We read with interest the report by Wong et al.1 of the Detecting Bowel Cancer in CKD (DETECT) study of colorectal cancer screening in 1706 patients with CKD. DETECT investigated the test performance characteristics of fecal immunochemical testing (FIT) to detect advanced colorectal neoplasia. The authors reported a high sensitivity (90%) and negative predictive value (99%), with a positivity rate of 22%.
Unfortunately, the estimates of sensitivity, specificity, and negative predictive values are unreliable because not all participants had the reference standard (colonoscopy) performed. In DETECT, only participants with a positive FIT were invited for colonoscopy (323 of 1706; 19%). The authors attempted to mitigate verification bias by using clinical follow-up at 2 years as an alternative reference standard. Although this provided colonoscopy outcomes for 98 of 1337 (7%) patients who were FIT negative at baseline, absence of neoplasia was presumed for the other 1239 patients if there were no symptoms or a subsequent positive FIT result. However, advanced neoplasia is usually asymptomatic, and the opportunity to identify it with colonoscopy was limited to a subgroup of patients, most of whom were FIT positive. This results in differential verification bias and inaccurate estimates of neoplasia prevalence, sensitivity, and specificity.2
The high FIT-positive rate (22%) is also notable. DETECT used a relatively low fecal hemoglobin threshold of 10 μg/g (many screening programs use >20 μg/g) in a population where aspirin and anticoagulant use were common. Unsurprisingly, specificity was relatively low (83%). Given the high rate of serious colonoscopy complications observed (1.5%), this gives cause for concern.
Further, the data from DETECT give an estimate of sensitivity for advanced adenoma (i.e., excluding the seven cases of cancer) of 89%. This is much higher than published general population studies with low risk of verification bias that used a 10 μg/g threshold (pooled sensitivity, 40%; specificity, 90%),3 and our previous study in patients with CKD (transplant) where all participants had colonoscopy (FIT positivity, 12%; sensitivity, 25%; specificity, 91%; negative predictive value, 90%).4 Such comparisons highlight the likely biased nature of the DETECT estimates.
We therefore believe that the reported diagnostic accuracy results from DETECT are misleading. Both sensitivity and negative predictive value are likely overestimated and, unfortunately, the data are insufficient to determine the accuracy of FIT. Without an unbiased estimation of sensitivity, and with the reported high false positivity rate and high colonoscopy complication rate, it remains unclear whether screening with FIT to detect premalignant advanced neoplasia in patients with CKD is appropriate.
Disclosures
None.
Footnotes
Published online ahead of print. Publication date available at www.jasn.org.
References
- 1.Wong G, Hope RL, Howard K, Chapman JR, Castells A, Roger SD, et al.: One-time fecal immunochemical screening for advanced colorectal neoplasia in patients with CKD (DETECT study). J Am Soc Nephrol 30: 1061–1072, 2019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.de Groot JA, Bossuyt PM, Reitsma JB, Rutjes AW, Dendukuri N, Janssen KJ, et al.: Verification problems in diagnostic accuracy studies: Consequences and solutions. BMJ 343: d4770, 2011 [DOI] [PubMed] [Google Scholar]
- 3.Imperiale TF, Gruber RN, Stump TE, Emmett TW, Monahan PO: Performance characteristics of fecal immunochemical tests for colorectal cancer and advanced adenomatous polyps: A systematic review and meta-analysis. Ann Intern Med 170: 319–329, 2019 [DOI] [PubMed] [Google Scholar]
- 4.Collins MG, Teo E, Cole SR, Chan CY, McDonald SP, Russ GR, et al.: Screening for colorectal cancer and advanced colorectal neoplasia in kidney transplant recipients: Cross sectional prevalence and diagnostic accuracy study of faecal immunochemical testing for haemoglobin and colonoscopy. BMJ 345: e4657, 2012 [DOI] [PMC free article] [PubMed] [Google Scholar]