Diagnostic performance of fecal immunochemical test and sigmoidoscopy for advanced right-sided colorectal neoplasms

Calibrating Natural History of Cancer Models in the Presence of Data Incompatibility: Problems and Solutions

The calibration of cancer natural history models is often challenged by a lack of representative calibration targets, forcing modellers to rely on potentially incompatible datasets. Using a microsimulation colorectal cancer model as an example, the purposes of this paper are to (1) highlight the reasons for uncertainty in calibration targets, (2) illustrate practical and generalisable approaches for dealing with incompatibility in calibration targets, and (3) discuss the importance of future research in the area of incorporating uncertainty in calibration. The low quality of data and differences in populations, outcome definitions, and healthcare systems may result in incompatibility between the model and the data. Acknowledging reasons for data incompatibility allows assessment of the risk of incompatibility before calibrating the model. Only a few approaches are available to address data incompatibility, for instance addressing biases in calibration targets and their adjustment, relaxing the goodness-of-fit metric, and validation of the calibration targets to the data not used in the calibration. However, these approaches lack explicit comparison and validation, and so more research is needed to describe the nature and causes of indirect uncertainty (i.e. uncertainty that cannot be expressed in absolute quantitative forms) and identify methods for managing this uncertainty in healthcare modelling.

How and why do we screen for colorectal cancer?

After almost 50 years of data analysis, screening for colorectal cancer has proven to be an effective tool in reducing colorectal cancer mortality. However, implementing the optimal strategy represents a challenge for many healthcare facilities around the world. There is much discussion regarding how screening should be done, the optimal tools that should be used, and the proper timing for screening procedures. Another essential step is to maintain the adherence of patients to screening programs. Also, the recommendation for lowering the age to initiate screening is in progress, as there is an increase in colorectal incidence in people born after 1970.

Time Trends in the Diagnosis of Colorectal Cancer With Obstruction, Perforation, and Emergency Admission After the Introduction of Population-Based Organized Screening

IMPORTANCE

Up to 30% of patients with a diagnosis of colorectal cancer (CRC) present as an emergency (an intestinal obstruction, perforation, or emergency hospital admission) (OPE). There are limited data about the association of organized, population-based colorectal cancer screening with the rate of emergency presentations.

OBJECTIVE

To examine the association of CRC screening with OPE at cancer diagnosis and time trends in the rate of OPE after the start of organized CRC screening using a highly sensitive fecal occult blood test.

DESIGN, SETTING, AND PARTICIPANTS

A historical cohort study was conducted among 1861 individuals 52 to 74 years of age with a diagnosis of CRC from January 1, 2007, to December 31, 2015, who lived in Winnipeg, Manitoba, a province with universal health care and an organized CRC screening program. Statistical analysis was performed from January 22, 2019, to February 26, 2020.

EXPOSURES

Variables included prior CRC screening, era of diagnosis, cancer stage at diagnosis, tumor site in the colon, area level mean household income, primary care continuity of care, and comorbidity.

MAIN OUTCOMES AND MEASURES

The primary outcomes were defined as an OPE. Logistic regression was used to evaluate factors associated with OPE at CRC diagnosis. Trends over time were calculated using Joinpoint Regression.

RESULTS

From 2007 to 2015, 1861 individuals 52 to 74 years of age (1133 men; median age, 65.1 years [interquartile range, 60.0-70.3 years]) received a diagnosis of CRC in Winnipeg. Most individuals had good continuity of care and moderate comorbidities. Overall, 345 individuals (18.5%) had an OPE. The rate of emergency hospital admissions decreased significantly from 2007 (the start of the organized, province-wide CRC screening program) to 2015 (annual change, -7.1%; 95% CI, -11.3% to -2.8%; P = .01). There was no change in the rate of obstructions or perforations or stage IV CRCs. Individuals who were up to date for CRC screening were significantly less likely to receive a diagnosis of an OPE (odds ratio, 0.38; 95% CI, 0.28-0.50; P < .001). The results were similar after adding emergency department visits and stage IV CRC at diagnosis to the outcome.

CONCLUSIONS AND RELEVANCE

This study suggests that the rate of emergency hospital admissions decreased over time for individuals who underwent CRC screening, but there was no change in the rate of obstructions and perforations. Individuals who were up to date for CRC screening were less likely to have a CRC diagnosis with an OPE.

Diagnostic Accuracy Of Fecal Occult Blood Tests For Detecting Proximal Versus Distal Colorectal Neoplasia: A Systematic Review And Meta-Analysis

Objective

We conducted a systematic review and meta-analysis aimed at evaluating the differences of diagnostic performance of fecal occult blood tests (FOBTs) in detecting advanced colorectal neoplasms located in the proximal versus distal colorectum.

Methods

PubMed, Embase, Cochrane Library, and Web of Science were searched for eligible articles published before August 17, 2018. Two independent reviewers conducted study assessment and data extraction. Diagnosis-related indicators of FOBT for detecting proximal and distal colorectal neoplasms were summarized, and further stratified by the type of FOBT (guaiac-based FOBT (gFOBT) and immunochemical FOBT (iFOBT)). Pooled sensitivities and specificities were calculated using a random effect model. Summary receiver operating characteristic curves were plotted and area under the curves were calculated.

Results

Overall, 31 studies meeting the inclusion criteria were included in this review. For gFOBT, we found no site-specific difference (proximally vs distally located) of pooled sensitivities observed in the colorectal cancer (CRC), advanced adenomas, and advanced neoplasms groups. As for iFOBT, pooled sensitivities for detecting CRC located in the distal colon/rectum were comparable with that in the proximal colon (proximal vs distal, 0.67, 95% CI 0.62-0.72 vs 0.72, 95% CI 0.68-0.75), while higher pooled sensitivities for detecting advanced adenomas and advanced neoplasms located in the distal colon/rectum than for detecting those in the proximal colon were observed for iFOBT with the values of 0.24 (95% CI 0.22-0.25) vs 0.32 (95% CI 0.30-0.34) and 0.25 (95% CI 0.23-0.28) vs 0.38 (95% CI 0.36-0.40), respectively. Summary receiver operating characteristic curve analyses showed similar patterns for both types of FOBT regarding the diagnostic accuracy for detecting colorectal neoplasms according to the anatomical sites of the colorectum.

Conclusion

iFOBT had higher sensitivity for detecting advanced adenomas and advanced neoplasia located in the distal colon/rectum than that for those in the proximal colon.

Effect of once-only flexible sigmoidoscopy screening on the outcomes of subsequent faecal occult blood test screening

OBJECTIVE

To investigate the outcomes of biennial guaiac faecal occult blood test (gFOBT) screening after once-only flexible sigmoidoscopy (FS) screening.

METHODS

Between 1994 and 1999, as part of the UK FS Screening Trial (UKFSST), adults aged 55-64 were randomly allocated to an intervention group (offered FS screening) or a control group (not contacted). From 2006, a subset of UKFSST participants (20,895/44,041 intervention group; 41,497/87,149 control group) were invited to biennial gFOBT screening by the English Bowel Cancer Screening Programme. We analysed gFOBT uptake, test positivity, yield of colorectal cancer (CRC), and positive predictive value (PPV) for CRC, advanced adenomas (AAs), and advanced colorectal neoplasia (ACN: AA/CRC).

RESULTS

Uptake of gFOBT at first invitation was 1.9% lower (65.7% vs. 67.6%, p < 0.01) among intervention versus control group participants. Positivity was 0.4% lower (2.0% vs. 2.4%, p < 0.01) and CRC yield was 0.08% lower (0.19% vs. 0.27%, p = 0.14). PPVs were also lower in the intervention versus control group, at 10.3% vs. 12.3% ( p = 0.44) for CRC, 22.7% vs. 31.4% ( p < 0.01) for AA, and 33.0% vs. 43.7% ( p < 0.01) for ACN. Among those who refused FS ( n = 5532), gFOBT uptake at first invitation was 47.7%, CRC yield was 0.25%, and PPV for ACN was 46.2%. Among FS attenders ( n = 15,363), uptake was 72.2%, CRC yield was 0.18%, and PPV for ACN was 27.9%.

CONCLUSIONS

Uptake, positivity and PPV of gFOBT screening were reduced following prior offer of FS screening. However, a quarter of FS screened participants receiving a diagnostic examination after positive gFOBT were diagnosed with ACN.

Early Colorectal Cancer Detected by Machine Learning Model Using Gender, Age, and Complete Blood Count Data

BACKGROUND

Machine learning tools identify patients with blood counts indicating greater likelihood of colorectal cancer and warranting colonoscopy referral.

AIMS

To validate a machine learning colorectal cancer detection model on a US community-based insured adult population.

METHODS

Eligible colorectal cancer cases (439 females, 461 males) with complete blood counts before diagnosis were identified from Kaiser Permanente Northwest Region's Tumor Registry. Control patients (n = 9108) were randomly selected from KPNW's population who had no cancers, received at ≥1 blood count, had continuous enrollment from 180 days prior to the blood count through 24 months after the count, and were aged 40-89. For each control, one blood count was randomly selected as the pseudo-colorectal cancer diagnosis date for matching to cases, and assigned a "calendar year" based on the count date. For each calendar year, 18 controls were randomly selected to match the general enrollment's 10-year age groups and lengths of continuous enrollment. Prediction performance was evaluated by area under the curve, specificity, and odds ratios.

RESULTS

Area under the receiver operating characteristics curve for detecting colorectal cancer was 0.80 ± 0.01. At 99% specificity, the odds ratio for association of a high-risk detection score with colorectal cancer was 34.7 (95% CI 28.9-40.4). The detection model had the highest accuracy in identifying right-sided colorectal cancers.

CONCLUSIONS

ColonFlag^® identifies individuals with tenfold higher risk of undiagnosed colorectal cancer at curable stages (0/I/II), flags colorectal tumors 180-360 days prior to usual clinical diagnosis, and is more accurate at identifying right-sided (compared to left-sided) colorectal cancers.

Distribution of colorectal polyps: Implications for screening

BACKGROUND

During the last decades data from different studies reported modifications of the topographic distribution of colorectal cancers (CRCs), with an increased frequency of tumours in proximal colonic segments. Given the documented link between adenomas and CRC, a proximal migration of adenomas over time could be expected as well.

AIM

To evaluate available evidence about the prevalence of adenomas and of sessile serrated polyps across colonic segments, the changing trends in their distribution across the colon and the diagnostic performance of screening tests currently adopted in population based screening programs for lesions located in different colonic segments.

METHODS

Literature search on PubMed, Embase, and Cochrane Database of Systematic Reviews, Database of Abstracts of Reviews of Effects with reference to preferred reporting items for systematic reviews and meta-analysis (PRISMA), considering all adult human studies in English, published between January 2000 and February 2017.

RESULTS

Cross-sectional analysis of endoscopy and pathology data-bases are consistently showing a trend toward an increase with age of the proportion of adenomas located in the proximal colon. Several observational studies analysed the site distribution of adenomas, testing the hypothesis of a proximal shift of these lesions, and most of them reported an increase in the proportion of right-sided adenomas over time, although a similar trend was not confirmed by others. Also the quality of the retrieved evidence was low. Both endoscopy and FIT are showing a different level of sensitivity for lesions arising in different colonic segments, depending also on screenees characteristics.

CONCLUSION

Available evidence is supporting the hypothesis of an increase in the proportion of right-sided adenomas with age, while a similar increase has not been reported for SSP/A, at least among subjects aged 50 or older. The trend toward a proximalization of colorectal adenomas over time, reported by some authors, likely results from improved diagnostic performances and/or the process of population ageing.

Identifying the optimal strategy for screening of advanced colorectal neoplasia

BACKGROUND AND AIM

To guide selection of optimal screening strategies for advanced colorectal neoplasia (ACRN), we evaluated the screening performances of sigmoidoscopy, fecal immunochemical test (FIT), and the Asia-Pacific Colorectal Screening (APCS) system, as well as combinations of these techniques.

METHODS

We reviewed the records of participants who had undergone a colonoscopy and FIT as part of a comprehensive health-screening program. Results of sigmoidoscopy were estimated from lesions under colonoscopic examination. The screening strategies were classified as sigmoidoscopy-based, FIT-based, and APCS-based. The sensitivity and reduction of number needed to colonoscope (NNScope) of each screening strategy were assessed.

RESULTS

Of 23 023 participants, 398 (1.7%) had ACRN. To predict the presence of ACRN, sigmoidoscopy was the most effective method (odds ratio [OR]: 22.96, 95% confidence interval [CI]: 18.26-29.07), followed by FIT (OR: 7.18, 95% CI: 5.28-9.68) and APCS (high risk [≥4] OR: 2.59, 95% CI: 1.87-3.58; moderate risk [2-3] OR: 1.66, 95% CI: 1.28-2.17). Regarding sigmoidoscopy-based strategies, "sigmoidoscopy positive alone" and "sigmoidoscopy positive or APCS ≥4" were optimal strategies (reduction of NNScope: 79-86%; sensitivity: 73-78%). In the case of FIT-based strategies, "FIT positive or APCS ≥4" was the optimal strategy (reduction of NNScope: 72%; sensitivity: 37%). With regard to the APCS system only, "APCS ≥2" and "APCS ≥4" were recommended as sensitivity-preferred and reduction in NNScope-preferred strategies, respectively.

CONCLUSIONS

The performance in ACRN prediction differs among screening methods (sigmoidoscopy, FIT, and APCS). Among the various single or combined screening strategies, several specific strategies can be chosen depending on medical resources, financial budgets, and screenee preference.

Diagnosis reliability of combined flexible sigmoidoscopy and fecal-immunochemical test in colorectal neoplasia screening

Background

Employing colonoscopy, the gold standard in colorectal cancer (CRC) diagnosis testing, for CRC screening presents a significant risk of complications. Alternative methods with a lower invasive-level and fewer risks are proposed in combination, though each with lower diagnosis performance when applied separately. The main objective of this cross-sectional pilot study was to evaluate the feasibility of a CRC screening program using combined flexible sigmoidoscopy and fecal-immunochemical test (FIT).

Methods

The patient population consisted of 2,201 consecutive-case symptomatic patients attending the gastroenterology outpatient clinic with mild complaints between 2012 and 2014. They were referred for FIT. A sample of 252 individuals underwent a subsequent colonoscopy, blind to FIT results, and theoretical sigmoidoscopy was simulated. On a subsample of 57 patients, real sigmoidoscopy was additionally performed. Prior probabilities in terms of patients’ compliance and CRC prevalence were estimated, together with predictive ability of FIT and sigmoidoscopy in screening population. We assessed the merit of a screening strategy employing two-stage serial multiple testing: a) first stage by combining two parallel tests, that is, flexible sigmoidoscopy and FIT and b) colonoscopy as the second diagnosis test. The scheme was validated using the actual predictive values derived from the study population.

Results

Colonoscopy found 75 (29.76%) individuals with advanced neoplasia. FIT was positive in 30.3% of advanced neoplasia cases, while between 23.73% and 28.28% met the theoretical sigmoidoscopy simulation criteria, with good concordance between real and theoretical sigmoidoscopy. The colonoscopy referral compliance rate was 52% among FIT-positives. Sensitivity and specificity of the first-stage test combination were better than sigmoidoscopy alone (McNemar test: P<0.001). Negative predictive values for low prevalence levels were between 81.5% and 90.12%.

Conclusion

Combining less resource challenging and less invasive testing procedures is worthwhile in colorectal neoplasia detection, improving sensitivity and specificity of either test alone, and leading to better posterior probabilities in usual screening scenarios.