Authors’ Reply

We appreciate the interest shown by Collins et al.¹ in our study of screening advanced colorectal neoplasia in people with CKD (DETECT). They assert that the performance characteristics of fecal immunochemical testing (FIT) arising from our study are unreliable because of verification bias and that high false positive rates were found due to low test thresholds.¹

We note that the authors have conducted a similar study but found lower test-positive and sensitivity values.² The observed differences in the test estimates are not unexpected. Their study was conducted in a single center, restricted only to transplant recipients, and of smaller sample size. A quantitative FIT (Eiken OC-Sensor) was the chosen screening tool for DETECT, because it is the screening test of choice by the National Bowel Cancer Screening program in Australia and Spain, rather than the InSure FIT (brush techniques) used by Collins et al.¹ Prior studies have also indicated lower positivity rates and sensitivity estimates for advanced colorectal neoplasia with the InSure FIT compared with the Eiken OC-Sensor FIT in the general population.^3,4

We suggest that the two-step reference standard (colonoscopy for FIT-positive patients and clinical follow-up for both FIT-positive and FIT-negative patients) is in fact the correct one and not colonoscopy for all for three reasons. First, given the appreciable risk of colonoscopy in potentially high-risk patients, we could not ethically justify subjecting FIT-negative patients to an unnecessary procedure. Second, what Collins et al.¹ have not considered is the potential for overdiagnosis. We acknowledge that this has only been recognized as a major issue recently, postdating their publication in 2012. Overdiagnosis occurs when the disease detected through screening does not cause morbidity and/or death.⁵ This is an important concept to consider, particularly in patients with limited life expectancy, such as those with CKD, and when competing events, such as cardiovascular diseases, predominate as the major cause of death. Overdiagnosis, the downside of cancer screening, will trigger a sequence of overtreatment, with the attendant adverse events and without benefit, because the screened individual will never experience the health consequences of the target condition. We anticipated that the harms associated with diagnostic colonoscopies are likely to be higher among those with CKD and thus, limit the benefits of routine screening. Our findings confirmed that colonoscopies and subsequent treatments, including polypectomy, incurred at least a 10-fold increased risk of major complications, including perforations and infections, in patients with CKD compared with those reported in the general population. Third, our two-step process is clinically feasible and allows for external generalizability.

In our study, a follow-up of 2 years was chosen to ensure that all clinically relevant colorectal cancers had enough time to progress to a detectable stage and that new (interval) cancers that develop after the index test (FIT) were also being detected. Importantly, only 14 additional patients (0.9%) were diagnosed with advanced colorectal neoplasia at the end of the 2-year follow-up, indicating that, even if differential verification bias may exist in theory, it is unlikely to be clinically relevant.

Disclosures

None.