Diagnostic performance of multi-target stool DNA and CT colonography for non-invasive colorectal cancer screening

Abstract

Background:

Multi-target stool DNA (mt-sDNA) screening has increased rapidly since simultaneous approval by the U.S. Food and Drug Administration (FDA) and Centers for Medicare and Medicaid Services (CMS) in 2014, whereas CT colonography screening remains underused without CMS coverage.

Purpose:

To report post-approval clinical experience with mt-sDNA screening for colorectal cancer (CRC) and compare results with CT colonography screening at the same center.

Materials and Methods:

In our retrospective cohort study, asymptomatic adults underwent clinical mt-sDNA screening over a 5-year interval (2014–2019). EMR search verified test results and documented subsequent optical colonoscopy (OC) and histopathologic findings. A similar analysis was performed for CT colonography screening over a 15-year interval (2004–2019), considering both 6-mm and 10-mm polyp size positivity thresholds. Statistical testing compared results for the two modalities.

Results:

3987 asymptomatic adults (mean age, 64±9 years; 2567 women) underwent mt-sDNA screening and 9656 (mean age, 57±8 years; 5200 women) underwent CT colonography. Test-positive rates for mt-sDNA, 6-mm-, and 10-mm-threshold CT colonography were 15.2%, 16.4%, and 6.7%, respectively. Similarly, OC follow-up rates for positive tests were 13.1%, 12.3%, and 5.9%, respectively. Positive predictive value (PPV) for any neoplasm ≥6 mm, advanced neoplasia, and CRC for mt-sDNA was 54.2%, 22.7%, and 1.9%; for 6-mm-threshold CT colonography was 76.8%, 44.3%, and 2.7%; and for 10-mm-threshold CT colonography was 84.5%, 75.2%, and 5.2%, respectively (p<0.001 for all except CRC 6-mm CT colonography). The overall detection and yield of histologically-proven advanced neoplasia was 2.7%, 5.0%, and 4.4%, respectively (p<0.001), and for CRC was 0.23%, 0.31%, and 0.30%, respectively.

Conclusion:

The relative yield of advanced neoplasia at CT colonography screening was nearly double that of mt-sDNA screening, presumably corresponding to increased cancer prevention. By adjusting the size threshold of positivity to 10 mm, CT colonography preferentially targets advanced neoplasia and cancer, and should be considered over a 6-mm threshold.

Introduction

Optical colonoscopy (OC) remains the dominant colorectal cancer (CRC) screening strategy in the U.S., providing both cancer prevention and detection. However, OC is an invasive and resource-intense primary screening examination. Less invasive CRC screening tests provide an alternative complementary option for population-based screening, with positive cases referred to OC as the de facto therapeutic standard. However, these non-invasive tests may differ in their ability to prevent CRC through the detection of advanced precancerous polyps. Currently available non-invasive screening examinations that are approved by the U.S. Food and Drug Administration (FDA) include various stool-based tests and CT colonography. Vast prior screening experiences with fecal occult blood and immunochemical testing (FOBT and FIT) have been previously published,(1) but the data are much more limited for multi-target stool DNA (mt-sDNA) and CT colonography screening, which were only recently included in the USPSTF guidelines.(2) Data are just now emerging on the post-approval clinical screening experience with the Cologuard (Exact Sciences, Madison, WI) multi-target stool DNA test.(3–5) Nonetheless, nationwide use of this stool-based test has dramatically increased since simultaneous approval for screening was granted in August 2014 by the FDA and the Centers for Medicare and Medicaid Services (CMS), with millions of screening test already performed. Published data on CT colonography screening is mainly limited to trial data and clinical practice at a single center.(6–9)

Clinical validation trials(8–10) can derive the sensitivity and specificity of these non-invasive tests for key outcomes such as advanced neoplasia and CRC, since every patient undergoes the OC reference standard. However, in actual clinical practice, diagnostic evaluation is limited to test positive rate, positive predictive value (PPV), and diagnostic yield (or detection rate, DR) as the key performance indicators, since OC is generally not performed unless the non-invasive test is positive. One critical difference between stool-based tests and CT colonography is the latter provides specific information on detected pathology that results in various degrees of positivity (e.g., small 6–9 mm polyps of debatable importance, large polyps ≥ 10 mm concerning for advanced adenomas, or frank masses ≥3 cm concerning for CRC). For positive cases, CTC also provides information on polyp location, morphology, and multiplicity. As such, CT colonography can operate at different target thresholds, whereas mt-sDNA and the other stool-based tests typically provide a binary positive or negative result in clinical practice, without insight into the nature or degree of positivity.

The purpose of our study was therefore to report our post-approval clinical screening experience to date with mt-sDNA, and to compare these results with our CT colonography clinical screening experience, with all studies performed at the same medical center, and applied to the same general screening population. The primary outcome measures of interest were the PPV and diagnostic yield for advanced neoplasia and CRC.

Materials and Methods

Our HIPAA-compliant study was approved by our institutional review board; the need for written informed consent was waived due to the retrospective nature of assessment.

Patients

The mt-sDNA screening cohort for our study was derived by identifying all outpatient adults undergoing asymptomatic clinical CRC screening with Cologuard (Exact Sciences, Madison, WI) at our academic center since the time of FDA approval in August 2014, using a dedicated EMR search. Previously, our center participated in the validation trial,(10) where patient recruitment ended in 2012; these trial patients were not included. Consecutive asymptomatic adults underwent post-approval mt-sDNA screening over a 5-year interval from November 2014 to May 2019. We excluded samples that could not be processed. The CT colonography screening cohort for comparison was derived from consecutive asymptomatic adults scanned over a 15-year interval from April 2004 to May 2019. We excluded examinations deemed technically inadequate in terms of colonic preparation or distention. For both screening cohorts, potential patients were excluded for known colorectal symptoms (e.g., rectal bleeding), weight loss, iron-deficiency anemia, or other concerning presenting issues beyond asymptomatic screening.

mt-sDNA screening

For mt-sDNA screening, after the provider ordered the test, the patients engaged with the Cologuard patient navigation system. The collection kits were directly mailed to the patient’s home address, where stool samples were collected and then shipped back to the Exact Sciences laboratory, where the samples were processed and analyzed. Results were then reported back to the ordering provider in a binary fashion (positive or negative), without any quantitative information. The recommendation for a positive mt-sDNA test is referral to OC, whereas repeat routine screening in three years is recommended for a negative mt-sDNA test. The EMR was reviewed (PMG, NDY, BW) for all patients to 1) ascertain the mt-sDNA test result, 2) confirm that the examination was performed for the purpose of asymptomatic CRC screening, and 3) to search for any subsequent colonoscopy examination, as well as histopathology results for any colorectal polyps/tumors found at OC. Polyp size data at OC was utilized. We also searched for any CT colonography examinations performed for a positive mt-sDNA test.

CT colonography screening

For CT colonography, we similarly included all asymptomatic outpatient adults referred for first-time CT colonography screening, identified through our dedicated CTC clinical database. Our CT colonography screening technique has been described in detail previously,(11, 12) but generally consists of a low-volume bowel preparation the evening before, followed by low-dose non-contrast CT scanning after colonic distention. Our standard bowel preparation consists of a cathartic agent (typically magnesium citrate), followed by oral contrast tagging with low-density barium (2% w/v) and a water-soluble iodinated agent.(13) Colonic distention is achieved via automated low-pressure CO₂.(14) Multi-detector CT scanning is performed at 120 kV, with variable mA to achieve low-dose but diagnostic CT colonography images (mean effective dose ~5 mSv). The protocol has changed very little over the years.

CT colonography interpretation was performed by one of 13 experienced abdominal radiologists at our center. Although positive results at CT colonography are reported on a polyp-by-polyp basis, the patient-level results are more relevant for comparison with sDNA screening and for clinical outcomes. In a prior study, we focused on factors affecting PPV at the per-polyp level,(7) which is beyond the scope of our current study. For patient-level results, each examination is assigned a CT colonography Reporting and Data System (C-RADS) category: C0 for technically inadequate, C1 (negative) for no non-diminutive polyps (>5 mm), C2 for small 6–9 mm polyps, C3 for large ≥10 mm polyps, and C4 for masses (≥3 cm).(15) Positive CT colonography cases can therefore be defined at different size thresholds: ≥6 mm, ≥10 mm, and ≥3 cm, which correspond to C-RADS C2-C4, C3-C4, and C4, respectively. Unlike mt-sDNA, patients positive at the C2 level (6–9 mm polyps) have the option to undergo 3-year CT colonography polyp surveillance in lieu of OC referral.(16) As with the mt-sDNA cohort, all results from OC follow-up following a positive CT colonography were reviewed and tabulated in a similar manner. A small subset of CT colonography screening patients avoided intermediate OC for obvious colorectal masses but rather underwent surgical resection for pathology proof. CT colonography data were extracted from our clinical database, with supplementary EMR review as needed. (PMG, NDY, BW).

Definitions

Beyond the test-positive rates (TPR) and OC referral rates, the main outcome measures of interest for our study were the PPV and DR (clinical yield) for advanced neoplasia and CRC. Both PPV and DR require subsequent OC for determination. Advanced neoplasia is defined as any advanced adenoma or invasive adenocarcinoma (CRC). Advanced adenomas are defined by large size (≥10 mm), or by advanced histologic features of high-grade dysplasia or a prominent villous component (tubulovillous or villous).(6) Large (≥10 mm) and/or dysplastic sessile serrated polyps (SSP) were also considered advanced. Non-advanced, non-diminutive (6–9 mm) tubular adenomas and SSPs constituted a secondary outcome measure of interest, whereas non-advanced diminutive polyps and non-neoplastic polyps represented the lowest polyp category, given their lack of clinical relevance. Metastatic disease and non-adenocarcinoma primary cancers were excluded. Each patient was classified according to the most aggressive/relevant polyp histology found at OC. Patients without available histology results (eg, polyp not resected or retrieved) were necessarily excluded from PPV calculations. Test results were further stratified according to age and gender categories to account for any demographic differences between the two screening cohorts.

Statistical Analysis

Comparison of proportions for statistical significance comparing results between the two screening modalities was performed using either the chi-squared test or two-sample t-tests, as appropriate. All p-values were based on two-sided tests and were considered statistically significant if less than 0.05. 95% confidence intervals (95% CI) were derive using Wilson’s test. All analyses were done using R software, version 3.3.2 (2016–10-31 (R Core Team (2016). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/).

Results

Patient Characteristics

From a total of 4033 consecutive asymptomatic adults who underwent post-approval mt-sDNA screening over a 5-year interval, 46 (1.1%) were excluded as the samples could not be processed by the laboratory, resulting in a final sDNA screening cohort of 3987 adults (mean age, 64±9 years; 1420 men, 2567 women). The CT colonography screening cohort was derived from 9719 consecutive asymptomatic adults scanned over a 15-year interval. After exclusion of 63 (0.6%) deemed technically inadequate, the final CT colonography cohort consisted of 9656 adults (mean age 57±8 years; 4456 men, 5200 women).

Test-positive rates

The test-positive rate for mt-sDNA was 15.2% (605/3987) and for CT colonography at the 6-mm, 10-mm, and 3-cm thresholds of positivity was 16.4% (1585/9656), 6.7% (647/9656), and 0.5% (49/9656), respectively (Fig 1 and Table 1). TPRs for CTC at 10-mm and 3-cm thresholds were significantly lower than mt-sDNA (p<0.001), but the rate was similar between 6-mm CTC and mt-sDNA (p=0.08).

Figure 1.

Flowcharts for (A) mt-sDNA and (B) CT colonography screening

Table 1.

Results of mt-sDNA and CT colonography Screening at a Single Center

	Test Positive Rate	OC Follow-up Rate	Positive Predictive Value (PPV)			Overall Yield (Detection Rate)
			Neoplasms ≥6 mm	Advanced Neoplasia	Invasive Cancer (CRC)	Advanced Neoplasia	Invasive Cancer (CRC)
mt-sDNA (n=3987)	15.2% (14.1–16.3%) (605/3987)	13.1% (12.2–14.4%) (522/3987)	54.2% (50.9–59.8%) (258/476)	22.7% (19.4–27.3%) (108/476)	1.9% (0.9–3.6%) (9/476)	2.7% (2.2–3.3%) (108/3987)	0.23% (0.11–0.44%) (9/3987)
CT colonography (n=9656)
6-mm threshold	16.4% (15.7–17.1%) (1585/9656)	12.3% (11.7–12.9%) (1192/9656)	76.8%* (74.3–79.6%) (842/1096)	44.3%* (41.2–47.2%) (486/1096)	2.7% (1.9–3.9%) (30/1096)	5.0%* (4.6–5.5%) (486/9656)	0.31% (0.22–0.44%) (30/9656)
10-mm threshold	6.7%* (6.2–7.2) (647/9656)	5.9%* (5.4–6.4%) (571/9656)	84.5%* (81.4–87.9%) (473/560)	75.2%* (71.3–79.3%) (421/560)	5.2%* (3.6–7.3%) (29/560)	4.4%* (4.0–4.8%) (421/9656)	0.30% (0.21–0.43%) (29/9656)
3-cm threshold	0.5%* (0.38–0.67%) (49/9656)	0.5%* (0.35–0.62%) (45/9656)	93.3%* (82.6–98.8%) (42/45)	93.3%* (82.6–98.8%) (42/45)	42.2%* (29.0–57.7%) (19/45)	0.43%* (0.32–0.58%) (42/9656)	0.20% (0.13–0.31%) (19/9656)

^*Indicates p<0.001 for CTC vs. mt-sDNA; see text for more details

Percentages in parentheses represent 95% CIs

OC follow-up rates

The overall OC follow-up rate from mt-sDNA screening for a positive test result was 13.1% (522/3987). A total of 83 patients with a positive mt-sDNA test had not yet undergone OC, with a mean clinical follow-up interval after positive sDNA test of 1.8 years. Findings at OC in the 522 referred patients included invasive cancers (CRC) in nine (Figure 2), advanced adenomas or SSPs in 99, small (6–9 mm) non-advanced tubular adenomas (TA) or SSPs in 150, diminutive or non-neoplastic polyps in 140, and no polyps in 78 patients. Pathology results were unavailable in 46 patients, typically for small polyps that were not resected or not retrieved, leaving 476 patients for PPV assessment. These results are shown in Figure 2, and summarized in Table 1.

Figure 2. Invasive cancer detected at OC following positive mt-sDNA screening test in two asymptomatic adults.

A. Large ulcerated cecal mass in a 51-year-old woman.

B. Malignant 2 cm sigmoid polyp in a 62-year-old woman.

For CT colonography, the overall OC follow-up rate for positive tests at the 6-mm, 10-mm, and 3-cm thresholds were 12.3% (1192/9656), 5.9% (571/9656), and 0.5% (45/9656), respectively. OC follow-up rates for CTC at 10-mm and 3-cm thresholds were significantly lower than mt-sDNA (p<0.001), but the rate was similar between 6-mm CTC and mt-sDNA (p=0.20). There were 393 patients with a positive CT colonography who did not undergo OC, most of whom opted for CT colonography polyp surveillance for small (6–9 mm) polyps. Findings at OC in the 1192 asymptomatic CT colonography patients referred included CRCs in 30, advanced adenomas/SSPs in 456, small (6–9 mm) TAs or SSPs in 356, diminutive or non-neoplastic polyps in 175, and no polyps in 79 patients. In 96 patients, pathology results were unavailable, leaving 1096 for PPV assessment. These results are also depicted in Figure 1, and summarized in Table 1. Pathology results could not be retrieved in 11 patients at the 10-mm threshold, leaving 560 for PPV assessment at that threshold.

Positive predictive value (PPV)

For the positive mt-sDNA cohort, the PPV for finding any neoplasm ≥6 mm at OC was 54.2% (258/476), PPV for finding any advanced neoplasia was 22.7% (108/476), and PPV for finding cancer was 1.9% (9/476). Similarly, for the CT colonography cohort positive at the 6-mm threshold, the PPV for finding any neoplasm ≥6 mm at OC was 76.8% (842/1096), for any advanced neoplasia was 44.3% (486/1096), and for CRC 2.7% (30/1096). For CT colonography at the 10-mm positive threshold, the PPV for advanced neoplasia was 75.2% (421/560), and for CRC was 5.2% (29/560). For CT colonography at the 3-cm threshold, the PPV for advanced neoplasia and CRC was 93.3% (42/45) and 42.2% (19/45), respectively. These PPV results are summarized in Table 1. All PPVs comparisons were significantly higher for CTC compared with mt-sDNA (p<0.001), with the exception of CRC for 6-mm CTC versus mt-sDNA (p=0.40). Test results according to age and gender categories are shown in Table 2.

Table 2.

Results of mt-sDNA and CT colonography Screening According to Age and Gender

	Test Positive Rate	OC Follow-up Rate	Positive Predictive Value (PPV)			Overall Yield (Detection Rate)
			Neoplasms ≥6 mm	Advanced Neoplasia	Invasive Cancer (CRC)	Advanced Neoplasia	Invasive Cancer (CRC)
mt-sDNA total cohort (n=3987)	15.2%	13.1%	54.2%	22.7%	1.9%	2.7%	0.23%
50–60 year-old men (n=521)	11.7% (9.2–14.8%) (61/521)	10.6% (8.2–13.5%) (55/521)	61.5% (48.0–73.5%) (32/52)	25.0% (15.2–38.2%) (13/52)	0% (0–0%) (0/52)	2.5% (1.5–4.2%) (13/521)	0% (0–0%) (0/521)
50–60 year-old women (n=846)	7.8% (6.2–9.8%) (66/846)	6.9% (5.3–8.8%) (58/846)	62.0% (48.2–74.1%) (31/50)	24.0% (14.3–37.4%) (12/50)	2.0% (0.4–10.5%) (1/50)	1.4% (0.8–2.5%) (12/846)	0.12% (0.02–0.7%) (1/846)
60–70 year-old men (n=492)	18.1% (14.9–21.7%) (89/492)	13.2% (10.5–16.5%) (65/492)	64.3% (51.2–75.6%) (36/56)	21.4% (12.7–33.8%) (12/56)	3.6% (1.0–12.1%) (2/56)	2.4% (1.4–4.2%) (12/492)	0.41% (0.1–1.5%) (2/492)
60–70 year-old women (n=938)	14.5% (12.4–16.9%) (136/938)	12.6% (10.6–14.9%) (118/938)	50.5% (41.2–59.7%) (55/109)	26.6% (19.2–35.6%) (29/109)	2.8% (0.94–7.8%) (3/109)	3.1% (2.2–4.4%) (29/938)	0.21% (0.11–0.94%) (3/938)
70+ year-old men (n=440)	20.7% (17.2–24.7%) (91/440)	18.2% (14.9–22.1%) (80/440)	50.6% (39.7–61.5%) (39/77)	23.4% (15.3–34.0%) (18/77)	0.0% (0–0%) (0/77)	4.1% (2.6–6.4%) (18/440)	0.00% (0–0%) (0/440)
70+ year-old women (n=750)	21.6% (18.8–24.7%) (162/750)	19.5% (16.8–22.5%) (146/750)	49.2% (40.9–57.7%) (65/132)	18.2% (12.5–25.6%) (24/132)	2.3% (0.8–6.5%) (3/132)	3.2% (2.2–4.7%) (24/750)	0.40% (0.14–1.2%) (3/750)
CTC (6-mm)* total cohort (n=9656)	16.4%	12.3%	76.8%	44.3%	2.7%	5.0%	0.31%
50–60 year-old men† (n=3126)	17.7% (16.4–19.1%) (554/3126)	12.7% (11.6–13.9%) (397/3126)	79.7% (75.3–83.5%) (295/370)	44.1% (39.1–49.1%) (163/370)	2.4% (1.3–4.6%) (9/370)	5.2% (4.5–6.1%) (163/3126)	0.29% (0.10–0.40%) (9/3126)
50–60 year-old women† (n=3754)	11.8% (10.8–12.8%) (442/3754)	9.2% (8.3–10.1%) (344/3754)	73.1% (68.0–77.7%) (231/316)	40.5% (35.2–46.0%) (128/316)	2.2% (1.1–4.5%) (7/316)	3.4% (2.9–4.1%) (128/3754)	0.19% (0.09–0.39%) (7/3754)
0–70 year-old men (n=1061)	22.8% (20.4–25.4%) (242/1061)	16.6% (14.5–19.0%) (176/1061)	78.9% (71.9–84.5%) (127/161)	47.2% (39.6–54.9%) (76/161)	3.7% (1.7–7.9%) (6/161)	7.2% (5.8–8.9%) (76/1061)	0.57% (0.26–1.2%) (6/1061)
0–70 year-old women (n=1143)	16.6% (14.6–18.9%) (190/1143)	13.6% (11.7–15.7%) (155/1143)	73.8% (66.1–80.3%) (107/145)	40.7% (33.0–48.8%) (59/145)	1.4% (0.4–4.9%) (2/145)	5.2% (4.0–6.7%) (59/1143)	0.17% (0.05–0.63%) (2/1143)
70+ year-old men (n=294)	30.6% (25.6–36.1%) (90/294)	22.1% (17.7–27.2%) (65/294)	78.2% (65.6–87.1%) (43/55)	58.2% (45.0–70.3%) (32/55)	3.6% (1.0–12.3%) (2/55)	10.9% (7.8–15.0%) (32/294)	0.68% (0.19–2.5%) (2/294)
70+ year-old women (n=278)	24.1% (19.5–29.5%) (67/278)	19.7% (15.5–24.9%) (55/278)	80.0% (66.4–88.5%) (39/49)	57.1% (43.3–70.0%) (28/49)	8.2% (3.2–19.2%) (4/49)	10.1% (7.1–14.2%) (28/278)	1.4% (0.6–3.6%) (4/278)

^*Results listed are for the 6-mm threshold of positivity

^†Also includes a small minority of patients under 50.

Overall yield (detection rate) of histologically-proven advanced neoplasia

When considering the entire screened population, the overall yield or DR of histologically-proven advanced neoplasia at mt-sDNA was 2.7% (108/3987) and for CT colonography (6-mm threshold) screening was 5.0% (486/9656); an 85.8% increase for the latter (p<0.001). The yield of invasive colorectal adenocarcinoma for mt-sDNA and CT colonography (6-mm positive threshold) was similar at 0.23% (9/3987) and 0.31% (30/9656), respectively (p=0.43). The increased yield for advanced neoplasia for CT colonography over mt-sDNA was observed for all age and gender groups (Table 2). When using a 10-mm threshold as positive for CT colonography, the yields were only minimally diminished for advanced neoplasia to 4.4% (421/9656; p<0.001), and for CRC to 0.30% (29/9656;p=0.43), respectively, despite a 59% reduction in the test-positive rate (from 16.4% to 6.7%) and 52% reduction in OC follow-up rate (from 12.3% to 5.9%). This reflects the fact that very few 6–9 mm are histologically advanced.(17) Additional DR data are shown in Tables 1 and 2.

Additional testing after mt-sDNA screening

A search of the mt-sDNA negative cohort (n=3382) revealed that 58 (1.7%) patients underwent subsequent OC. Colonoscopy in these sDNA-negative patients found anal cancer in one patient (Figure 3), advanced adenomas in two, non-diminutive non-advanced neoplasms in nine, diminutive and non-neoplastic polyps in 21, and no polyps in 25 patients. A detailed search of the mt-sDNA positive cohort revealed that 18 patients underwent CT colonography, of which 13 went on to OC (Figure 4). One positive sDNA patient with negative OC was found to have metastatic esophageal cancer (Figure 5).

Figure 3. Rectal mass (anal squamous cell carcinoma) in a 67-year-old woman with a negative mt-sDNA screening test (asymptomatic at the time of screening).

A. IV contrast-enhanced transverse (axial) abdominal CT performed for a palpable right groin mass nine months after screening sDNA test shows low attenuation right inguinal lymphadenopathy (arrow) and a large rectal mass (*). Subsequent OC (not shown) confirmed the large rectal mass, which proved to be anal squamous cell carcinoma on biopsy.

B. Transverse (axial) CT image through the upper abdomen also showed multiple low-attenuation liver tumors (arrowhead) that proved to be metastatic disease on ultrasound-guided biopsy (not shown). The patient died one year later after failing to respond to radiation and chemotherapy.

Figure 4. Colon mass (4-cm tubular advanced adenoma) in an asymptomatic 64-year-old woman with positive mt-sDNA and CT colonography examinations.

A and B. 3D endoluminal (A) and 2D coronal (B) images from CT colonography demonstrate a large 4-cm mass (blue arrow in B) in the ascending colon.

C. Image from subsequent OC re-demonstrates the mass, which required right hemi-colectomy for complete resection, but no malignant focus was found at pathology.

Figure 5. Metastatic esophageal adenocarcinoma (extracolonic aerodigestive primary) in a 69-year-old woman with positive mt-sDNA (for asymptomatic screening) and extracolonic CT colonography findings.

A. Axial image from CT colonography one month after positive sDNA test demonstrates a large lobulated mass in the gastrohepatic region (arrowheads). CT colonography was performed instead of colonoscopy due to anticoagulation medication.

B and C. Subsequent IV contrast-enhanced abdominal CT (B) better delineates bulky heterogeneous lymphadenopathy (arrowheads), which was biopsied with ultrasound guidance (C) and revealed poorly-differentiated adenocarcinoma. No concerning colorectal findings were seen at CT colonography or subsequent OC.

D. Image from EGD shows ulcerated mass in distal esophagus, which proved to be the primary site of malignancy. The patient died less than one year later.

Discussion

Optical colonoscopy remains the dominant CRC screening strategy in the United States, but also serves in a diagnostic referral role following positive non-invasive/non-therapeutic CRC test results. Multi-target stool DNA and CT colonography are considered among the most effective of the available non-invasive CRC screening tests in practice. We had the unique opportunity to assess and report on a large single-center experience for comparing these screening tests. Our PPV results from clinical screening with multi-target stool DNA matched closely with those from the large clinical trial by Imperiale et al.,(10) as well as other recently reported post-approval clinical experiences.(3–5) At our institution, as well as many others across the U.S., the utilization of this stool-based CRC screening test has markedly increased since the time of FDA approval (and simultaneous coverage from the CMS for Medicare beneficiaries) in August 2014. In fact, millions of mt-sDNA screening tests have been performed since approval in the U.S. In comparison, CT colonography screening remains vastly underused, perhaps largely due to its uncovered status by CMS for Medicare.(18)

As our results confirm, the differences in PPV and advanced neoplasia detection between mt-sDNA and CT colonography screening are primarily related to differential detection of large/advanced adenomas. In our practice, the PPV for advanced neoplasia was 22.7% for mt-sDNA, 44.3% for 6-mm threshold CTC, and 75.2% for 10-mm threshold CTC. In the mt-sDNA screening trial published by Imperiale in 2014,(10) the PPV for advanced neoplasia was 23.6%, with a 42% sensitivity for detection. This compares with detection rates >90% for CT colonography.(8, 9, 19) This logically accounts for the near doubling in advanced neoplasia detection and yield we observed with clinical CT colonography screening (at 6-mm threshold) compared with mt-sDNA screening (5.0% vs. 2.7%) within the same general asymptomatic screening population. In clinical trials, the sensitivity for CRC detection with mt-sDNA was 92%,(10) compared with 96% at CT colonography,(20) which parallels the narrower difference in CRC detection rates between mt-sDNA and CTC in our practice (0.2% vs. 0.3%). Of note, reported sensitivity of FIT for advanced neoplasia and CRC appear to be somewhat lower than mt-sDNA, but specificity appears to be higher.(1)

In terms of balancing clinical efficacy and cost-effectiveness, the TPR, PPV, DR, and cost characteristics of a non-invasive CRC screening test all combine to determine its overall performance for advanced neoplasia and cancer prevention. For CT colonography, reserving a 10-mm threshold for OC referral is highly effective for targeting advanced neoplasia, with a PPV of >75% (compared with <25% for mt-sDNA), while maintaining a significantly higher detection rate (4.4% vs. 2.7% for mt-sDNA), despite a much lower OC follow-up rate (5.9% vs. 13.1%for mt-sDNA). The resulting higher advanced neoplasia yield despite lower OC utilization favors this 10-mm CTC strategy. When coupled with cost considerations, CT colonography further strengthens its case as the dominant test (ie, more cost effective and more clinically efficacious). Specifically, our departmental billing personnel quote a Medicare reimbursement rate of $509 for mt-sDNA (CPT code 81528), compared with a range of $225-$462 for CT colonography (global fee for diagnostic examination without contrast; CPT code 74261). Screening CT colonography is not yet covered by CMS, but reimbursement would likely be similar to the diagnostic examination. In addition, the initial screening interval for negative CT colonography is 5 years,(15, 21) compared with 3 years for mt-sDNA, which further impacts cost-effectiveness.

Because mt-sDNA is commonly thought of as a “cancer detection test”, it may come as somewhat of a surprise that fewer than 2% of positive cases were due to cancer in our practice, similar to other clinical experiences.(3–5) Put another way, >98% of positive mt-sDNA tests could be considered falsely positive for cancer, although this misses the point of CRC screening. In contrast, if CRC detection were the only goal of screening (which it is not), the 3-cm positive threshold for CT colonography would yield a similar number of cancers as mt-sDNA screening (0.20% vs. 0.23%), at a PPV of >40%, but would send fewer than 1% of those screened to colonoscopy (0.5% vs. 13.1% for mt-sDNA). This strategy, however, would largely lose the benefit of cancer prevention, with an advanced neoplasia detection rate dropping to <1%. Therefore, it is critical for primary care providers and their patients to understand the prime importance of cancer prevention via detection and removal of large advanced adenomas.

As with FIT and OC screening, both mt-sDNA and CT colonography have certain advantages and disadvantages that should be considered by providers and patients when embarking on an individualized screening strategy. In general, both mt-sDNA and CT colonography avoid the invasiveness of OC, with neither requiring any needles, pain or sedation medications, driver, or recovery time. This distinction from OC is further accentuated by the COVID-19 pandemic, especially with regard to exposure risks related to sedation and fecal viral load. Furthermore, specific advantages of mt-sDNA screening over OC and CT colonography include the lack of any bowel preparation, (presumed) avoidance of discomfort or embarrassment, and the ability to perform the test at home. As noted above, the main disadvantages of mt-sDNA (compared with both CT colonography and OC) are the lower PPV and detection rate for advanced adenomas, leading to lower rates of cancer prevention. Other relative disadvantages of mt-sDNA (and all stool-based testing in practice) compared with CT colonography include the binary test result approach, whereby positive tests are not further stratified according to level of importance, and the inability for same-day polypectomy. Unlike CTC, the lack of insight into the nature or degree of positivity (eg, polyp or mass size, morphology, and location) in turn leads to more unnecessary optical colonoscopy examinations, where either no polyps or only clinically unimportant polyps are found. For this relatively common scenario of a negative diagnostic colonoscopy following positive mt-sDNA testing, no additional clinical evaluation is required for this presumed false-positive result, according to the company itself.

Beyond the increased detection of advanced adenomas and the different positivity thresholds , other relative advantages of CT colonography screening over mt-sDNA include the possibility for same-day polypectomy,(12, 22) detection of unsuspected relevant extracolonic findings,(23–25) and the potential for other incidental screening opportunities.(26–30) In terms of opportunistic screening at CT colonography, measures of bone mineral density, aortic calcification, visceral fat, muscle (sarcopenia screening), and hepatic steatosis can all be automated from CT scans,(31–35) and can provide prognostic value.(36, 37) Potential disadvantages of CT colonography include the associated radiation exposure, albeit low-dose,(38) and additional testing related to extracolonic findings that ultimately prove to be insignificant.(39, 40)

Given the relatively low PPV (and specificity) of a positive mt-sDNA test, one might consider CT colonography as a reasonable intermediate next step, especially for those who are reluctant to undergo OC, or are anticoagulated. This would prevent a large number of unnecessary OC examinations, especially if the threshold for polypectomy referral were set at 10 mm. This approach might also provide more reassurance that a rare non-CRC aerodigestive malignancy was not the cause for the positive mt-sDNA result.(41, 42) Without the cross-sectional imaging provided by CT colonography, many patients and providers may be understandably confused as to which test was ‘‘correct’’ in the discordant scenario of a positive mt-sDNA and negative OC. Potential concern over missed cancers related to a DNA-based test could lead to inappropriate additional testing, such as repeat colonoscopy, unnecessary upper endoscopy, additional cross-sectional imaging, or even a PET examination. Such increased utilization could undermine the desired benefits of a non-invasive screening test.

Our study had some limitations. Despite reporting on the largest combined clinical screening experience with both mt-sDNA and CT colonography, this was a retrospective cohort study, with the potential for bias inherent to such a study design. In terms of our study population, the overall mean age of the mt-sDNA screening cohort was significantly older than the CT colonography screening cohort (64 vs. 57 years), which should impact advanced neoplasia prevalence. However, the female predominance in the sDNA cohort was less pronounced compared with the CT colonography cohort (36% vs. 46%), which may partially offset the age difference. Furthermore, we stratified results according to age and sex to account for these differences. We are not aware of other meaningful differences between the two screening cohorts, but the lack of prospective randomization and differing time horizons are limitations. The cohort of CT colonography patients who underwent in vivo surveillance for 6–9 mm polyps (C-RADS C2 category) likely led to an underestimation of disease prevalence in the CT colonography arm. Given the interpretive nature of CT colonography, generalizability may be more of an issue compared with mt-sDNA, although we have shown uniform performance within our group.(43) Importantly, we did not perform a cost-effectiveness analysis, but such a comparison is warranted. Given the recent FDA and CMS approval for sDNA, data from repeat testing are currently limited. Finally, this was a single-center experience, and our results cannot be directly extrapolated to other screening centers, especially in terms of CT colonography screening.

In summary, we found that the diagnostic performance of multi-target stool DNA screening in clinical practice matches well with other recently published data. The relative yield of advanced neoplasia was higher at CT colonography compared with mt-sDNA screening, largely due to improved detection of advanced adenomas, which has important implications for cancer prevention. In terms of optimizing clinical efficacy (and cost-effectiveness), a 10-mm threshold for positivity at CT colonography proved to be highly favorable, with lower test-positive and OC follow-up rates, higher PPVs, and a sustained yield for advanced neoplasia. As with other U.S.-based screening centers, mt-sDNA screening has greatly increased, whereas CT colonography remains underused. When choosing among the various non-invasive screening options, diagnostic performance should be considered, along with other relevant factors such as cost, patient convenience, safety, and incidental screening opportunities.

Summary Statement.

CT colonography screening nearly doubled the detection and yield of advanced neoplasia compared with multi-target stool DNA screening, which translates into improved cancer prevention.

Key Results.

1. Positive predictive value (PPV) for any advanced neoplasia (AN) and colorectal cancer (CRC) for multi-target stool DNA (mt-sDNA) in 3987 adults was 23%, and 2%, respectively; for 6-mm-threshold CT colonography, PPV was 44% and 3%; and for 10-mm-threshold CT colonography was 75%, and 5%, respectively, in 9656 adults.

2. The overall detection and yield of histologically-proven advanced neoplasia at mt-sDNA and 6-mm CT colonography screening was 2.7% and 5.0% (p<0.001), and for CRC was 0.23% and 0.31%, respectively (p=0.48).

Abbreviations:

mt-sDNA

multi-target stool DNA

CRC

colorectal cancer

PPV

positive predictive value

DR

detection rate

OC

optical colonoscopy