standard drink/day, or >10 standard drink/week); ref. 16] was recorded.

DNA extracted from plasma samples was bisulphite-converted then assayed in triplicate with a multiplexed real-time PCR assay as described previously (12). A plasma sample was deemed positive if at least one PCR replicate was positive for methylated BCAT1 and/or IKZF1 DNA.

Each PCR plate included a 7-point serial dilution of fully methylated DNA, which was used for quantitation of methylated DNA and circulating DNA. When consideration was given to levels of methylated DNA, this was expressed as the mass of methylated BCAT1 and IKZF1 DNA measured in a plasma sample (pg/sample). The qPCR assay also amplified a DNA region in ACTB (DNA control parameter) and the mean estimated mass measured across the triplicate ACTB PCR reactions was used for determination of cfDNA levels (ng/mL).

Descriptive analyses were conducted to determine medians, interquartile ranges (IQRs), or frequencies as appropriate. Univariable logistic regression analyses using the methylation test result as the outcome event were performed against the various demographic and clinical variables compiled for each patient to identify the variables associated with the methylation test result. Variables found to be significantly associated with detectable methylated DNA in circulation, or those that were considered clinically relevant, were included in a multivariable logistic regression analysis model. For the biomarker case–control study, the analysis was performed with additional demographic and clinical variables included such as: smoking status, body mass index (BMI, kg/m²), alcohol intake, country of birth, medications, comorbidities as well as clinical findings such as previous cancers, new cancer diagnosis, and death within 12 months of blood testing. Socioeconomic status (index of relative socioeconomic disadvantage score) was determined from the participant's residential postcode through linkage with the Australian 2016 census data (17) and expressed as above or below the median score for Australia, with below considered disadvantaged areas. Any missing data were omitted from analyses. Results were reported as odds ratios (ORs) and 95% confidence intervals (CIs). Mann–Whitney rank-sum test and chi-square analyses were used for subpopulation comparisons. All statistical tests were two-sided and P values less than 0.05 were considered statistically significant. All analyses were performed using Stata version 13.0 (StataCorp.).

Of the 1,593 patients, 114 (7.2%) had colorectal cancer findings at colonoscopy. Of the 1,479 patients without colonoscopically evident colorectal cancer, 126 (8.5%) had a blood sample positive for methylated BCAT1 and/or IKZF1 (median age = 68.1 years; IQR = 61.0–74.8; 50% male; Fig. 1). Table 1 lists the clinical and demographic variables compiled for the 1,479 patients without colorectal cancer evident at colonoscopy (median age = 63.7 years; IQR = 55.3–71.3 years; 47.9% male).

Univariable and multivariable logistic regression analyses were performed to determine if any of the examined variables were associated with detection of methylated BCAT1/IKZF1 in the 1,479 patients without colonoscopically evident colorectal cancer. Univariable analysis showed that all age groups ≥60 years, hypertension, and cfDNA levels were significantly associated with detection of methylated BCAT1 and/or IKZF1 DNA in circulation (Table 2). Following multivariable analysis including age, sex, cfDNA, hypertension, and findings at colonoscopy within the model, only the age group ≥70 years (OR, 4.31; 95% CI, 1.50–12.41) and cfDNA levels (OR, 1.16; 95% CI, 1.10–1.22) were independently associated with detection of methylated BCAT1 and/or IKZF1 DNA (Table 2). As aging is known to positively correlate with higher cfDNA levels (18), multivariable analysis was followed by a statistical test for collinearity. There was no linear relationship between cfDNA levels and age, and their association with detection of the biomarkers in blood from patients without colonoscopically evident colorectal cancer.

In the patients without colonoscopically evident colorectal cancer, there were no significant differences in the disposition of demographics and clinical variables at the time of the colonoscopy between age- and sex-matched biomarker positive cases (n = 126) and the randomly selected biomarker negative controls (n = 252; Supplementary Table S1).

In the assessment of noncolorectal neoplastic states that were either present at the time of blood collection or were diagnosed within the subsequent 12 months, there were 17 individuals with confirmed cancer (11/126 cases and 6/252 controls, P = 0.005). In the 11 cases with a positive blood test for methylated BCAT1 and/or IKZF1 and absence of colonoscopically evident colorectal cancer, the cancers included two prostate cancers (one with bony metastases), one with metastases from colorectal cancer, breast cancer with metastases, pancreatic cancer, lung adenocarcinoma with metastases, B-cell chronic lymphocytic leukemia, nasopharyngeal tumor, SCC of the tonsils with metastases, and a neuroendocrine tumor in the small intestine. In the controls negative for both biomarkers, the cancers diagnosed included metastatic ovarian cancer, three lung cancers (one mesothelioma, one small cell carcinoma, and one adenocarcinoma), a urothelial papillary bladder cancer, and metastatic cancer with the pancreas as the suspected primary. Metastatic cancer diagnosed within 12 months of blood testing, as well as cfDNA levels, were significantly associated with the detection of methylated BCAT1 and/or IKZF1 (Table 3). The use of analgesics or metformin, or death within 12 months of blood testing showed trends but failed to reach significance (P = 0.076–0.095). Country of birth, smoking, alcohol intake, and BMI were not associated with the detection of the methylated biomarkers (P > 0.05).

Of the whole cohort (n = 1,593), 196 (12.3%) were positive for at least one of the methylated biomarkers. Of the 196 positive results, 70 patients were diagnosed with colorectal cancer (a positive predictive value of 35.7%). True positive rate (sensitivity for colorectal cancer) was 61.4% (70/114) whereas the false-positive rate was 8.5% (126/1,479). The assayed levels of methylated BCAT1 and/or IKZF1 in circulation in patients with a true positive (with colorectal cancer) was significantly higher than that found in those without colorectal cancer at colonoscopy [median = 92.9 pg/blood sample (IQR = 18.8–825.3) vs. 5.9 pg/blood sample (IQR = 0.3–16.3), respectively, P < 0.001].

The methylated biomarker profiles in all patients who had a positive test result were also assessed. In patients diagnosed with colorectal cancer at colonoscopy, 45.7% (32/70) were positive for both methylated biomarkers. In the 126 cases without evident colorectal cancer, only 7.9% (10/126) were found to have both methylated biomarkers present (P < 0.001, Table 4). Most of the biomarker positive cases without colorectal cancer were positive only for methylated BCAT1 (65.9%, P < 0.001, Table 4). The total levels of methylated DNA did not correlated with the cfDNA concentration in the 126 patients without evident colorectal cancer (r = −0.013; P = 0.888).

There were four patients where the level of methylated DNA far exceeded the median level of methylated DNA measured in patients diagnosed with colorectal cancer at the time of colonoscopy, and where either methylated IKZF1 or both biomarkers were detected in the blood sample (Fig. 2). All four patients had metastatic adenocarcinoma within 12 months either side of the blood test (Supplementary Table S2).

Limiting the cases to a screening population (i.e., the only indication for colonoscopy being a positive FIT with age between 50–75 years), of the whole cohort there were 265 cases, and 29 (10.9%) had detectable circulating methylated BCAT1 and/or IKZF1. Of these, 22 (75.9%) did not have colorectal cancer diagnosed at colonoscopy. This group had a median methylated DNA level of 2.8 pg/sample (IQR = 0.4–8.3). Two patients had cancer diagnosed close to the time of colonoscopy, including breast cancer with suspicion for metastatic spread in the bones (methylated DNA level = 172.7 pg/sample), and a nasopharyngeal tumor with lymphatic involvement (methylated DNA level = 77.6 pg/sample).

This study identified variables that are associated with the detection of circulating DNA methylated in BCAT1 or IKZF1 in patients undergoing colonoscopy for various indications and in whom colorectal cancer was not evident at the procedure. Analysis of data from the full cohort of cases without colonoscopically evident colorectal cancer demonstrated an association of these methylated biomarkers with age (≥70 years) and cfDNA levels. The age-sex biomarker case-control study, which explored additional variables, found an association with a diagnosis of metastatic adenocarcinoma within 12 months of colonoscopy as well as confirming the association with cfDNA levels.

Prior to the implementation of any new biomarker for colorectal cancer detection, whether it be applied to testing of symptomatic cases to prioritize who undergoes colonoscopy, to surveillance of subjects at increased risk, or to population screening, it is important to understand patient and other variables that might be associated with a positive biomarker result in the absence of colonoscopically evident colorectal cancer. We found that in a population of 1,593 patients that underwent colonoscopy, there were 196 patients who were positive for methylated BCAT1 and/or IKZF1 DNA. Of these, 70 (36%) results were considered true positives, with colorectal cancer diagnosed at colonoscopy. In those without evident colorectal cancer, on further examination of medical records, there were 11 patients (5.6%) that had another cancer type present at the time of colonoscopy, or who developed another cancer or colorectal cancer in the diagnosed subsequent 12 months. Many of these cancers were adenocarcinomas with distant metastases, with logistic regression showing a strong association between metastatic cancer and the presence of at least one of the methylated blood biomarkers (OR, 14.6). Methylation of BCAT1 and IKZF1 in cancers other than colorectal cancer have been noted previously (19–22).

This left 115 cases (58.7%) considered to have “false-positive” blood test results. Although variables such as smoking, hypertension, and medications have been reported to alter other methylation biomarkers for cancer detection (23, 24), these factors were not found to be associated with detection of methylated BCAT1 and/or IKZF1 DNA. Another methylated biomarker used for colorectal cancer detection and monitoring (SEPT9) has been previously found to be elevated in certain disease states such as diabetes and arteriosclerosis (24). Our study found no significant association between the detection of the circulating DNA methylated in BCAT1 and/or IKZF1 and common comorbidities. However, age ≥70 years was significantly associated with a positive test result in patients without colonoscopically evident colorectal cancer. A similar association between age and methylated biomarkers has also been observed for methylated SEPT9 (25, 26). This may be related to the reported observations of a relationship between aging and accumulation of genome-wide DNA methylation (27–29).

The important clinical question that arises from any new biomarker for colorectal cancer, whether it is fecal- or blood-based, is whether further examinations are needed in the positive cases without evident colorectal cancer. With FIT, results that are considered to be a false positive do not typically have further investigations unless patients also have iron-deficiency, however, a more cautious approach might be warranted for blood-based biomarkers as the source of methylated DNA may be from cancer not present in the colorectum. In our cohort of patients undergoing colonoscopy for a range of indications, 5.6% of the “false positives” were associated with another synchronous or subsequently-diagnosed cancer. Four of these patients had a methylation level above the median level measured in patients with colorectal cancer. All four patients were found to have metastatic adenocarcinoma present. It was also observed that three were positive for both markers and one was positive for methylated IKZF1. These support that the biomarker profile and levels may be used to determine which cases need further follow up. Because more than two-thirds of the biomarker positive cases without evident colorectal cancer were positive only for methylated BCAT1, then a sole positive BCAT1, especially if levels are low, and in the older patient, may not warrant further detailed clinical investigation. However, if the total methylation levels are in the range typically observed in patients with colorectal cancer, then further examinations should be considered.

The observation that epigenetic biomarkers are present for cancers other than colorectal cancer is not new. This has been observed with methylated SEPT9 in blood (30–32). In this study, the four cases with high levels of methylated BCAT1/IKZF1 included adenocarcinomas with metastatic spread. Patients were undergoing colonoscopy for a range of clinical reasons, sometimes in a search for the primary in patients suspected to have cancer somewhere. This situation is less likely to arise in the screening context, with our subgroup analysis showing that one patient had high methylated DNA levels and was diagnosed with breast cancer. The majority of findings of other cancers is likely to occur in patients undergoing testing associated with symptoms or within a high-risk surveillance program. These results suggest that detection of circulating methylated BCAT1 or IKZF1 in patients in the absence of colorectal cancer at colonoscopy should not be ignored and a search for cancer elsewhere is warranted if methylation levels are high. However, at this point in time, it is difficult to determine a threshold value for each biomarker to trigger additional investigations for patients with a positive test result but with no colonoscopically evident colorectal cancer. Larger prospective studies are warranted to identify this.

Our study revealed a significant association between cfDNA levels and the presence of at least one of the methylated biomarkers, but found no correlation between the levels of cfDNA and amount of DNA methylated. Some studies have shown cfDNA concentrations to be higher in patients with cancer (33), however our results showed that the four cases with high methylation levels did not have particularly high cfDNA levels.

This study focused on patients covering a wide spectrum of pathologies encountered in the colorectum. The implications of finding an association with age ≥70 years and cfDNA yield are relevant to use of such a blood test for testing of symptomatic cases to prioritize who undergoes colonoscopy, to surveillance of subjects at increased risk, or to population screening. However, a limitation is that as our study population included a range of indications for colonoscopy, we cannot make conclusions regarding the proportion of false-positive results that would be expected to be encountered in a screening population. Instead, we have provided results for what was seen in our cohort undergoing colonoscopy for a screening FIT. A further limitation is that because the design was retrospective and identification of additional variables was based on hospital medical records, there might be some missed associations. The variables identified to be associated with detection of methylated BCAT1 and/or IKZF1 in patients without evident colorectal cancer at colonoscopy now need prospective investigation in specific contexts to determine what clinical action might be appropriate. It is important to conduct prospective studies, particularly in a screening cohort, to determine how many false positives will be encountered, how many need further investigations for other cancers (new or recurrent disease), and whether these investigations will allow treatment and improved survival. These additional studies will allow the evaluation of the cost effectiveness of the methylated BCAT1/IKZF1 test, which is needed before introducing the test into population screening or clinical use with symptomatic and surveillance cases.

In conclusion, this retrospective study has identified aging and cfDNA levels to be associated with detection of circulating DNA methylated in BCAT1 and IKZF1 in the absence of colonoscopically evident colorectal cancer. Further, very high levels of circulating DNA methylated in IKZF1 and/or BCAT1 in cases without colonoscopically evident colorectal cancer might indicate presence of adenocarcinoma elsewhere. Isolated detection of methylated BCAT1 was more common in the absence of colorectal cancer than if colorectal cancer was present, demonstrating that this biomarker, rather than IKZF1, is the major contributor to loss of specificity for colorectal cancer. Further research in the contexts of colorectal cancer population screening, surveillance of those at increased risk for colorectal cancer and triaging symptomatic cases for colonoscopy, is needed to fully understand the impact of these findings.

G.P. Young reported personal fees from Clinical Genomics Inc., other from Clinical Genomics Inc. during the conduct of the study, and other from Eiken Chemical Company outside the submitted work. E.L. Symonds reported grants and nonfinancial support from Clinical Genomics during the conduct of the study; grants and nonfinancial support from Eiken Chemical Company outside the submitted work; and E.L. Symonds had a patent for WO2020150311A1 licensed to Flinders University. No disclosures were reported by the other authors.

H. Saluja: Data curation, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. G.P. Young: Conceptualization, resources, supervision, funding acquisition, investigation, methodology, writing–review and editing. F. Kholmurodova: Formal analysis, funding acquisition, writing–review and editing. E.L. Symonds: Conceptualization, data curation, formal analysis, supervision, writing–original draft, project administration, writing–review and editing.

The authors acknowledge Susan Byrne, Kathryn Cornthwaite, and Lorraine Sheehan-Hennessy of Cancer Research, Flinders Health and Medical Research Institute, College of Medicine and Public Health, Flinders University, Bedford Park, South Australia, and Susanne Pedersen of Clinical Genomics, North Ryde, New South Wales, Australia, for their contribution to this project. F. Kholmurodova was supported by a grant funded by the financial support of Cancer Council SA's Beat Cancer Project on behalf of its donors and the State Government of South Australia through the Department of Health together with the support of the Flinders Medical Centre Foundation, its donors and partners. G.P. Young received grants from the National Health and Medical Research Council of Australia (Grant Nos. APP1006242 and APP1017083) and from Clinical Genomics Pty Ltd. (North Ryde, NSW, Australia).

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.