Low Incidence of Aerodigestive Cancers in Patients With Negative Results From Colonoscopies, Regardless of Findings From Multitarget Stool DNA Tests

Abstract

BACKGROUND & AIMS:

We aimed to compare the incidence of aerodigestive cancers in persons with negative results from colonoscopies and positive vs negative results from multitarget stool DNA tests for colorectal cancer and vs expected incidence.

METHODS:

We performed a retrospective cohort study of 1216 subjects with comprehensive patient records and/or cancer registry data from 3 medical centers in North America. Subjects had no neoplasia or only nonadvanced adenomas, based on screening colonoscopy, and either negative results (concordant with colonoscopy, n = 1011) or positive results (discordant colonoscopy, n = 205) from the multitarget stool DNA test. Outcomes included aerodigestive cancers in discordant vs concordant groups and comparison of observed aerodigestive cancer incidence between the groups and compared with expected incidence for the population, based on the Surveillance, Epidemiology, and End Results (SEER) data.

RESULTS:

Median follow-up times were comparable between subjects in the discordant (5.3 y; interquartile range, 3.5–5.8 y) and concordant (5.4 y; interquartile range, 3.7–5.8 y) groups. Aerodigestive cancers developed in 5 subjects in the discordant group vs 11 subjects in the concordant group (crude risk ratio, 2.3; 95% CI, 0.8–6.6; adjusted risk ratio, 2.2; 95% CI, 0.8–6.2; P = .151). The incidence of aerodigestive cancer was lower in the concordant group than the expected incidence based on SEER data (risk ratio, 0.4; 95% CI, 0.2–0.6; P = .0008). The incidence of aerodigestive cancer was not significantly greater in the population in the discordant group than the expected incidence based on SEER data (risk ratio, 0.8; 95% CI, 0.3–1.9; P = .599).

CONCLUSIONS:

In a retrospective study with a median follow-up time of 5.4 years, incident aerodigestive cancers were uncommon among subjects with negative findings from colonoscopies, regardless of discordant or concordant results from multitarget stool DNA tests. Patients with negative results from high-quality colonoscopies therefore should not undergo further testing.

Despite decreases in incidence and mortality over the past 10 to 15 years, colorectal cancer (CRC) remains the second most common and lethal cancer in US men and women combined.¹ Screening for CRC is recommended by multiple evidence-based guidelines, all of which support more than 1 modality and strategy.^2-5 Although screening for CRC reduces both disease incidence and mortality, adherence to screening among persons age 50 to 75 remains static at 65% in the United States.⁶ The multitarget stool DNA (mt-sDNA) test was developed to provide both a noninvasive screening test with high sensitivity for average-risk CRC screening and a process to help address the need for increased adherence to screening through patient navigation. Since Food and Drug Administration approval in 2014, more than 2.2 million individuals have been screened with mt-sDNA (Exact Sciences Laboratories data, Madison, WI).

In the pivotal study for this new technology (NCT01397747),⁷ individuals with a positive mt-sDNA test who were found to have advanced colorectal neoplasia on a subsequent diagnostic colonoscopy were considered to be true positives, whereas those with only nonadvanced adenoma or no colorectal neoplasia were considered false positives. In screening studies, 7% to 13%^7,8 of mt-sDNA–screened patients had a positive test with a negative colonoscopy, which includes no polyps of any kind, hyperplastic polyps, and nonadvanced adenomas. A discordant result (mt-sDNA positive/colonoscopy negative) raises the question of whether additional clinical evaluation for other aerodigestive cancers (ADCs) is warranted.

A previous meta-analysis of false-positive stool DNA methylation alone did not show increased ADC risk; however, the studies included were heterogeneous with respect to test panel and patient risk factors, and full mt-sDNA panel results were not available for all patients.⁹ The objective of this study was to determine whether asymptomatic patients with false-positive mt-sDNA results are at increased risk for ADCs (lung and digestive tract). To accomplish this objective, we compared this group with those with negative colonoscopy and negative mt-sDNA results and compared both groups with age- and sex-adjusted Surveillance, Epidemiology, and End Results (SEER) data.¹⁰

Methods

With institution-specific review board approval, 5 of 90 pivotal study sites participated in this retrospective cohort study. This study follows the Strengthening the Reporting of Observational Studies in Epidemiology guidelines for reporting of cohort studies.¹¹ The participating sites were the University of Calgary, Alberta Canada (Calgary); Mayo Clinic (Rochester MN, Jacksonville FL, Phoenix AZ [Mayo]); and the Wisconsin Center for Advanced Research, Milwaukee, WI (WCAR). These sites were selected on the basis of full access to comprehensive patient records and/or cancer registry data.

Site investigators were provided with lists of all site-specific pivotal study participants with negative colonoscopic findings and positive (discordant) or negative (concordant) mt-sDNA tests from the pivotal study’s central database. The pivotal study colonoscopy was performed within 90 days of informed consent. Participants with negative colonoscopies included those with no colorectal neoplasia or with only nonadvanced adenomas (conventional adenoma or sessile serrated polyp <1.0 cm in size, without advanced histologic features) or hyperplastic polyps as the most significant findings. Participant characteristics, including age, sex, race, ethnicity, colonoscopy findings, and follow-up interval, were compared across study sites to determine baseline comparability with all pivotal study participants. Sites had full access to comprehensive patient records and/or cancer registry data.

Follow-up data were obtained on all current study participants to determine whether a subsequent aerodigestive cancer had been identified. Although all 5 sites had electronic medical records systems, sites varied in the availability of additional sources and methods to enhance the completeness and accuracy of follow-up evaluation. Site follow-up times varied; site-specific follow-up information was obtained as follows.

The 3 Mayo Clinic sites performed an Accurint database (LexisNexis, New York, NY) search to determine participant vital status. Participants or their next of kin were invited by mail for a structured telephone interview to document any new cancer diagnoses. A medical records review was performed on those who declined an interview and those who did not respond; chart review also was performed on a random 10% subset of those interviewed to measure the type of and frequency of discrepant information, as previously described.⁹ Patient follow-up evaluation was censored at the time of death, last documented health care encounter, or finalization of the study database on December 31, 2015.

The WCAR site has an active referral and primary care practice, low patient turnover, and extensive medical records. WCAR contacted patients by telephone who had fewer than 3 years of medical record follow-up evaluation to optimize data collection. Medical record reviews and interviews were completed by September of 2016.

The University of Calgary site is a regional (Alberta, Canada) colonoscopy referral center. All study participants were linked to the Alberta Cancer Registry using their unique Personal Health Number to identify new diagnoses of CRC and death from any type of cancer. Linkage was completed in August 2015 and a final second review was completed in January 2018. The Alberta Cancer Registry captures all invasive cancers, except nonmelanoma skin cancer, diagnosed in the province of Alberta.

For all sites, subsequent cancer diagnoses were confirmed by review of endoscopic, radiographic, and histopathologic reports. If database, registry, or medical records indicated a patient death in an individual not diagnosed with ADC during the study period, the underlying cause of death was documented to properly exclude ADC.

The follow-up interval was the shorter of the time between the informed consent for the pivotal study and either the diagnosis of ADC or medical records/cancer registry review. Participants were considered lost to follow-up evaluation if there were no available medical records beyond 6 months of the pivotal study colonoscopy unless these patients were reached by other means.

The group with concordant results (true negatives) served as the reference group. The incidence of ADC was defined as the number of events per year of follow-up evaluation. Estimated risk ratios adjusted for the length of follow-up evaluation were calculated for each group with corresponding 95% CIs. In addition to this direct comparison, we compared the observed incidence of aerodigestive cancer in the concordant and discordant groups with the expected incidence of age- and sex-adjusted SEER data.¹⁰ Continuous data were summarized as a median with corresponding 25th and 75th percentiles (interquartile range [IQR]), and comparisons between patient subgroups were made with the Wilcoxon rank sum test. Categoric data were summarized as a percentage of group totals, and comparisons between subgroups were based on the Fisher exact test for proportions.

Participants with complete follow-up data were included in the primary analysis (complete data approach) and participants lost to follow-up evaluation were not included in the primary analysis. To address potential bias resulting from participants lost to follow-up evaluation, 2 imputation analyses were performed. First, the follow-up time for each person lost to follow-up evaluation was imputed from a uniform distribution (range, 6 mo to 6 y) (imputation method 1). For the end point of ADC, subjects randomly were assigned the ADC outcome from a Bernoulli distribution with a probability of 0.03. This probability was based on the expected number of ADC events based on SEER data assuming all subjects lost to follow-up evaluation had 6 years of follow-up evaluation. For the second approach, follow-up time and ADC outcome were selected randomly as a pair from subjects with complete follow-up data (imputation method 2). The random selection of the outcome pairs was stratified by group status (ie, discordant and concordant strata). Each of the imputation methods was performed 10,000 times and the results were combined using Rubin’s¹² method.

Results

Of 1230 participants with a negative colonoscopy, 1216 (99%) had sufficient follow-up evaluation for inclusion (≥6 mo or a diagnosis of ADC). The study population represents 13.3% (1216 of 9166) of all pivotal study participants with negative colonoscopies and includes 205 (16.7%) discordant and 1011 (83.1%) concordant results. An additional 2 participants with discordant results and 12 participants with concordant results had no incident ADCs but had insufficient postpivotal study follow-up evaluation (0–6 mo) for inclusion in the study population (Figure 1).

Figure 1.

Study flow chart. mt-sDNA, multitarget stool DNA.

The study cohort of 1230 had demographic differences compared with all pivotal study participants with a negative colonoscopy (N = 7936). The median age was 67 years (IQR, 65–70 y) compared with 65 years (IQR, 55–70 y) (P < .001), 51% vs 44% were men (P < .001), and 92.9% vs 82.6% were white (P < .001), for this study and the pivotal study, respectively. However, demographics were similar between the discordant and concordant subgroups. Although statistically significantly different, patient median ages were similar at 68 years (IQR, 65–71 y) and 67 years (IQR, 65–70 y) for individuals with discordant and concordant results, respectively (P = .005). Participants with a positive mt-sDNA had nonadvanced adenomas on colonoscopy more frequently than participants with negative mt-sDNA tests (41.0% vs 30.5%; P = .002) (Table 1). The median follow-up time was 5.3 years (IQR, 3.5–5.8 y) in the discordant group and 5.4 years (IQR, 3.7–5.8 y) in the concordant group (P = .408); 93.2% (191 of 205) of discordant and 92.4% (934 of 1011) of concordant cases had at least 3 years of follow-up evaluation (Table 2).

Table 1.

Comparative Demographics and screening Colonoscopy Findings Among 1205 Participants With Negative Colonoscopies and Either Positive (Discordant) or Negative (Concordant) mt-sDNA Tests From 5 Clinical Sites Participating in the DeeP-C mt-sDNA Pivotal Study

Demographic	Discordant cases (N = 205)a	Concordant cases (N = 1011)a	P value
Age
Median (IQR), y	68 (65–71)	67 (65–70)	.005
50–64 y, n (%)	43 (21)	246 (24)	.303
65-85 y, n (%)	162 (79)	765 (76)
Sex
Male, n (%)	107 (52)	520 (51)	.842
Female, n (%)	98 (48)	491 (49)
Race
White, n (%)	194 (95)	937 (93)	.514
Black, n (%)	4 (2)	20 (2)
Other, n (5)	7 (3)	754 (5)
Ethnicity
Hispanic, n (%)	1 (0.5)	11 (1.1)	.428
Negative colonoscopy findingsb
NAA (NAA detection rate)b	84 (41)	308 (30)	<.001
No colorectal neoplasia	121 (59)	703 (70)

DeeP-C, detection of colorectal advanced adenomatous polyps and cancer; IQR, interquartile range; mt-sDNA, multitarget stool DNA; NAA, nonadvanced adenoma.

^aAn additional 2 discordant and 12 concordant participants were lost to follow-up evaluation within 6 months of informed consent (0–6.0 mo).

^bParticipants with NAA as the most significant findings are considered false positives for mt-sDNA screening, whereas colorectal cancer and advanced adenomas are true positives. Participants with negative screening colonoscopy findings have no colorectal neoplasia or have NAA or hyperplastic polyps as the most significant findings. NAAs are colorectal adenomas or sessile serrated polyps less than 1.0 cm in diameter, without high-grade dysplasia, or a 25% or greater villous component.

Table 2.

Follow-Up Observation Intervals of 1216 Individuals With Negative Colonoscopy Examinations and Either Positive or Negative mt-sDNA Tests, by DeeP-C Study Site

Site	Discordant cases	Concordant cases
WCAR
Cases, n	35	147
Median follow-up period, y (IQR)	3.8 (3.7–4.6)	4.1 (4.0–4.9)
Calgary
Cases, n	113	594
Median follow-up period, y (IQR)	5.8 (5.5–5.9)	5.7 (5.5–5.9)
Mayo (3 sites)
Cases, n	57	270
Median follow-up period, y (IQR)	3.2 (3.0–3.4)	3.3 (3.0–3.5)
Total
Cases, n	205	1011
Median follow-up period, y (IQR)	5.3 (3.5–5.8)	5.4 (3.7–5.8)

DeeP-C, detection of colorectal advanced adenomatous polyps and cancer; IQR, interquartile range; mt-sDNA, multitarget stool DNA; WCAR, Wisconsin Center for Advanced Research.

Within the discordant group, 5 (2.4%) ADCs (incidence rate, 0.5%/y; 95% CI, 0.2–1.2%/y) were diagnosed at 0.3 (pancreatic), 2.9 (low-grade parotid), 3.0 (colon), 2.9 (lung), and 3.6 (lung) years after pivotal study informed consent. The expected number of ADCs within the discordant group was 6, resulting in a risk ratio of 0.8 (95% CI, 0.3–1.9; P = .599) relative to the SEER population. Among the concordant group, 11 (1.1%) ADCs (incidence rate of 0.2%/y; 95% CI, 0.1%–0.4%/y) were diagnosed at 0.5 (pancreas), 3.7 (pancreas), 2.0 (head and neck), 3.2 (head and neck), 0.7 to 4.4 (lung; n = 6), and 4.6 (liver) years after informed consent (Table 3). The expected number of ADCs for the concordant group was 30, resulting in a risk ratio of 0.4 (95% CI, 0.2–0.7; P < .001) relative to the SEER population. The estimated risk ratio for the discordant group relative to the concordant group was 2.2 (95% CI, 0.8–6.2; P = .151) after adjusting for the age- and sex-specific rate of ADC based on SEER, which was similar to the age- and sex-adjusted risk ratio of 1.9 (95% CI, 0.7–5.7; P = .168) and to the unadjusted risk ratio of 2.3 (95% CI, 0.8–6.6; P = .123). Imputation of the missing follow-up data for 14 subjects (2 discordant and 12 concordant) using imputation methods 1 and 2 showed a general tendency for the complete data approach to overestimate the risk ratios of ADC in the discordant group relative to the concordant group for both the crude and population-adjusted estimates (Table 4).

Table 3.

Incident Aerodigestive Cancers Diagnosed in 1205 mt-sDNA DeeP-C Pivotal Study Participants With Negative Colonoscopies and Positive or Negative mt-sDNA Tests During the Follow-Up Period

mt-sDNA/ colonoscopy finding	DeeP-C colonoscopy finding	Diagnosis	Age at DeeP-C study informed consent	Sexa	Follow-up period, y	Interval of informed consent to diagnosis, y	Interval of informed consent to death, y	Cause of death
Positive/negative	No biopsy	Pancreatic cancer	70	Man	0.6	0.3	0.6	Pancreatic cancer
Positive/negative	NAA (2), 6–9 mm	Colorectal cancerb cecal	82	Man	3.0	3.0	Alive	NA
Positive/negative	No biopsy	Parotid gland, low-grade mucoepidermoid cancer	60	Woman	4.7	2.9	Alive	NA
Positive/negative	No biopsy	Lung, adenocarcinoma, NOS	68	Man	3.1	2.9	3.1	Leukemia
Positive/negative	NAA (3), 6–9 mm	Lung, squamous cell carcinoma, well differentiated	72	Man	5.2	3.6	5.2	Not coded
Negative/negative	No biopsy	Head and neck-nasopharynx	65	Man	4.6	3.2	4.6	Nasopharyngeal cancer
Negative/negative	No biopsy	Head and neck-tongue, squamous cell carcinoma	66	Man	5.7	2.0	Alive	NA
Negative/negative	No biopsy	Liver, hemangiosarcoma	70	Man	4.7	4.6	4.7	Not coded
Negative/negative	Hyperplastic polyp <10 mm	Lung cancer	70	Man	2.4	2.4	Alive	NA
Negative/negative	No biopsy	Lung, adenocarcinoma NOS	65	Man	0.8	0.7	0.8	Noncancer related
Negative/negative	No biopsy	Lung, adenocarcinoma NOS	65	Woman	3.6	2.6	3.6	Lung cancer
Negative/negative	No biopsy	Lung, adenocarcinoma NOS	73	Man	4.9	4.4	4.9	Not coded
Negative/negative	No biopsy	Lung, small cell carcinoma	73	Man	4.4	4.3	4.4	Lung cancer
Negative/negative	NAA (1), ≤5 mm	Lung, squamous cell carcinoma, NOS	72	Woman	3.2	2.7	3.2	Lung cancer
Negative/negative	Hyperplastic polyp <10 mm	Pancreatic cancer	71	Woman	0.5	0.5	0.5	Pancreatic cancer
Negative/negative	No biopsy	Pancreatic, malignant carcinoid	67	Woman	5.4	3.7	Alive	NA

DeeP-C, detection of colorectal advanced adenomatous polyps and cancer; mt-sDNA, multitarget stool DNA; NA, not applicable; NAA, nonadvanced adenoma; NOS, not otherwise specified.

^aAll patients with incident aerodigestive cancer were Caucasian and non-Hispanic.

^bColorectal cancer was detected on the 3-year surveillance examination indicated by the previous DeeP-C colonoscopy finding of 2 adenomas.

Table 4.

Sensitivity Analysis: Imputation of Missing Data

Relative risk estimates	Imputation method 1	Imputation method 2	Complete data
Crude estimate	1.96 (95% CI, 0.69–5.61)	1.99 (95% CI, 0.70–5.64)	2.29 (95% CI, 0.79–6.62)
Adjusted estimate	2.08 (95% CI, 0.73–5.89)	2.11 (95% CI, 0.74–6.0)	2.17 (95% CI, 0.75–6.24)

Among the 5 participants with discordant results and incident ADC, 2 were alive and 3 had died: 1 from the incident cancer, 1 from leukemia, and 1 was not coded. Among the 11 participants with concordant results and incident ADC, 3 were alive and 8 had died: 5 from the cancer, 1 not related to cancer, and 2 were not coded (Table 3). Among the subjects missing from follow-up evaluation, a final review (December 2018) of medical records and publicly available sources showed no evidence of death for the 2 missing discordant subjects but 2 of the 12 concordant subjects were found to have died 4.1 years and approximately 6 years after informed consent, although the cause of death was not noted.

Discussion

In this retrospective cohort study on a screening population, we found that the rates of subsequent ADCs were no different between patients with false-positive and true-negative mt-sDNA tests. The hazard rates between the discordant and concordant participant groups, adjusted for length of follow-up evaluation, were not significantly different (P = .151). Furthermore, when compared with SEER-derived expected rates of aerodigestive cancers, both subgroups had lower than expected rates of these cancers. These findings indicate that additional diagnostic investigation in mt-sDNA false-positive patients who have had high-quality colonoscopy does not appear warranted.

Colorectal cancer screening with mt-sDNA is a recent addition to clinical practice (August 2014) and is included in the US Preventive Services Task Force guidelines (2016),² referred to as fecal immunochemical test-DNA, as one of several equally recommended screening tests. mt-sDNA was designed with an a priori false-positive rate of approximately 10% of screened individuals with no colorectal neoplasia to optimize the detection of early stage CRC¹³ (American Joint Committee on Cancer stages I + II).¹⁴ This performance subsequently was evaluated in the mt-sDNA pivotal screening study, which found high sensitivity for CRC overall (92%), for CRC stages I + II (94%), and for high–grade dysplasia (69%). In the pivotal study, mt-sDNA false positives included those with positive mt-sDNA tests in individuals with nonadvanced adenomas or hyperplastic polyps as the most significant finding on colonoscopy, in addition to those with no evidence of colorectal epithelial neoplasia.

Our findings are consistent with previous studies on this topic. Ahlquist et al¹⁵ performed esophagogastroduodenoscopy and abdominopelvic computed tomography scans on false-positive patients with a negative colonoscopy after a positive prototype stool DNA test that detected specific DNA mutations in APC, TPp53, and KRAS, evidence of microsatellite instability (BAT-26), and an assessment of stool DNA strand length. This study was terminated by the local internal review board after 70 consecutive negative follow-up examinations. Within the mt-sDNA Food and Drug Administration approval process for use in CRC screening of individuals at average risk for CRC, case-control studies were performed that analyzed stool from patients with cancers including noncolorectal aerodigestive and other anatomic sites. These studies projected that screening with mt-sDNA would be associated with a very low rate of false-positive tests owing to non-CRC sources (1.2 cases/1000 false positives) with only liver, pancreatic, and gynecologic cancers contributing to this 0.02% decrease in specificity.¹⁶ In addition, Cotter et al⁹ performed a meta-analysis of false-positive stool DNA methylation tests that did not show increased ADC risk; however, the included studies were heterogeneous with respect to test panel and patient risk factors; only 2 methylated DNA markers were common to all included studies, each of which had different designs.

The mt-sDNA pivotal study showed the performance characteristics of mt-sDNA for CRC screening in individuals at average risk for CRC. Every participant received an mt-sDNA test followed by a colonoscopy within 90 days of study informed consent, thus providing cohorts of patients with negative colonoscopies and either mt-sDNA–positive or –negative tests for review. We specifically focused on subgroups of these individuals with accessible medical records at several larger enrolling sites.

We found that incident aerodigestive cancers occurred rarely in asymptomatic average-risk screening patients with false-positive mt-sDNA results. Single cases of pancreatic cancer, CRC, and low-grade parotid tumor, and 2 cases of lung cancer were identified in that group. The CRC was stage I (pT2, N0), located in the cecum, and discovered when the patient developed iron-deficiency anemia shortly before he was due for a 3-year surveillance colonoscopy for multiple nonadvanced adenomas found on the index colonoscopy. Although this may have been a de novo cancer, it also could represent a false-negative index colonoscopy. This possibility speaks to the importance of a meticulous examination of the colon and rectum after a positive mt-sDNA test. In the pivotal study, the routine colonoscopy screening component was performed blinded to the study mt-sDNA result. Johnson et al¹⁷ reported on the increase in colonoscopic detection of neoplastic precursor lesions and specifically those located in the right colon with flat/sessile morphology in typical clinical practice when gastroenterologists were aware of the mt-sDNA result in comparison with findings by gastroenterologists in the same practice who performed the pivotal study colonoscopy blinded to the mt-sDNA-results.

Among participants with negative mt-sDNA tests, 11 incident ADCs were found including 6 lung cancers, 2 head and neck cancers, 1 hepatic hemangiosarcoma, and 2 pancreatic cancers. Most incident cancers in both groups occurred 3 or more years after the mt-sDNA tests.

This study had several limitations. First, the cohort was predominantly Caucasian. Although the mt-sDNA test performance in non-white populations^7,8,18 appears similar to white populations, long-term follow-up evaluation in more diverse patients is not yet available for comparison with the current results. Second, we examined a subgroup of all persons from the pivotal study based on our ability to obtain clinical follow-up data from selected sites. Identification of postpivotal study cancers was dependent on availability of clinical and tumor registry records, and it is possible that some cancers were missed. This limitation was mitigated by corroboration with both a survey and a national vital status registry (Accurint) for the Mayo clinic sites, by local and provincial tumor registry for the Calgary site, and by medical records and telephone survey for the WCAR site. ADC rates were similar between study sites with few patients lost to follow-up evaluation, providing a measure of internal validation of the different methods for cancer ascertainment. A third limitation was the relatively small study population with a limited number of aerodigestive cancers, precluding a more robust conclusion about the study findings. Some might consider the duration of follow-up evaluation to be short; however, more than 93% of participants with discordant results were followed up for more than 3 years. Additional events occurring thereafter have a decreasing temporal association with mt-sDNA test results and therefore a decreasingly plausible association. Incident cancers in both groups occurred at a median time of 2.9 years (IQR, 2.1–3.7 y) after mt-sDNA testing.

In conclusion, the incidence of ADCs in a cohort of individuals at average risk for CRC who participated in a screening study, with negative screening colonoscopies and either positive or negative mt-sDNA results, was very low and not significantly different over a median of 5.4 years of follow-up evaluation. Further investigation of asymptomatic individuals solely based on false-positive mt-sDNA results after high-quality negative colonoscopy does not appear to be warranted.

What You Need to Know.

Background

We aimed to compare the incidence of digestive cancers in persons with negative results from colonoscopies and positive vs negative results from multitarget stool DNA tests for colorectal cancer and vs expected incidence.

Findings

In a retrospective study, we found that incident digestive cancers were uncommon among subjects with negative findings from colonoscopies, regardless of discordant or concordant results from multitarget stool DNA tests.

Implications for patient care

Patients with negative results from high-quality colonoscopies should not undergo further testing.

Abbreviations used in this paper:

ADC

aerodigestive cancer

CRC

colorectal cancer

IQR

interquartile range

mt-sDNA

multi-target stool DNA

SEER

Surveillance, Epidemiology, and End Results

WCAR

Wisconsin Center for Advanced Research