Abstract
BACKGROUND
Colorectal cancer (CRC) rates in low-resource countries, which typically lack CRC screening programs, are rising. Here we determine whether a risk model for patients with rectal bleeding can identify patients with curable CRC.
METHODS
This prospective cross-sectional study evaluated a model constructed from data from one hospital and validated at two other hospitals. The primary endpoint was ability of the model to predict CRC, as diagnosed by colonoscopy, from clinical characteristics. The secondary endpoint was to determine the percentage of patients who had CRC.
RESULTS
Consecutive patients 45 years of age or older with self-reported rectal bleeding for more than one week were evaluated. From 1/2014-7/2016, 362 patients answered a questionnaire and underwent colonoscopy. In the validation cohort, 56% of patients with rectal bleeding, weight loss, and change in bowel habits had CRC, compared with 2% of patients with bleeding alone.
Overall, 18.2% of patients had CRC and 8.6% had adenomas. The proportion of CRC patients with potentially curable stage II or III disease was 74%, compared with a historical rate of 36%. The combination of rectal bleeding with both symptoms significantly predicted CRC in the validation set (odds ratio 12.8, 95% confidence interval 4.6, 35.4; P < 0.001).
CONCLUSIONS
In low-resource settings, patients with rectal bleeding, weight loss, and change in bowel habits should be classified as high risk for CRC. Patients with a high risk score should be prioritized for colonoscopy to increase the number of patients diagnosed with potentially curable CRC.
Keywords: colorectal cancer, cancer screening, colonoscopy, rectal bleeding
Condensed abstract
Colorectal cancer is a growing public health problem in low- and middle-income countries. This study, proving a risk model can identify early-stage patients, is the first prospective screening or surveillance colorectal cancer trial in Sub-Saharan Africa.
INTRODUCTION
Cancer is a major public health issue in most low- and middle-income countries (LMICs). Colorectal cancer (CRC), traditionally thought to exist primarily in high-income countries, is increasing in many LMICs as industrialization and lifestyle changes alter risk factors.1, 2 The short- and long-term outcomes for patients in LMICs with CRC are poor.3, 4 In Nigeria, the location of the present study, we found, prospectively, that patients diagnosed with CRC have a 1-year survival rate of 47% (Alatise, Kingham, unpublished data). As many as 99% of patients with CRC in LMICs present emergently with locally advanced (stage III) or metastatic (stage IV) disease.4 Survival is strongly related to stage at diagnosis: 5-year survival is greater than 90% for patients with stage I CRC, but only 10% for stage IV CRC.5-7 In addition, the cost of treatment rises significantly as the stage of the disease increases.8 Two strategies are commonly used to detect early-stage CRC: population-based CRC screening and early diagnosis strategies in symptomatic patients.9, 10 Screening for CRC leads to a reduction in emergency presentations, a reduction in the proportion of cancers to adenomatous polyps (adenomas), and a shift to diagnosis at earlier stages.11 Screening also allows an increase in the proportion of patients managed with curative intent, even in resource-limited communities.11
Recent modeling studies suggest that CRC screening can be effective in LMICs.12, 13 In Sub-Saharan Africa, however, there are no published prospective studies testing CRC screening methods, no screening programs, and no evidence-based screening guidelines utilizing prospectively collected data. Screening of asymptomatic individuals for microscopic blood in the stool may appear to be cost-effective in LMICs like Nigeria, but this method is challenging because of a high prevalence of non-cancer–related rectal bleeding and limited resources to justify colonoscopies that seldom diagnose CRC.14, 15 An alternative strategy is symptom-based screening to identify high-risk patients, as has been suggested for high-income countries to facilitate the diagnosis of early-stage CRC.16-23 In resource-limited environments, however, there are no prospective data on the outcomes of symptom-based referrals for colonoscopy.
Given the importance of generating local evidence for CRC screening, we conducted a prospective, multicenter, cross-sectional study in 3 academic institutions in Nigeria with facilities for routine endoscopy. The trial consisted of colonoscopy in high-risk patients with rectal bleeding. On the basis of our retrospective colonoscopy data,15 we hypothesized that colonoscopy would demonstrate that 10%-20% of these patients have CRC or pre-malignant colorectal adenomatous polyps.
METHODS
Study Design and Setting
The setting was 3 hospitals from the African Research Group for Oncology (ARGO) consortium in southwest Nigeria: Obafemi Awolowo University Teaching Hospitals Complex (OAUTHC), Ile-Ife, Osun State; University College Hospital (UCH), Ibadan, Oyo State; and University of Ilorin Teaching Hospital (UITH), Ilorin, Kwara State. The human subjects committee of each institution approved the study. Patients provided written informed consent. The study was registered with clinicaltrials.gov as NCT03032874.
The design was a prospective, multicenter, cross-sectional study involving patients with rectal bleeding who underwent complete colonoscopy from January 2014 to July 2016 at the endoscopy units at the participating institutions. We used patients from OAUTHC to constitute a set for a training model to predict cancer risk, and patients from UCH and UITH to validate this model. Since OAUTHC accrued much faster than UCH and UITH, we set a target of at least 75 patients for the validation set and allowed the training set to accrue until the validation enrollment targets were reached. With this strategy, 217 patients were accrued at OAUTHC. While it is difficult to provide power calculations for building a predictive model, we will have approximately 83% power to detect a univariate association with a binary predictor and the outcome of CRC diagnosis. This calculation assumes 5% Type 1 error, and CRC rates of 5% and 22% in the 2 symptom groups created by the binary predictor. At all 3 centers, a questionnaire on demographics and symptoms was completed for each of the patients. No patients dropped out between the time of questionnaire completion and the time of colonoscopy, as they were performed the same day. The questionnaire was based on National Cancer Institute CRC risk factors for patients in high-income countries.24
Before the research began at each center, an update course was conducted for community health workers and primary care physicians who practiced within a 45-minute driving distance from the teaching hospitals. This ensured education about rectal bleeding and CRC for physicians at many of the community health centers and along the patient referral pathway. In addition, educational talks by the authors (O. Alatise, T. Kingham, O. Ayandipo, A. Adeyeye) were broadcast on television and radio in the cities where the studies were conducted.
Eligibility Criteria
Inclusion criteria included: age 45 years or older, rectal bleeding lasting at least 1 week, and having undergone a complete colonoscopy. Exclusion criteria included a history of inflammatory bowel disease, or colorectal polyps, or CRC; or family history of hereditary polyposis syndromes; or hereditary nonpolyposis colorectal cancer; or individual history of colectomy.
Pre-procedure Evaluation and the Study Procedure
A custom-designed questionnaire was used to gather demographic data, rectal bleeding history, and any relevant medical history. This questionnaire was pilot tested in a previous study examining attitudes of rectal bleeding in Nigerian patients.25 All colonoscopies were performed by endoscopists who perform at least 100 procedures per year. Complete colonoscopy was defined as visualization of the cecum, confirmed by the following landmarks: iliocecal valve, ileal intubation, or identifying the appendiceal orifice.26-28 All colonoscopies at OAUTHC were performed by O. Alatise, all colonoscopies at UCH were performed by O. Ayandipo and A. Akere, and all colonoscopies at UITH were performed by A. Adeyeye and M. Bojuwoye. All the procedures were done under conscious sedation. A polyethylene glycol bowel prep was used. Patients who were diagnosed with CRC or polyps underwent biopsy or polypectomy, respectively. Local pathologists evaluated the specimens. ARGO staff, based at OAUTHC, collated completed questionnaires from the other 2 institutions after every 5 patients. Incomplete data sets were highlighted and reported to the respective coordinators. To obtain missing data, the research coordinators contacted patients. The protocol included financial support for patients diagnosed with CRC to undergo surgery, which included coverage of staging scans and surgery (usually less than $600 in total). All cancer patients were staged by computed tomography scan, during surgery, or by a combination of the two.
Data Analysis
The primary endpoint was the ability of the model developed in this study to predict CRC. Secondary endpoints were the percentage of study participants who had CRC diagnosed upon colonoscopy, the percentage with adenomas, the percentage of patients presenting with stage I and II CRC compared to historical controls in Nigeria, and the safety of colonoscopy in the study cohort. Change in bowel habits was coded as any change in bowel habit (yes/no). Unintentional weight loss within the last 6 months (yes/no) was described to patients as looseness of clothes, belts, and wristwatches. Some patients provided objective weight-loss information for the preceding 6 months.
Data were stored in a password-protected database on a secure website. We created the predictive model from the training set by first conducting univariate tests of association where categorical variables were compared using the χ2-test or Fisher’s exact test, while quantitative variables were compared using the Kruskal-Wallis test. Variables significantly associated in this univariate analysis were included in a logistic regression model to build a predictive model. Predictions using this model were obtained on the validation set patients, and the ability of these predictions to discriminate patients with cancer from those without cancer was evaluated using the c-index.
All statistical tests were two-sided, and P values below 0.05 were considered significant. Statistical analysis was carried out using SAS 9.4 and R 3.3.1.
RESULTS
In total, 362 patients were enrolled: 217 (60%) at OAUTHC and 145 (40%) at the other 2 participating institutions (UCH and UITH). The patients at OAUTHC constituted the training set for a prediction model, and those at UCH and UITH were used as a validation set. In the combined cohorts, most commonly, patients saw blood in their stool once or twice per week (n=171; 47.2%; Table 1). Patients frequently believed that hemorrhoids were the cause of their rectal bleeding (n=172; 47.5%). The training and validation sets had no statistically significant differences in median age, sex distribution, or body mass index (BMI), but they had some significant differences in patient beliefs regarding etiology of bleeding and symptoms, such as frequency of blood in stool (Table 1).
TABLE 1.
Patient Characteristics and Symptoms in the 2 Study Groups. Values Are Number (%) or Median (Range)
| Characteristic | Overall (n=362) | Training (n=217) | Validation (n=145) | P value | |
|---|---|---|---|---|---|
| Age, yrs | 59.5 (44, 95) | 60 (45, 95) | 59 (44, 87) | 0.14 | |
| BMI, kg/m2 | 24.5 (12.6, 49.6) | 24.3 (12.6, 49.6) | 25.1 (14.9, 43.3) | 0.089 | |
| Sex | 0.37 | ||||
| Female | 126 (34.8) | 80 (36.9) | 46 (31.7) | ||
| Months since the blood in stool started | 6 (0.25, 588) | 6 (0.25, 588) | 5 (0.25, 324) | 0.53 | |
| Frequency of blood in stool | 0.001 | ||||
| 1-2 per week | 171 (47.2) | 101 (46.5) | 70 (48.3) | ||
| 3-5 per week | 82 (22.7) | 37 (17.1) | 45 (31) | ||
| 6-7 per week | 109 (30.1) | 79 (36.4) | 30 (20.7) | ||
| Patient believes hemorrhoids are the cause of bleeding | 0.024 | ||||
| No | 186 (51.4) | 101 (46.5) | 85 (58.6) | ||
| Yes | 172 (47.5) | 114 (52.5) | 58 (40) | ||
| Missing | 4 (1.1) | 2 (0.9) | 2 (1.4) | ||
| Change in bowel habits | >0.95 | ||||
| Yes | 174 (48.1) | 104 (47.9) | 70 (48.3) | ||
| Type of change in bowel habits | 0.010 | ||||
| Alternating | 50 (13.8) | 23 (10.6) | 27 (18.6) | ||
| Constipation | 25 (6.9) | 17 (7.8) | 8 (5.5) | ||
| Diarrhea | 28 (7.7) | 12 (5.5) | 16 (11) | ||
| Pellet-like stool | 71 (19.6) | 52 (24) | 19 (13.1) | ||
| None | 188 (51.9) | 113 (52.1) | 75 (51.7) | ||
| Weight loss over 6 months | 0.28 | ||||
| Yes | 153 (42.3) | 97 (44.7) | 56 (38.6) | ||
| Family members with CRC | 0.28 | ||||
| No | 330 (91.2) | 206 (94.9) | 124 (85.5) | ||
| Yes | 15 (4.1) | 7 (3.2) | 8 (5.5) | ||
| Missing | 17 (4.7) | 4 (1.8) | 13 (9) | ||
| Seen physician for rectal bleeding | 0.57 | ||||
| Yes | 302 (83.4) | 183 (84.3) | 119 (82.1) | ||
Abbreviations: BMI, body mass index; CRC, colorectal cancer
Only 4 patients (1.1%) reported complication from the colonoscopy procedure. These complications included excessive abdominal distension and severe vomiting from sedatives. None required hospital admission, and all resolved within 24 hours of the procedure.
The source of bleeding, diagnosed by colonoscopy, was most commonly hemorrhoids (n=120; 33.1%), followed by diverticulosis (n=81; 22.4%) and CRC (n=65; 18.0%). In about 13% of patients, the cause of bleeding could not be ascertained by colonoscopy. Most colonoscopy findings were not significantly different between the training and validation groups (Table 2). The overall rate of adenoma detection was 8.6% (n=31) and the overall rate of cancer detection was 18.2% (n=66). Nine adenomas (29%) and 8 CRCs (12%) were found in the right colon. Most CRCs were not metastatic (29% stage II, 45% stage III). The percentage of stage II patients was significantly higher at OAUTHC than at the other 2 centers (42% vs 11%; P = 0.005), but the percentage of stage IV cancers was similar (29% vs 21%).
TABLE 2.
Colonoscopy Findings in the 2 Study Groups. Values are Number (%)
| Finding | Overall (n=362) | Training (n=217) | Validation (n=145) | P value |
|---|---|---|---|---|
| Internal hemorrhoids | <0.0001 | |||
| No | 134 (37) | 33 (15.2) | 101 (69.7) | |
| Yes | 228 (63) | 184 (84.8)a | 44 (30.3)b | |
| Grade of internal hemorrhoids | 0.007 | |||
| 1 | 14 (6.1) | 6 (3.3) | 8 (18.2) | |
| 2 | 165 (72.4) | 137 (74.1) | 28 (63.6) | |
| 3 | 46 (20.2) | 39 (21.2) | 7 (15.9) | |
| 4 | 3 (1.3) | 3 (1.6) | 0 (0) | |
| Polyps | 0.1 | |||
| No | 277 (76.5) | 173 (79.7) | 104 (71.7) | |
| Yes | 85 (23.5) | 44 (20.3) | 41 (28.3) | |
| Adenomatous polyp | 0.44 | |||
| No | 331 (91.4) | 196 (90.3) | 135 (93.1) | |
| Yes | 31 (8.6) | 21 (9.7) | 10 (6.9) | |
| Pathology of polyps | 0.051 | |||
| Adenomatous polyp | 31 (36.5) | 21 (47.7) | 10 (24.4) | |
| Hyperplastic polyps | 6 (7.1) | 1 (2.3) | 5 (12.2) | |
| Inflammatory | 43 (50.6) | 21 (47.7) | 22 (53.7) | |
| Not available | 5 (5.9) | 1 (2.3) | 4 (9.8) | |
| Classification of adenomatous polyps | 0.72 | |||
| Tubular | 16 (51.6) | 13 (61.9) | 3 (30) | |
| Tubulovillous | 9 (29) | 6 (28.6) | 3 (30) | |
| Villous | 1 (3.2) | 1 (4.8) | 0 (0) | |
| Missing | 5 (16.1) | 1 (4.8) | 4 (40) | |
| Grade of the adenomatous polyps | >0.95 | |||
| High | 1 (3.2) | 1 (4.8) | 0 (0) | |
| Low | 30 (96.8) | 20 (95.2) | 10 (100) | |
| Adenomatous polyp location | 0.780 | |||
| Left colon | 21 (67.7) | 13 (61.9) | 8 (80.0) | |
| Right colon | 9 (29) | 7 (33.3) | 2 (20.0) | |
| Transverse | 1 (3.2) | 1 (4.8) | 0 (0) | |
| CRC | 0.68 | |||
| No | 296 (81.8) | 179 (82.5) | 117 (80.7) | |
| Yes | 66 (18.2) | 38 (17.5) | 28 (19.3) | |
| Location of CRC | >0.95 | |||
| Left colon | 58 (87.9) | 33 (86.8) | 25 (89.3) | |
| Right colon | 8 (12.1) | 5 (13.2) | 3 (10.7) | |
| Stage of CRC | 0.003 | |||
| II | 19 (28.8) | 16 (42.1) | 3 (10.7) | |
| III | 30 (45.5) | 11 (28.9) | 19 (67.9) | |
| IV | 17 (25.8) | 11 (28.9) | 6 (21.4) | |
| Pathology of CRC | 0.069 | |||
| Moderately differentiated | 22 (33.3) | 14 (36.8) | 8 (28.6) | |
| Poorly differentiated | 5 (7.6) | 5 (13.2) | 0 (0) | |
| Well differentiated | 39 (59.1) | 19 (50) | 20 (71.4) |
Abbreviations: CRC, colorectal cancer
One patient received 2 grades
One grade on a hemorrhoid in this group is missing
To determine factors that can predict a finding of CRC, we evaluated the association of demographic and clinical characteristics with colonoscopy findings in the training set (Table 3). Patients with CRC had a significantly lower BMI (22.9 kg/m2) than patients without CRC (24.9 kg/m2; P = 0.009). Patients with CRC also had a significantly lower proportion of internal hemorrhoids (61% vs 90%; P < 0.001). Weight loss and change of bowel habits were more common in CRC patients than in those without CRC (P < 0.0001). When building the model on the training set, weight loss (odds ratio 7.9; 95% confidence interval 2.8, 21.8; P < .0001) and change of bowel habits (odds ratio 6.3; 95% confidence interval 2.3, 17.6; P = .0004) were significant predictors. There was also evidence of a synergistic interaction between these variables, though in a traditional multivariate model, the interaction term was not statistically significant. We therefore opted for a univariate model based upon a symptom score, coded as 0, 1, or 2 according to whether the patient had none, one, or both of the symptoms of weight loss or change in bowel habits. Overall, CRC was present in only 1.4% of patients with no weight loss or change in bowel habits, but in 49% of patients with a symptom score of 2 (i.e., both weight loss and change in bowel habits). Symptom score was highly significant in predicting CRC in the validation set with patients with both symptoms compared to no symptoms (odds ratio 12.8; 95% confidence interval 4.6, 35.4; P < 0.001). The concordance index in the validation set was 0.875 (Table 4). When age was added to the final model, the concordance was 0.7467, so its incremental value is not significant. If a symptom score of 2 was used as positive, then specificity for diagnosing CRC was 83% and sensitivity was 89%. Symptom score was not associated with stage (P = 0.50; Table 5). For example, weight loss and change in bowel habits had a similar distribution among stages II through IV.
TABLE 3.
Univariate Analysis of Association of Patient Characteristics with Colonoscopy Findings in Patients in the Training Set. Data Are Number (%) or Median (Range)
| Characteristic | No CRC (n=179) | CRC (n=38) | P value |
|---|---|---|---|
| Age, yrs | 60 (45, 95) | 60 (45, 82) | 0.52 |
| BMI, kg/m2 | 24.9 (12.6, 49.6) | 22.9 (16.6, 30.8) | 0.009 |
| Months since start of bleeding | 6 (0.25, 588) | 6 (0.5, 84) | 0.52 |
| Sex | 0.095 | ||
| Female | 61 (34.1) | 19 (50) | |
| Male | 118 (65.9) | 19 (50) | |
| Frequency of bleeding episodes | 0.32 | ||
| 1-2 per week | 86 (48) | 15 (39.5) | |
| 3-5 per week | 32 (17.9) | 5 (13.2) | |
| 6-7 per week | 61 (34.1) | 18 (47.4) | |
| Any change in bowel habits | <0.001 | ||
| No | 108 (60.3) | 5 (13.2) | |
| Yes | 71 (39.7) | 33 (86.8) | |
| Type of change in bowel habits | <0.001 | ||
| Alternating | 13 (7.3) | 10 (26.3) | |
| Constipation | 16 (8.9) | 1 (2.6) | |
| Diarrhea | 8 (4.5) | 4 (10.5) | |
| Pellet-like stool | 34 (19) | 18 (47.4) | |
| None | 108 (60.3) | 5 (13.2) | |
| Weight loss over last 6 months | <0.001 | ||
| No | 115 (64.2) | 5 (13.2) | |
| Yes | 64 (35.8) | 33 (86.8) | |
| Internal hemorrhoids | <0.001 | ||
| No | 18 (10.1) | 15 (39.5) | |
| Yes | 161 (89.9) | 23 (60.5) | |
| Diverticulum | 0.001 | ||
| Absent | 113 (63.1) | 34 (89.5) | |
| Present | 66 (36.9) | 4 (10.5) |
Abbreviations: BMI, body mass index
TABLE 4.
Results for the Model Created from the Training Cohort to Predict CRC from Symptom Score (Based Upon Weight Loss and Change in Bowel Habits) in Patients with Rectal Bleeding 45 Years of Age or Older (n=145)
| Validation Cohort | |||
|---|---|---|---|
| Symptom Score | No CRC | CRC | Total |
| 0 | 63 (98%) | 1 (2%) | 64 |
| 1 | 34 (94%) | 2 (6%) | 36 |
| 2 | 20 (44%) | 25 (56%) | 45 |
| Total | 117 | 28 | 145 |
| Concordance index: 0.875 | |||
Abbreviations: CRC, colorectal cancer
TABLE 5.
Distribution of Number of Symptoms and Stage of CRC for Patients Diagnosed with Cancer (n=28) in the Validation Set
| CRC Stage | ||||
|---|---|---|---|---|
| Symptom Score | II | III | IV | Total |
| 0 | 1 | 0 | 0 | 1 |
| 0 | 0 | 1 | 0 | 1 |
| 1 | 1 | 1 | 0 | 2 |
| 2 | 2 | 17 | 6 | 25 |
| Total | 3 | 19 | 6 | 28 |
| Fisher P value = .4954 | ||||
Abbreviations: CRC, colorectal cancer
DISCUSSION
Screening offers a means of shifting CRC diagnosis to earlier stages, thus improving the poor outcomes in Sub-Saharan Africa. However, large-scale asymptomatic CRC screening is unlikely to be possible in the near future in this region. We prospectively examined the utility of identifying high-risk Nigerian patients with rectal bleeding, and then evaluating them with colonoscopy. We found that 18.2% of patients over 45 years of age with rectal bleeding of more than 1 week had CRC. For a patient with rectal bleeding, a simple symptom score based upon weight loss and change in bowel habits was associated with CRC, with an odds ratio of 12.8. In the validation cohort, fully 89% of the cancers were in patients with a symptom score of 2, suggesting that this model may have excellent sensitivity among patients with rectal bleeding. Overall, CRC was present in only 1.4% of patients with a symptom score of 0, but in 49% of patients with a symptom score of 2 (i.e., both weight loss and change in bowel habits). These findings demonstrated that in low-resource environments, it is possible to identify patients likely to harbor CRC using a combination of symptoms. Most importantly, 74% of cancers were stage II or III, compared to historical data in Nigeria showing that only 36% of CRCs were stage II or III at diagnosis (Alatise, Kingham unpublished data).
Outside of Sub-Saharan Africa, numerous studies have evaluated patients with rectal bleeding to determine the risk of polyps and CRC in this population. The CRC risks associated with rectal bleeding, weight loss, and change in bowel habits in our study were higher than those reported in most studies outside Sub-Saharan Africa. The positive predictive value of rectal bleeding in our study (18.2%) contrasts with the values reported in a meta-analysis in which values ranged from 2.2% to 16%. The higher positive predictive value seen in our study is likely multifactorial. Many of the patients in our study had other symptoms in addition to bleeding. It is possible that the patients referred for colonoscopy are selected at community health centers or by primary physicians as high risk because of these symptoms or their clinical suspicion for CRC. This stresses the need for a community-based study of asymptomatic patients with rectal bleeding in West Africa, as it is unknown what regional differences may exist for CRC risk in this population. This meta-analysis showed a positive predictive value for rectal bleeding that ranged from 2.2% to 16% in the 13 included studies, and the pooled positive predictive value reached 8.1% in those ≥ 50 years of age. The analysis also revealed a higher risk when rectal bleeding was combined with other symptoms (e.g., pooled positive likelihood ratio of 1.9 for weight loss and 1.8 for change in bowel habits).29 The positive likelihood ratio was higher for our validation cohort (2.32 for change in stool and 3.89 for weight loss). One study at tertiary hospitals in the United Kingdom found 181 CRC patients (6.7%) among 2697 subjects with rectal bleeding, change in bowel habits, and abdominal pain.30
Cancer control plans are currently being generated in many countries in Sub-Saharan Africa. The data presented here provide an important component of CRC control planning. Recognition of the importance of the symptom complex of rectal bleeding, weight loss, and change in bowel habits should lead to systematic education of primary care physicians to refer patients with these symptoms early for surveillance colonoscopies. It is possible this will lead to a stage-migration so that patients with polyps have prevention of CRC and so that patients with CRC are diagnosed at an earlier, more curable stage. Similar programs exist in high-income countries, where, as adjunct to organized CRC screening programs, they reduce morbidity and mortality.16,14,17-22,31 In this context, some practice guidelines, such as the updated National Institute for Health and Care Excellence guidelines in the United Kingdom, recommend urgent referral of all patients 40 years of age and older who report rectal bleeding with a change toward looser stools and/or increased stool frequency persisting for 2 weeks or more.32
The incidence of CRC in most LMICs is lower than in high-income countries,33 but it is rising. While this may justify organized screening for CRC in the future, the use of symptom-based surveillance is an initial step toward reducing mortality from CRC in Nigeria and other LMICs. In addition, given the limited access to colonoscopy in Sub-Saharan Africa, the need for its prioritization cannot be overemphasized. In Nigeria, for example, there are approximately 100 endoscopists (Alatise, personal data).
The argument against the use of symptom-based surveillance for CRC is that these symptoms are common in populations without cancer and so may have poor sensitivity for CRC.34, 35 Hence, evaluating patients at low risk of CRC may lead to unnecessary harm, which could include cost, patient anxiety, and iatrogenic injury.32 However, the high yield of CRC in this study (18.2%) shows that the risk of CRC among Nigerians with a median of 6 months of rectal bleeding may actually be higher than previously estimated and so justify recommending referral of patients much earlier than 6 months into their bleeding so they can undergo colonoscopy surveillance. In addition, the high rate of CRC compared to adenomatous polyps in this patient population again confirms this is surveillance of a high-risk group. It could also be argued that the symptom complex (rectal bleeding, weight loss, and change in bowel habits) connotes late presentation. Our data clearly demonstrate that the symptoms are not unique to late presentation.
In this study, over 80% of the study cohort had consulted a physician before enrollment. Unfortunately, most patients were not originally referred for a colonoscopy. In previous studies in both high-income countries and LMICs, only 28%-41% of patients with rectal bleeding consulted a physician.34, 36-38 A reason for the higher consultation rate for rectal bleeding in the current study is the Nigerian cultural belief that seeing blood from any part of the body is an ominous sign that requires early treatment. This makes rectal bleeding a useful tool for investigating CRC in Nigeria.
In our cohorts, the right colon was the location for about 30% of adenomatous polyps and 12% of CRCs. Most of these cases would have been missed by sigmoidoscopy. This suggests that colonoscopy may be superior to sigmoidoscopy for screening patients with rectal bleeding. Similarly high proportions of right-sided adenomatous polyps and CRCs have also been reported in a single-center study from Iran.16
Strengths of this study are that it is both multicentric and prospective. It begins to fill the massive gap in data to form the basis for evidence-based CRC screening in Africa. The study also has limitations. Some patients were given bowel prep supplies but never returned to be enrolled in the study, so they did not complete a questionnaire or undergo a colonoscopy. These patients were not followed, so the characteristics of these patients are unclear. There was likely selection bias among the patients referred into the study, given the high rates of patients with CRC (18%) and adenomatous polyps (9%). It is possible that the education level of referring physicians had an effect on the profile of patients referred. This pattern was consistent, however, among all 3 cities. In addition, the study did not evaluate the surveillance method for cost effectiveness in LMICs like Nigeria. Despite these limitations, this is the first prospective cross-sectional study to determine the CRC downstaging in patients with rectal bleeding, stratified by the symptoms of weight loss and change in bowel habits.
CONCLUSIONS
The risk model developed and validated in this study has high sensitivity and specificity. It can be used to provide evidence-based recommendations for cancer control plans in Sub-Saharan Africa, with the hope that such usage will lead to an improvement in the poor outcomes for patients with CRC in LMICs.
Acknowledgments
The authors thank Dr. Bernard Levin for his advice and critical appraisal of the manuscript. The authors also thank Dr. Janet Novak for her critical review and editing, African Research Group for Oncology staff members Olawumi Yejide Olajide, Olawale Olalude, and Memorial Sloan Kettering Cancer Center research staff Paula Garcia, Jeremy Constable, and Liana Langdon-Emery.
FUNDING SUPPORT
This study was performed with support from the Thompson Family Foundation, MSK Global Cancer Disparity Initiative, Surgeons OverSeas, and NIH/NCI Cancer Center Support Grant No. P30 CA008748. The funding agencies had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Footnotes
Anatomic site: Gastrointestinal Disease
Invitation indication: Not invited
Trial Registration: clinicaltrials.gov Identifier: NCT03032874 (Community Based Screening for Colorectal Cancer in an Underserved High Risk Population in Nigeria; https://clinicaltrials.gov/ct2/show/)
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.
Conflict of interest information: The authors have no conflict of interest disclosures to report, and this manuscript is not under consideration elsewhere.
AUTHOR CONTRIBUTIONS
Olusegun Isaac Alatise: Conceptualization, methodology, formal analysis, investigation, data curation, writing–original draft, and writing–review and editing.
Omobolaji O. Ayandipo: Conceptualization, methodology, formal analysis, investigation, data curation, and writing–review and editing.
Ademola Adeyeye: Conceptualization, methodology, formal analysis, investigation, data curation, and writing–review and editing.
Ken Seier: Formal analysis, investigation, data curation, writing–original draft, and writing–review and editing.
Akinwunmi O. Komolafe: Formal analysis, investigation, data curation, and writing–review and editing.
Matthew O. Bojuwoye: Formal analysis, investigation, data curation, and writing–review and editing.
Oludapo O. Afuwape: Formal analysis, investigation, data curation, writing–original draft, and writing–review and editing.
Ann Zauber: Conceptualization, methodology, formal analysis, investigation, data curation, and writing–review and editing.
Adeleye Omisore: Formal analysis, investigation, data curation, writing–original draft, and writing–review and editing.
Samuel Olatoke: Formal analysis, investigation, data curation, writing–original draft, and writing–review and editing.
Adegboyega Akere: Formal analysis, investigation, data curation, writing–original draft, and writing–review and editing.
Olusola Famurewa: Formal analysis, investigation, data curation, writing–original draft, and writing–review and editing.
Mithat Gonen: Formal analysis, investigation, data curation, writing–original draft, and writing–review and editing.
David O. Irabor: Formal analysis, investigation, data curation, writing–original draft, and writing–review and editing.
T. Peter Kingham: Conceptualization, methodology, formal analysis, investigation, data curation, writing–original draft, and writing–review and editing.
References
- 1.Arnold M, Sierra MS, Laversanne M, Soerjomataram I, Jemal A, Bray F. Global patterns and trends in colorectal cancer incidence and mortality. Gut. 2017 Apr;66(4):683–691. doi: 10.1136/gutjnl-2015-310912. [DOI] [PubMed] [Google Scholar]
- 2.Torre LA, Siegel RL, Ward EM, Jemal A. Global Cancer Incidence and Mortality Rates and Trends–An Update. Cancer Epidemiol Biomarkers Prev. 2016 Jan;25(1):16–27. doi: 10.1158/1055-9965.EPI-15-0578. [DOI] [PubMed] [Google Scholar]
- 3.Kingham TP, Alatise OI, Vanderpuye V, et al. Treatment of cancer in sub-Saharan Africa. Lancet Oncol. 2013 Apr;14(4):e158–167. doi: 10.1016/S1470-2045(12)70472-2. [DOI] [PubMed] [Google Scholar]
- 4.Saluja S, Alatise OI, Adewale A, et al. A comparison of colorectal cancer in Nigerian and North American patients: is the cancer biology different? Surgery. 2014 Aug;156(2):305–310. doi: 10.1016/j.surg.2014.03.036. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Keane MG, Johnson GJ. Early diagnosis improves survival in colorectal cancer. Practitioner. 2012 Jul-Aug;256(1753):15–18, 12. [PubMed] [Google Scholar]
- 6.O’Connell JB, Maggard MA, Ko CY. Colon cancer survival rates with the new American Joint Committee on Cancer sixth edition staging. J Natl Cancer Inst. 2004 Oct 06;96(19):1420–1425. doi: 10.1093/jnci/djh275. [DOI] [PubMed] [Google Scholar]
- 7.Usher-Smith JA, Walter FM, Emery JD, Win AK, Griffin SJ. Risk Prediction Models for Colorectal Cancer: A Systematic Review. Cancer Prev Res (Phila) 2016 Jan;9(1):13–26. doi: 10.1158/1940-6207.CAPR-15-0274. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Huang HY, Shi JF, Guo LW, et al. Expenditure and financial burden for common cancers in China: a hospital-based multicentre cross-sectional study. Lancet. 2016 Oct;388(Suppl 1):S10. [Google Scholar]
- 9.Scottish Intercollegiate Guidelines Network (SIGN) Diagnosis and management of colorectal cancer (SIGN publication no 126) http://www.sign.ac.uk/guidelines/fulltext/126/index.html. Accessed October 26, 2017.
- 10.Zauber AG, Knudsen AB, Rutter CM, et al. AHRQ Technology Assessments . Cost-Effectiveness of CT Colonography to Screen for Colorectal Cancer. Rockville (MD): Agency for Healthcare Research and Quality (US); 2009. [PubMed] [Google Scholar]
- 11.Mansouri D, McMillan DC, Crearie C, Morrison DS, Crighton EM, Horgan PG. Temporal trends in mode, site and stage of presentation with the introduction of colorectal cancer screening: a decade of experience from the West of Scotland. Br J Cancer. 2015 Jul 28;113(3):556–561. doi: 10.1038/bjc.2015.230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Ginsberg GM, Lauer JA, Zelle S, Baeten S, Baltussen R. Cost effectiveness of strategies to combat breast, cervical, and colorectal cancer in sub-Saharan Africa and South East Asia: mathematical modelling study. BMJ. 2012 Mar 02;344:e614. doi: 10.1136/bmj.e614. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Siripongpreeda B, Mahidol C, Dusitanond N, et al. High prevalence of advanced colorectal neoplasia in the Thai population: a prospective screening colonoscopy of 1,404 cases. BMC Gastroenterol. 2016 Aug 23;16:101. doi: 10.1186/s12876-016-0526-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Adedayo OT, Babatunde AS, Adekoya OA, Nwokoro CC. Rectal bleeding amongst Medical Students: Prevalence and Consultation Behaviour. East and Central African Journal of Surgery. 2011;16(2):146–150. [Google Scholar]
- 15.Alatise OI, Arigbabu AO, Agbakwuru EA, Lawal OO, Ndububa DA, Ojo OS. Spectrum of colonoscopy findings in Ile-Ife Nigeria. Niger Postgrad Med J. 2012 Dec;19(4):219–224. [PubMed] [Google Scholar]
- 16.Bafandeh Y, Khoshbaten M, Eftekhar Sadat AT, Farhang S. Clinical predictors of colorectal polyps and carcinoma in a low prevalence region: results of a colonoscopy based study. World J Gastroenterol. 2008 Mar 14;14(10):1534–1538. doi: 10.3748/wjg.14.1534. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Blumenstein I, Tacke W, Bock H, et al. Prevalence of colorectal cancer and its precursor lesions in symptomatic and asymptomatic patients undergoing total colonoscopy: results of a large prospective, multicenter, controlled endoscopy study. Eur J Gastroenterol Hepatol. 2013 May;25(5):556–561. doi: 10.1097/MEG.0b013e32835d1ef4. [DOI] [PubMed] [Google Scholar]
- 18.Hamilton W, Green T, Martins T, Elliott K, Rubin G, Macleod U. Evaluation of risk assessment tools for suspected cancer in general practice: a cohort study. Br J Gen Pract. 2013 Jan;63(606):e30–36. doi: 10.3399/bjgp13X660751. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Hippisley-Cox J, Coupland C. Identifying patients with suspected colorectal cancer in primary care: derivation and validation of an algorithm. Br J Gen Pract. 2012 Jan;62(594):e29–37. doi: 10.3399/bjgp12X616346. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Leung WK, Ho KY, Kim WH, Lau JYW, Ong E, Hilmi Iea. Colorectal neoplasia in Asia: a multicenter colonoscopy survey in symptomatic patients. Gastrointestinal Endosc. 2006;64:751–759. doi: 10.1016/j.gie.2006.06.082. [DOI] [PubMed] [Google Scholar]
- 21.National Institute for Health and Care Excellence. Suspected cancer: recognition and referral (NG12) https://www.nice.org.uk/guidance/ng12. Accessed October 26, 2017. 27 March 2017.
- 22.Schoepfer A, Marbet UA. Colonoscopic findings of symptomatic patients aged 50 to 80 years suggest that work-up of tumour suspicious symptoms hardly reduces cancer-induced mortality. Swiss Med Wkly. 2005 Nov 19;135(45-46):679–683. doi: 10.4414/smw.2005.11033. [DOI] [PubMed] [Google Scholar]
- 23.Rabeneck L, Horton S, Zauber A, Earle C. Colorectal Cancer. In: Gelband H, Jha P, Sankaranarayanan R, Horton S, editors. Cancer: Disease Control Priorities. Washington, DC: International Bank for Reconstruction and Development/The World Bank; 2016. pp. 101–119. [Google Scholar]
- 24.National Cancer Institute. Colorectal Cancer Risk Assessment Tool. https://www.cancer.gov/colorectalcancerrisk/colorectal-cancer-risk.aspx. Accessed March 15, 2018.
- 25.Alatise OI, Fischer SE, Ayandipo OO, Omisore AG, Olatoke SA, Kingham TP. Health-Seeking Behavior and Barriers to Care in Patients With Rectal Bleeding in Nigeria. Journal of Global Oncology. 2017 Feb 17; doi: 10.1200/JGO.2016.006601. Epub ahead of print. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Rex DK, Petrini JL, Baron TH, et al. Quality indicators for colonoscopy. Am J Gastroenterol. 2006 Apr;101(4):873–885. doi: 10.1111/j.1572-0241.2006.00673.x. [DOI] [PubMed] [Google Scholar]
- 27.Baxter NN, Sutradhar R, Forbes SS, Paszat LF, Saskin R, Rabeneck L. Analysis of administrative data finds endoscopist quality measures associated with postcolonoscopy colorectal cancer. Gastroenterology. 2011 Jan;140(1):65–72. doi: 10.1053/j.gastro.2010.09.006. [DOI] [PubMed] [Google Scholar]
- 28.Brahmania M, Park J, Svarta S, Tong J, Kwok R, Enns R. Incomplete colonoscopy: maximizing completion rates of gastroenterologists. Can J Gastroenterol. 2012 Sep;26(9):589–592. doi: 10.1155/2012/353457. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Astin M, Griffin T, Neal RD, Rose P, Hamilton W. The diagnostic value of symptoms for colorectal cancer in primary care: a systematic review. Br J Gen Pract. 2011 May;61(586):e231–243. doi: 10.3399/bjgp11X572427. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Tong GX, Chai J, Cheng J, et al. Diagnostic value of rectal bleeding in predicting colorectal cancer: a systematic review. Asian Pac J Cancer Prev. 2014;15(2):1015–1021. doi: 10.7314/apjcp.2014.15.2.1015. [DOI] [PubMed] [Google Scholar]
- 31.Gondos A, Bray F, Brewster DH, et al. Recent trends in cancer survival across Europe between 2000 and 2004: a model-based period analysis from 12 cancer registries. Eur J Cancer. 2008;44:1463–1475. doi: 10.1016/j.ejca.2008.03.010. [DOI] [PubMed] [Google Scholar]
- 32.Olde Bekkink M, McCowan C, Falk GA, Teljeur C, Van de Laar FA, Fahey T. Diagnostic accuracy systematic review of rectal bleeding in combination with other symptoms, signs and tests in relation to colorectal cancer. Br J Cancer. 2010 Jan 05;102(1):48–58. doi: 10.1038/sj.bjc.6605426. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Ferlay J, Soerjomataram I, Dikshit R, et al. Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012. Int J Cancer. 2015 Mar 1;136(5):E359–386. doi: 10.1002/ijc.29210. [DOI] [PubMed] [Google Scholar]
- 34.Thompson, Pond, Ellis, Beach, Thompson Rectal bleeding in general and hospital practice; ‘the tip of the iceberg’. Colorectal Dis. 2000 Sep;2(5):288–293. doi: 10.1046/j.1463-1318.2000.00141.x. [DOI] [PubMed] [Google Scholar]
- 35.Rubin G, Hamilton W. Alarm features of colorectal cancer. Gut. 2009 Jul;58(7):1026. author reply 1026-1027. [PubMed] [Google Scholar]
- 36.Crosland A, Jones R. Rectal bleeding: prevalence and consultation behaviour. BMJ. 1995 Aug 19;311(7003):486–488. doi: 10.1136/bmj.311.7003.486. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Thompson MR, Asiimwe A, Flashman K, Tsavellas G. Is earlier referral and investigation of bowel cancer patients presenting with rectal bleeding associated with better survival? Colorectal Dis. 2011 Nov;13(11):1242–1248. doi: 10.1111/j.1463-1318.2010.02438.x. [DOI] [PubMed] [Google Scholar]
- 38.Thompson MR, Perera R, Senapati A, Dodds S. Predictive value of common symptom combinations in diagnosing colorectal cancer. Br J Surg. 2007 Oct;94(10):1260–1265. doi: 10.1002/bjs.5826. [DOI] [PubMed] [Google Scholar]