Screening and prevention of colorectal cancerBMJ 2021; 374 doi: https://doi.org/10.1136/bmj.n1855 (Published 15 September 2021) Cite this as: BMJ 2021;374:n1855
- 1Division of Gastroenterology, University of Utah, Salt Lake City, UT, USA
- 2Department of Internal Medicine, University of Utah, Salt Lake City, UT, USA
- 3Huntsman Cancer Institute, University of Utah, Salt Lake City, UT, USA
- Correspondence to P Kanth
Mortality from colorectal cancer is reduced through screening and early detection; moreover, removal of neoplastic lesions can reduce cancer incidence. While understanding of the risk factors, pathogenesis, and precursor lesions of colorectal cancer has advanced, the cause of the recent increase in cancer among young adults is largely unknown. Multiple invasive, semi- and non-invasive screening modalities have emerged over the past decade. The current emphasis on quality of colonoscopy has improved the effectiveness of screening and prevention, and the role of new technologies in detection of neoplasia, such as artificial intelligence, is rapidly emerging. The overall screening rates in the US, however, are suboptimal, and few interventions have been shown to increase screening uptake. This review provides an overview of colorectal cancer, the current status of screening efforts, and the tools available to reduce mortality from colorectal cancer.
Colorectal cancer (CRC) remains a major health burden with high mortality throughout the world. Globally, there were 1.8 million cases and 880 792 deaths from CRC in 2018.1 An increase in rates of CRC in individuals under 50 has been recently described. Most cases of CRC are sporadic, whereas 20-30% are familial. Inherited causes (with known genetic mutation) contribute up to 6-10% of all cases.23 CRC mortality is preventable and the goal of this review is to provide a comprehensive understanding of the screening and prevention efforts in controlling CRC incidence and mortality. The review provides an overview of the worldwide prevalence, risk factors, precursor lesions, and pathways of CRC development, followed by discussion of available screening modalities, colonoscopy quality metrics, and chemoprevention of CRC.
We identified literature through a search of PubMed, Medline, and Embase from 1980 to 2021. Key words searched included “colon cancer”, “colorectal cancer”, “screening”, “prevention”, “FIT”, “FOBT”, and “colonoscopy”. We selected articles published in peer review journals with high impact factor, and by reputed societies. Whenever possible we chose large randomized controlled trials (RCTs) published in high impact and peer review journals over retrospective studies. If no RCTs were available then similar criteria were used for observational studies.
National and international clinical guidelines were included as indicated.
Epidemiology of colorectal cancer
In the US, an estimated 149 500 new cases of CRC occurred in 2021, and 52 980 deaths.45 It is the third and fourth most common cancer diagnosed in the US and the UK, respectively, and is the second leading cause of cancer related deaths in both men and women in both the countries.67 While the incidence and mortality of CRC have been declining in some countries in North America, Australia, and northern Europe, an opposing trend of increases is seen in parts of Asia and South America.8910Figures 1 and 2 depict the incidence of CRC and mortality among men in selected countries since 2000.11 CRC incidence appears to increase with a rise in the human development index of a country, and as less developed countries adopt a more western lifestyle.12
The incidence rate of CRC varies around the world. The age standardized incidence rate (ASRi) is 38.7 per 100 000 persons in the US (2012-16) and it ranges from 20 to 40 per 100 000 persons in Europe (2015).1413 The ASRi for Asia varies widely and is as low as 4.4 (India) to as high as 44.5 (Korea). The overall ASRi for African countries and Latin America is around 8.0 and 16.8 per 100 000 persons, respectively.114 In general, CRC remains a major health and socioeconomic burden in the developed world. Screening for CRC is largely opportunistic in the US, compared with many European countries which have implemented an organized approach. Opportunistic systems mostly comprise an ad hoc or fee based service, while population based organized approaches requires measurement and reporting of screening quality at every step of the process.13
Both organized and opportunistic CRC screening have contributed to the declining trend in the incidence and mortality in many countries. Stool based screening modalities such as the fecal occult blood test (FOBT) were recommended by the US Preventive Services Task Force (USPSTF) in 1996.15 Screening colonoscopy was introduced in the US in 1997, and has been a benefit covered by the Center for Medicare and Medicaid Services since 2001. CRC incidence and mortality has declined in the US since these recommendations (fig 1). Modeling studies suggest around a 50% decline in CRC incidence and mortality can be attributed to implementation of screening.1617 Increased uptake of organized screening was associated with a notable decline in CRC incidence and mortality in a large community based study from the US.18 A modeling study from the Netherlands calculated that the introduction of organized screening could lead to a clinically significant decrease in the incidence and mortality of CRC over 20 to 30 years.19
Early onset colorectal cancer
One of the most alarming current healthcare issues is the rise in CRC incidence in individuals aged 20-49.202122232425 Data from the North American Association of Central Cancer Registries from 2012 to 2016 show a 1.8% increase in the annual incidence in individuals under 50.42627 Conversely, incidence of CRC among individuals over 50 has declined during the same period (fig 3). However, it is important to note that the absolute CRC rate for individuals younger than 50 remains low overall (increased from 7.29 to 8.36 per 100 000) compared with those 50 and older (declined from 123.00 to 111.61 per 100 000) from 2013 to 2017 (fig 3). The rise in rates of CRC in the younger population is not limited to the western hemisphere, but is a worldwide phenomenon.9102224
CRC tends to occur mostly in the distal colon (splenic flexure to the rectum) in younger people.2829 Certain risk factors include male sex, and Black or Asian race; however, the cause of the rise of young onset cancer is not completely understood.303132 A minority of these cancers originates from underlying genetic conditions such as Lynch syndrome and other less common genetic mutations.313334 A study of 430 individuals under 50 diagnosed with CRC showed that one in five people carried a known germline mutation associated with CRC. The authors recommended the use of a multigene panel to screen individuals for the presence of these pathogenic mutations in this cohort.35
Most early onset cancers do not have an identifiable genetic cause and are not caused by Lynch syndrome or other hereditary conditions. A major concern is that CRC in many younger individuals is diagnosed at a late stage and follows an aggressive course.2936 The National Comprehensive Cancer Network (NCCN) recommends swift endoscopic evaluation for alarming signs and symptoms such as rectal bleeding, anemia, or change in bowel habits in younger individuals.37 Earlier diagnosis is of utmost importance as shown by a study projecting a statistically significant reduction in mortality from cancer by detecting CRC at an earlier stage.38 NCCN recommends genetic risk counseling evaluation for all individuals younger than 50 who have a CRC diagnosis.39
Precursor lesions and neoplasia pathways
Most CRCs arise from two pre-cancerous polyp subtypes; adenomatous polyps or sessile serrated lesions (previously termed sessile serrated adenoma/polyps).4041 Sessile serrated lesions (SSLs) have been included in the World Health Organization classification of tumors since 2010.42 Pre-cancerous polyps typically have distinct features identifiable by colonoscopy. Adenomatous polyps are usually well demarcated with an elevated appearance and may have a stalk or pedicle, while sessile serrated lesions are flat (non-pedunculated) and typically have a “mucus cap” with indistinct polyp margins43 (figs 4a, 4b, 4c). Because of these features, the likelihood of missing sessile serrated lesions on colonoscopy is higher than with adenomatous polyps. Sessile serrated lesions also have histological features that overlap with features of benign hyperplastic polyps. Many studies have shown high inter-observer variability and poor consensus among pathologists in the accurate classification of sessile serrated lesions4445 that can lead to errors in the recommendations for post-polypectomy surveillance.
These different polyp subtypes lead to cancer development through distinct neoplasia pathways. The adenoma-carcinoma pathway contributes up to 60-70% of all CRCs, while the serrated neoplasia pathway can lead to 15-30% of all CRCs.1946 The adenoma-carcinoma pathway occurs through alterations in the APC and RAS genes and is microsatellite stable.47 The serrated neoplasia pathway associated with sessile serrated lesions is incompletely understood and is thought to occur predominantly via BRAF and CpG-island methylator phenotype (CIMP).464849 Serrated pathway cancers can be microsatellite stable or unstable.49 A separate carcinoma pathway is noted in those arising from Lynch syndrome, which results from defects in the mismatch repair genes and leads to microsatellite unstable cancers.50
Sessile serrated lesions have also been described as a potential factor in the development of post-colonoscopy colorectal cancer (PCCRC) or interval cancers,515253 defined as CRC occurring before the next recommended colonoscopy exam. These are believed to be the result of missed or incompletely removed pre-cancerous lesions. Prior studies have suggested the roles of microsatellite instability (MSI) and CpG-island methylation in interval CRCs, and both phenomena have been implicated in cancers arising from the serrated pathway.54 Suspected serrated pathway cancers (MSS/MSI-low, CIMP-positive, BRAF-mutated, KRAS-wild) are considered to be aggressive and are associated with greater disease-specific mortality compared with MSI-high, non-CIMP, BRAF-wild cancers.55 CIMP-high tumors may also respond poorly to the standard 5-fluorouracil-based adjuvant therapy for CRC.56 Overall, serrated cancers carry a poor prognosis.575859
Risk factors such as age, race, history of inflammatory bowel disease, underlying predisposing genetic mutations, and family history of colorectal advanced polyps or cancer are non-modifiable. However, several modifiable risk factors are associated with the development of colon polyps and cancer. Commonly modifiable lifestyle and environmental risk factors include diet, smoking, alcohol intake, physical activity, and elevated body mass index (BMI). WHO in 2015 reviewed 800 epidemiological studies and attributed an 18% excess risk of CRC development with a high intake of red and processed meat.60 A case-control study of more than 500 000 individuals followed for 10 years showed a decrease in CRC risk among subjects with higher physical activity.61 Adherence to the World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) 2018 lifestyle (diet, nutrition, physical activity) recommendations based on two large prospective datasets that included more than 68 000 women from the Nurses’ Health Study and around 45 000 men from the Health Professionals Follow-up Study followed for over 25 years have been reported. The authors found a significantly lower CRC risk in men who adhered to the lifestyle recommendations, but a weaker association in women.62 Another study that evaluated WCRF/AICR lifestyle scores in two large Italian cohorts reported similar findings.63 A long term lifestyle assessment using data from the Nurses’ Health Study found that smoking, high BMI, and alcohol intake were associated with an increased risk of precancerous colon polyps.64 These potentially modifiable risk factors are clinically relevant and could be addressed during patient-provider visits.
The effect of the gut microbiome in the development of CRC has recently been explored. Intestinal microbiome dysbiosis is associated with several lifestyle factors including diet.65 Several microbes have been associated with CRC risk, the most significant being Fusobacterium nucleatum and enterotoxigenic Bacteroides fragilis.66676869 Research in this area is still preliminary and no specific clinical recommendations suggest influencing gut flora as a means to reduce CRC mortality.
Various screening modalities have been examined to decrease the incidence and mortality of CRC in individuals at average risk.18707172737475 These include stool based options, a semi-invasive method using computed tomography (CT) imaging or capsule endoscopy, and direct visualization of the distal or entire colon with sigmoidoscopy or colonoscopy.76 Screening using stool based or endoscopic options was introduced in the late 1990s and early 2000s in the US.15 The quality of evidence varies for these modalities. Guaiac based fecal occult blood tests (gFOBT) and sigmoidoscopy have been shown to reduce CRC mortality through multiple randomized controlled trials. No controlled clinical trials of colonoscopy as a screening tool have been published; however, its use is supported by prospective cohort and case-control studies.
Fecal immunochemical testing (FIT) has not been tested through randomized controlled trials, but its effectiveness is reasonable to assume since this is a form of fecal occult blood testing with greater sensitivity and specificity for CRC than gFOBT. CT colonography and multi-target stool DNA tests have not been studied in either prospective or retrospective studies to determine whether the incidence or mortality of CRC is reduced through screening; however, their ability to detect CRC has been compared with colonoscopy in several studies.7778
The US Preventive Services Task Force (USPSTF) recommends screening without identifying a preferred option, whereas the US Multi-Society Task Force (USMSTF) provides a tier-based approach and suggests colonoscopy or FIT as the first tier followed by other options.76
Worldwide, in most countries where CRC screening is conducted, screening is recommended to initiate at age 50.79 In the US, USMSTF and the American College of Physicians recommended screening to start at age 50 among individuals at average risk for developing CRC.767980 African Americans are recommended to start screening at age 45 by the USMSTF.76 The most recent CRC screening guideline from the American Cancer Society (ACS) recommends initiation of screening at age 45 for individuals at average risk irrespective of race because of the substantial increase in rates of early onset CRC.81 The recent draft recommendation by the USPSTF also lowers the age at which to initiate screening to 45, although it rated this a grade B level of recommendation while reserving the grade A recommendation for screening between the ages of 50 and 75.82 The USPSTF final recommendations will be released later in 2021. Clinical practice guidelines from the American College of Gastroenterology suggest lowering the screening age to 45 as a conditional recommendation based on very low quality evidence.83 Both the ACS and USPSTF recommendations are based on the epidemiology of CRC in addition to computer simulation modeling analyses84; however, no studies confirming the effectiveness of CRC screening in individuals between the ages of 45 and 49 (inclusive) have been published. Other individuals who have an increased risk of CRC development are recommended to begin screening at an earlier age, including those with a first degree relative who developed CRC. Screening in this population is recommended to begin at age 40, or 10 years before the youngest age of CRC diagnosis in the family.76
The age at which screening is initiated, and the intervals between successive screening examinations differ for individuals with inflammatory bowel disease (ulcerative colitis and Crohn’s colitis), a family history of CRC, or carriers of genetic mutations, because of the increased lifetime CRC risk in these cohorts.397685868788
Strong recommendations (for example, grade A USPSTF) for CRC screening between the ages of 50 and 75 are consistent among all society guidelines. In general, most societal recommendations advise against screening after 85. Screening should be tailored in individuals between age 75 and 85: individuals are recommended to undergo screening if they have not previously undergone screening, or if their estimated life expectancy is 10 years or more.7679 A study evaluating screening in the age group ≥85 years showed minimal gain (15%) in life expectancy and suggests careful evaluation of risks and benefits in much older people.89 A summary of screening recommendations by various US societies is shown in table 1.
Screening rates and disparities
Screening in the US is mostly opportunistic, whereas certain European countries have taken an organized, population based approach. Although CRC screening rates in the US have been gradually increasing, the proportion adherent to guideline recommendations is still below 70% in most geographical areas.90 The National Colorectal Cancer Roundtable of the American Cancer Society’s initiative of screening goal of “80% by 2018” has now changed to “80% for every community” since this goal was not obtained in that year.91 This adherence target was based on a computer simulation study that showed if the screening rates increased to 80%, 277 000 new cases and 203 000 deaths from CRC would be avoided by 2030.92 Poor education of patients and providers is among the major barriers to implementing widespread screening.9394 Racial disparities in CRC outcomes are based on deficits in access, education, and advocacy for screening.949596 The risk of post-colonoscopy CRC has been documented to be higher in black people than in white people.97 Race disparities are also observed in CRC treatment and survival, with Black individuals having reduced rates of survival.9899 Many socioeconomic barriers exist, including lack of medical insurance, access to care, and affordability of care.96100 No evidence suggests that racial differences in genetics or other immutable factors are responsible for the cancer outcome disparities, and a study from the US Veterans Administration health system showed that CRC mortality did not differ between black and white people in the cohort studied, showing that if insurance and access to healthcare are equivalent, racial disparities are mitigated.101
Identification and surveillance in hereditary colorectal cancer syndromes
This review focuses on screening among individuals at average risk for developing CRC; however, 6-10% of all CRCs arise from inherited CRC syndromes and clinicians should be familiar with familial adenomatous polyposis (FAP) because of its striking clinical presentation (FAP presents with many colon polyps and is easy to recognize on colonoscopy) and Lynch syndrome since it is one of the most common inherited gastrointestinal cancer syndromes.2 Other uncommon polyposis conditions contribute to the familial and inherited causes of CRC (table 2).102 FAP is a polyposis syndrome caused by a mutation in the APC gene that has a 100% lifetime risk of CRC in the absence of colectomy.3103 The prevalence of FAP is around one in 10 000 individuals and it is easily recognizable on colonoscopy that reveals hundreds to thousands of adenomatous polyps.103 Gardner syndrome is a historical term for the FAP spectrum polyposis, with extra-intestinal presentations such as osteomas and benign skin lesions.
By contrast, Lynch syndrome is a non-polyposis condition owing to mismatch repair gene defects leading to MSI-high colorectal (lifetime risk up to 80%) and extra-colonic cancers (table 2). Three per cent of all CRCs are from Lynch syndrome, and with a prevalence of around one in 279-300 individuals, it is considered one of the most common hereditary syndromes.104105 The identification of Lynch syndrome in the general population largely depends on obtaining a detailed cancer history and will be missed if the family history is not obtained. It may be clinically recognized using criteria such as Amsterdam II, Revised Bethesda, or predictive models, with many relatives presenting with young onset cancer involving different organs106107108 (table 3). Thus, obtaining a thorough family history with referral to genetic counseling is highly recommended in individuals with personal or family history of early onset CRC.39109 Interestingly, in the era of multigene panel testing, more Lynch mutations are being diagnosed in individuals who do not fulfill clinical (Amsterdam) criteria.110 Patients with FAP, Lynch syndrome, and other rare polyposis syndromes require intense surveillance throughout their lifetimes.39 Screening should begin 10 years before the age at which the earliest CRC was diagnosed and colonoscopy is the only recommended screening modality.
Stool based screening tests
Stool based options include gFOBT, FIT, and multi-target stool DNA that includes FIT.7377111112113114115 The main principle of the stool based method is that CRC causes occult bleeding that may be detected in stool. The benefits of stool based screening are the non-invasiveness, low cost, and the ability to perform these tests without leaving home. gFOBT detects pseudo peroxidase activity and is an indirect measurement of hemoglobin in the stool. It requires using three stool cards, and certain diets and medications may cause false positive results.116117 The single-test sensitivity of Hemoccult Sensa (high sensitivity FOBT) for CRC is up to 76% with a specificity >85%.115118119 Multiple randomized clinical trials have shown the benefit of using fecal based screening in reducing CRC mortality.7374120121 The Minnesota colon cancer control study has shown a sustained reduction in CRC mortality for individuals undergoing fecal based screening.72122 These clinical trials show the effectiveness of programmatic screening using annual or biennial screening.
FIT directly measures human hemoglobin in the stool with greater sensitivity and specificity than gFOBT. Most guidelines recommend annual fecal testing for the detection of CRC, although the American College of Physicians recommends gFOBT or FIT every two years instead of annually. This is based on a study that compared annual and biennialFOBT with no difference in mortality in the screening cohort.80122 FIT has several advantages over FOBT including the recommendation to use one instead of three stool samples, and absence of dietary restrictions before the test. Several studies have compared FIT with gFOBT and colonoscopy.123124125126127128129 A meta-analysis of 19 studies from 1996 to 2013 in the screening population showed an overall pooled sensitivity and specificity of 79% and 94% for FIT in the detection of CRC.130 Many qualitative and quantitative FIT kits are available worldwide with variable threshold values of hemoglobin in feces for the detection of CRC.115 The sensitivity to detect CRC varies considerably based on the threshold value used.113115131132 Most FIT brands have >20 µg Hb/g feces as threshold value with sensitivity ranging from around 50% to 88%.115133134 Greater than 87% sensitivity was noted for different brands when a threshold of ≥10 µg/g was used.127135136 A study comparing nine quantitative FIT brands with a threshold from 2 to 15 µg/g showed a sensitivity ranging from 56% to 81% and specificity up to 97% in detecting CRC.113 A study from Taiwan compared two FIT brands with the same threshold of 20 µg/g in more than 900 000 individuals undergoing screening and noted variable sensitivity of 68% versus 80%.112 Another meta-analysis of 31 studies with more than 120 000 participants reviewed 18 FIT brands. A threshold value of ≥10 µg/g showed a 91% sensitivity when compared with 71% sensitivity for brands using a threshold of ≥20 µg/g.114
The multi-target stool DNA test (commercially sold as Cologuard) was introduced in 2014 and detects hemoglobin and DNA shed in the stool (methylation markers BMP3, NDRG4, mutant KRAS, and β-actin).77 This test is approved by the US Food and Drug Administration (FDA) and is recommended for screening every 1-3 years.7679 The sensitivity to detect CRC is 92% with a specificity of around 85%. Eighty per cent of the test detection is determined by FIT and 20% is contributed by the DNA markers. The advantage of a multi-target stool test is less frequent screening compared with FIT or gFOBT, with higher sensitivity in single-sample CRC detection. The disadvantages are higher cost than FIT and lower specificity that leads to unnecessary colonoscopies.
Stool options may also detect pre-cancerous adenomas, but the detection rate remains low with most studies describing a sensitivity around 50% or less.114123124125133137 The multi-target stool DNA test has higher sensitivity in detecting advanced adenomas compared with FIT (46% versus 27%).138 Sessile serrated lesions have the lowest detection rate using stool modalities. A study comparing a multi-target stool DNA test with FIT detected only 7% of SSLs in the FIT cohort and 40% in the multi-target stool screening cohort.138 Another FIT study of more than 1800 individuals from Australia showed a sensitivity of 16% in detecting SSLs with a 20 µg/g cut-off threshold.139 Another study of around 6000 participants with different FIT cut-offs of 10, 15, and 20 µg/g had the best sensitivity of 18% in the detection of SSL compared with 32% for adenomas.140 It is unclear if FIT detection could have differences based on sex and race or ethnicity. One study with more than 600 000 participants showed higher sensitivity for men at a 20 µg/g cut-off.141 Studies comparing races for FIT detection rates are limited and have not shown a difference between White and Black people.142 Sensitivity and specificity of stool tests in CRC and pre-cancerous polyp detection are summarized in table 4.
FIT has a better participation rate than gFOBT, likely because of the advantages mentioned above. A study from Scotland compared initiating a FIT screening program with a participation rate of 63% for FIT compared with 56% for FOBT.143 A prior study evaluating the role of small value financial incentives failed to show increased participation in FIT screening.144 Adherence to colonoscopy was lower, but increased participation was noted when patients were offered both stool-based and colonoscopy screening as options.145
Any positive stool test needs a follow-up colonoscopy and the clinical impact of time to colonoscopy after positive stool based screening has been studied.146147 Prior US studies reported that a greater than six month delay in colonoscopy following a positive stool test increased the rate of cancer mortality.148149 Similarly, a Taiwanese study showed a higher risk of any CRC and stage III-IV colon cancer if colonoscopy was performed more than six months after a positive FIT.146 A microsimulation model study calculated a decrease of 10% in the overall screening benefit when follow-up colonoscopy was delayed by 12 months.147 The reasons for failure to undergo colonoscopy after positive stool screening tests are several.150151 Patient factors such as not showing up to a scheduled pre-procedure visit or the colonoscopy appointment itself are the most common reasons; however, provider factors, including not ordering a colonoscopy after a positive FIT, or system factors such as the absence of a tracking system to identify all positive FIT, also frequently contribute.150
Indirect visualization screening tests
CT colonography (CTC) or virtual colonoscopy is a semi-invasive modality to evaluate for CRC and has been endorsed by the USPSTF and the USMSTF for evaluation every five years.7679 The sensitivity varies from 67% to 94% for lesions larger than 10 mm.7879 The negative aspects associated with this modality are high cost, incidental lesions requiring more resources for evaluation with minimal overall benefit, and unavailability of an infrastructure to perform colonoscopy the same day as CTC in most centers, which requires a second bowel preparation for a follow-up colonoscopy if a polyp is detected.
Colon capsule endoscopy
A newer non-invasive modality that has been described is colon capsule endoscopy (CCE). This is a small capsule that is swallowed after bowel preparation through which multiple pictures of the colon are taken.152153 The second generated CCE called Pillcam-2 was introduced in 2009 and has a sensitivity and specificity of 88% and 82% for lesions larger than 6 mm.153154155 CCE has been approved by the FDA for patients who had incomplete colonoscopy for reasons technical or other. CCE has not been recommended for screening by any of the national societies. Some of the drawbacks include a potentially high cost, lack of reimbursement, and low sensitivity for detection of sessile serrated lesions.
Blood based screening tests
Blood tests to detect early CRC and pre-cancerous polyps are highly desired to increase uptake and reduce the harms and costs of screening. A blood test detecting methylated septin 9 DNA is an FDA approved test and is currently available for screening in individuals who decline other screening modalities.156 A recent meta-analysis of 19 studies in the screening population described a pooled sensitivity and specificity of 69% and 92% in the detection of CRC.157 This is not endorsed by USMSTF or USPSTF because of its lower sensitivity, cost, and risk of missing high risk lesions.7679 Other potential blood based markers such as detection of circulating tumor DNA (ongoing ECLIPSE trial; ClinicalTrials.gov identifier: NCT04136002) or multiomics blood based marker (ongoing PREEMPT CRC trial; ClinicalTrials.gov identifier: NCT04369053) are being studied. Other avenues, such as use of volatile organic compounds detected by smell, are being explored, although these may take many years to be adopted in clinical practice.158
Direct visualization screening tests
Flexible sigmoidoscopy as a screening tool has been shown to enable a substantial reduction in deaths from CRC.159 However, colonoscopy remains the most commonly used modality for early detection and prevention of CRC in the US. Removing pre-cancerous polyps, especially advanced lesions, and diagnosing CRC at the earliest stages are advantages of this direct visualization tool.160 Although cohort studies suggest that colonoscopy may substantially decrease CRC incidence and mortality, no randomized clinical trials confirm effectiveness.5170161 Poland’s national CRC screening program reported a reduction in CRC incidence of 72% and mortality by 81% for individuals undergoing colonoscopy compared with the general population during a follow-up of more than 17 years.70 At least four international randomized clinical trials are currently comparing colonoscopy with stool based screening or no screening: CONFIRM (US) that compares annual FIT with colonoscopy over 10 years; NordICC (Nordic countries, Netherlands, and Poland) that compares one-time colonoscopy with no screening; COLONPREV (Spain), and SCREESCO (Sweden) that compare FIT with colonoscopy. The results of these trials will not be available for several years.
Colonoscopy as screening is a tier-one recommendation by the USMSTF along with FIT testing.76 Colonoscopy is not perfect, however, and risks missing early and advanced lesions leading to PCCRC or interval CRC.53162163 The prevalence of PCCRC has been estimated to range from 3.7% to 8.6%.163164 A meta-analysis of 43 studies and more than 150 000 tandem colonoscopies showed a missed rate of almost 25% for adenomas and sessile serrated lesions.162 Colonoscopy quality metrics, tools, and practices have improved considerably in the last decade and many older studies may reflect higher miss rate when compared with recent studies. Some other reasons for the interval CRCs could include the poorly understood serrated polyp biology, incomplete resection of polyps, or poor quality colonoscopy.52165166167 Colonoscopy carries low but important risks, such as bleeding and colonic perforation, and risks associated with sedation administered during colonoscopy procedures, with the rate of bleeding or perforation of 0.28% and 0.05%-0.06%, respectively.168169170 Evidence supporting screening strategies for stool based and direct visualizations tests are summarized in tables 5(a) and 5(b).
Interventions to increase rates of screening
Adherence to screening is low despite the evidence that screening reduces CRC mortality. A survey study of 2500 individuals from five European countries who had never undergone colonoscopy showed that more than half of the respondents did not know colonoscopy is used as a tool for screening and prevention.171 Interventions to increase CRC screening adherence include outreach, navigation, education of patients or providers, reminders, and financial incentives. Outreach is defined as the active dissemination of screening away from the primary care setting, such as distribution of FIT by mailing kits to individuals’ homes, or calling patients to schedule a screening colonoscopy. Navigation employs trained personnel to guide an individual through screening, including obtaining and performing the test, interpreting test results, and scheduling a follow-up colonoscopy if the screening test is positive. Patient education may take the form of brochures, videos, and websites and may include features such as motivational interviewing (directed, patient-centered counseling designed to enable behavior change through resolving ambivalence), cancer risk stratification, or decision aids to identify a patient’s preferred screening strategy. Patients trust the education from their primary care provider, and this trust is highly predictive of adherence to screening.172 Provider education often consists of reports of screening rates in their patient panel with a comparison with their peers. Reminders to healthcare providers can be in the form of electronic health record dashboards that highlight patients due for screening. Financial incentives may consist of monetary incentives for patients to complete screening tests, either as fixed payments or via a lottery such as a one in 10 chance to win $50 after undergoing screening. Individual preference for specific screening strategies has also been examined in several trials, with a choice of screening strategies (annual FIT or colonoscopy) being more effective than recommending a single strategy.173
Clinical trials have shown that the two most effective means of increasing CRC screening rates are outreach with mailed fecal blood tests (relative risk (RR), 2.26; 95% confidence interval (CI) 1.81 to 2.81) and patient navigation (RR, 2.01; 95% CI, 1.64 to 2.46), both of which have been shown to double the rate of screening uptake compared with usual care.174 These studies have highlighted the ability of outreach to increase screening rates irrespective of sex, race, and ethnicity, patient insurance status, or use of other interventions. Navigation increases screening uptake, and results are similar when using lay or peer navigators as when navigation is provided by healthcare professionals. Moreover, screening uptake is increased when navigation is combined with outreach.173 Compared with usual care, participants randomized to receive outreach and navigation had 31% greater time during which they were up to date with CRC screening (incidence rate ratio 1.31; 95% CI, 1.25 to 1.37).
Quality measures for colonoscopy
Colonoscopy has undergone significant changes and improvements in the past decade. High quality colonoscopy is critical for optimal detection and removal of pre-cancerous lesions, and is essential for prevention of post-colonoscopy colorectal cancer.53175 Multiple quality measures have been developed by gastrointestinal societies.176177178 Some of the candidate quality measures include the adenoma detection rate (ADR), cecal intubation rate, and correct recommendation for post-polypectomy surveillance frequencies. The ADR is a calculation of the proportion of individuals at average risk for development of CRC in whom at least one adenoma is detected and removed during a screening colonoscopy. The minimum ADR recommended for men and women is 30% and 20%, respectively.176 The ADR has been extensively studied and low ADRs have been associated with increased risk of CRC after colonoscopy in multiple studies.179180181 Unfortunately, ADRs may not be a perfect benchmark of quality colonoscopy. They do not measure “procedure laxity” that may occur after an endoscopist finds the first polyp in a screening colon, and the ensuing dip in concentration that may lead to missing additional polyps. One study observed that endoscopists with adequate ADRs could still have low detection rates for advanced adenomas.182 ADR, however, remains a major quality measure, and improvements in ADRs have been associated with decreased CRC mortality.180 ADRs can be improved with feedback and performance reports cards to providers performing colonoscopies. This was shown in a study of 20 endoscopists performing more than 12 000 screening colonoscopies who were provided report cards of their ADRs, which led to an improvement in ADR of 3% after a two year follow-up.183
Sessile serrated lesion detection rate (SDR) has been described, but is not included in the ADR calculation and a consensus benchmark for SDR has not been established.184185 Results from a large colonoscopy registry from New Hampshire (more than 29 000 screening colonoscopies) from 28 facilities and 77 proceduralists suggested a SDR of at least 7% as a benchmark.186 Studies related to SDR however, have shown variable detection results among endoscopists.187188 The morphology of SSLs may contribute to their miss rate during colonoscopy since most are flat lesions with indistinct margins that may be difficult to detect. Furthermore, high inter-observer variation is seen among pathologists in defining SSLs, and many SSLs are misdiagnosed as hyperplastic polyps and are not included in the SDR. It is also not clear if having a high ADR suggests equally appropriate sessile serrated lesion detection for endoscopists. The New Hampshire registry study showed 25% of endoscopists with adequate ADR did not meet the 7% detection rate for SDR. Another recent study of 3000 colonoscopies performed by 26 endoscopists from five US centers showed a low SDR for many providers despite their having an adequate ADR.189
Other metrics have been described, such as adequate bowel preparation, documentation of a bowel preparation scoring system, reaching cecum in greater than 90%-95% of the time, evaluating proximal colon diligently either by retroflexion of the scope or taking “two looks” in this segment.176 Cecal intubation and scope withdrawal time are now measured at many centers and an average withdrawal time of ≥ 6 minutes has been shown to detect more pre-cancerous lesions. Some data support the efficacy of mechanical tools (such as Endocuff or EndoRings) which have been studied for better mucosal visualization, and have aided an increase in detection of adenoma.190 The use of high definition scopes over standard definition has been shown to help in adenoma detection and is now being preferred at many facilities.191 Chromoendoscopy (use of dyes) in increasing detection of polyps has been studied in the screening population with some benefit.192 Most studies, however, compared chromoendoscopy with standard definition scopes and not high definition colonoscopy, which is currently the standard for use in screening colonoscopy. The British Society of Gastroenterology position statement on serrated polyps in the colon and rectum suggests a minimum withdrawal time of six minutes and chromoendoscopy to enhance serrated lesion detection, although both are weak recommendations with low quality evidence.193
Chromoendoscopy has not been recommended by most US guidelines194 or has not been addressed.768081 The main drawback has been an increase in the procedure time, without clinically relevant benefit, that may preclude incorporation in busy endoscopy practices.
The use of artificial intelligence (AI) in endoscopy is being explored to increase polyp detection, and endoscopically differentiate pre-malignant lesions from hyperplastic polyps.195196197 Some forms of AI utilize real time computer image analysis to assist the endoscopist to increase polyp detection.198199 One meta-analysis of six randomized controlled trials, however, did not show an increased rate of advanced adenoma detection using AI.200 Data also remain unclear and unexplored in the detection of SSLs. Potential benefits of implementing AI include standardization of endoscopy procedure documentation, making it more automated, and improved data utilization for research.201 This technology is also being used for real time pathology diagnosis to identify non-pre-cancerous diminutive lesions that do not need to be removed, thereby reducing pathology costs.197 The daily use of AI in clinical practice, however, remains elusive. Larger, more generalizable data are required to confirm whether reduction in CRC mortality is gained by using artificial intelligence.
One of the greatest quests in the prevention of CRC is to discover effective chemotherapeutic agents. Aspirin has been the most widely studied in this domain. Chemoprevention with aspirin is effective only if taken for more than 10 years; with this, aspirin has been shown to decrease CRC incidence by 20% and mortality by 33% over 20 years.202203204 Aspirin has been recommended for CRC prevention in LS.39205206 Data from the CAPP2 trial show that regular long term use of aspirin decreases CRC risk in Lynch syndrome.206 The major issue is to identify the minimal effective dose, as the initial dosage used for CAPP2 was 600 mg, high enough to cause an increased risk of gastrointestinal bleeding. The ongoing CAPP3 trial is exploring optimal dosage for CRC prevention in this cohort. Similarly, other non-steroidal anti-inflammatory drugs (NSAIDs) have been explored in CRC prevention and have been shown to reduce risk when taken for longer than five years.207 As with aspirin, however, NSAIDs are not recommended in clinical practice, as the risk-benefit ratio may not be clear in the average risk screening cohort. Vitamin D deficiency has been linked with CRC, but supplementation is currently not recommended for the average risk screening population.208
Multiple screening guidelines are offered by various societies for colon cancer prevention. Most of these follow USPSTF and GI-MSTF recommendations. They are enumerated in table 1.
Prevention and screening remain priorities for countries with high CRC incidence and mortality. Several strategies for screening have been shown to reduce cancer mortality, including stool based and direct visualization tests. Epidemiological studies have shown that screening has contributed to the observed decrease in CRC incidence. However, uptake of screening remains suboptimal in the US and barriers to screening are associated with racial, ethnic, and socioeconomic disparities in CRC outcomes.
It is hoped that advances in understanding the genetic basis of sporadic and familial CRC will provide greater opportunities for precision diagnostics that allow for low cost, non-invasive screening tests. Chemoprevention is emerging as a viable cancer control strategy; however, the effectiveness is insufficient to preclude the need for population screening. Quality of screening, including colonoscopic detection and removal of pre-cancerous polyps, and accurate classification of cancer risk are current areas for improvement that will lead to reductions in cancer mortality. At present, however, it is clear that public health implementation of good screening practices with a focus on awareness and education for patients and providers is needed to achieve the goal of reducing death, disparity, and morbidity from CRC.
Questions for future research
How can screening uptake in the community be increased (with a goal of at least 80% of the population up to date with screening)?
What are the genetic and environmental mechanisms of the serrated pathway to colorectal cancer?
Can artificial intelligence improve quality of colonoscopy?
What role does the intestinal microbiome play in colorectal cancer development, and can cancer risk be reduced by changing the microbial environment?
How can identifying individuals with a genetic predisposition for colorectal cancer be achieved at the population level?
Series explanation: State of the Art Reviews are commissioned on the basis of their relevance to academics and specialists in the US and internationally. For this reason they are written predominantly by US authors
Competing interests: We have read and understood the BMJ policy on declaration of interests and declare the following interests: none.
Patient involvement: no patients were directly involved in the creation of this article.
Provenance and peer review: commissioned; externally peer reviewed.