Digestive and Liver Disease 48 (2016) 321–326

Contents lists available at ScienceDirect

Digestive and Liver Disease journal homepage: www.elsevier.com/locate/dld

Oncology

Predictors of advanced colorectal neoplasia at initial and surveillance colonoscopy after positive screening immunochemical faecal occult blood test Edoardo Botteri a,∗ , Cristiano Crosta b , Vincenzo Bagnardi a,c , Darina Tamayo b , Angelica Maria Sonzogni d , Giuseppe De Roberto b , Annalisa de Leone b , Albert Lowenfels e , Patrick Maisonneuve a a

Division of Epidemiology and Biostatistics, European Institute of Oncology, Milan, Italy Division of Endoscopy, European Institute of Oncology, Milan, Italy c Department of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milan-Bicocca, Milan, Italy d Division of Pathology, European Institute of Oncology, Milan, Italy e Department of Surgery, New York Medical College, Valhalla, New York, United States b

a r t i c l e

i n f o

Article history: Received 22 June 2015 Accepted 19 November 2015 Available online 2 December 2015 Keywords: Colonoscopy Colorectal cancer Lifestyle Screening

a b s t r a c t Background: Characteristics such as gender and lifestyle are not taken in account in colorectal cancer screening and surveillance recommendations. Aims: To identify factors associated with advanced neoplasia at initial and surveillance colonoscopy. Methods: In this observational study, 750 individuals with positive faecal occult blood test, aged 50–74 years, underwent a first screening colonoscopy in 2007–2009. We collected anthropometric data as well as data on physical activity, smoking and drinking habits, fruit and vegetable consumption and low-dose aspirin use through a questionnaire. Results: At initial colonoscopy advanced neoplasia (n = 399, 53.2%) was positively associated with age, male gender, smoking and alcohol drinking, and inversely associated with physical activity, fruit and vegetables consumption and long-term use of aspirin. These 7 factors were used to calculate a risk score, ranging from 0 (no unfavourable characteristics) to 7 (all unfavourable characteristics present), which was significantly associated with advanced neoplasia (odds ratio 1.55 for one point increase, P < 0.01). Among the 372 adenoma patients who returned for follow-up surveillance colonoscopy, the score remained associated with advanced neoplasia (odds ratio 1.28 for one point increase, P = 0.01). Conclusion: Besides age and gender, modifiable factors such as lifestyle and aspirin use were associated with the risk of advanced neoplasia at initial and surveillance colonoscopy. © 2015 Editrice Gastroenterologica Italiana S.r.l. Published by Elsevier Ltd. All rights reserved.

1. Introduction Colorectal cancer (CRC) is the third most commonly diagnosed cancer in males and the second in females, with over 1.2 million new cases and 608,700 deaths estimated to have occurred in 2008 worldwide [1]. The detection and removal of precancerous lesions through CRC screening and the intervention on modifiable risk factors for CRC, such as physical activity, dietary fibre intake, alcohol consumption and smoking habits [2–6], are strategies for

∗ Corresponding author at: Division of Epidemiology and Biostatistics, European Institute of Oncology, Via Ripamonti 435, 20141 Milano, Italy. Tel.: +39 02 57489820; fax: +39 02 94379221. E-mail address: [email protected] (E. Botteri).

reducing both CRC incidence and mortality. It was estimated that CRC screening and changes in risk factors accounted each for 50% of the CRC incidence rates decline in USA during 1975–2000 [7]. While CRC screening is widely accepted, there is no consensus on the preferred screening strategy, therefore stool tests and endoscopic exams are commonly prescribed alone or in combination. Clinical trials demonstrated that immunochemical faecal occult blood test (iFOBT) has higher participation rate and detection rate of advanced colorectal neoplasm than guaiac-based faecal occult blood test in the screening population [8,9], and currently iFOBT is considered the most valid stool test option in most guidelines [10–12]. Whether the performance of iFOBT depends on the characteristics and habits of the participants and whether subsequent surveillance of patients with adenoma should be scheduled in accordance to those factors is unknown.

http://dx.doi.org/10.1016/j.dld.2015.11.020 1590-8658/© 2015 Editrice Gastroenterologica Italiana S.r.l. Published by Elsevier Ltd. All rights reserved.

322

E. Botteri et al. / Digestive and Liver Disease 48 (2016) 321–326

In the present study, we investigated whether age, gender, lifestyle and use of daily low-dose aspirin can predict the presence, type and location of colorectal neoplasia in iFOBT-positive individuals who participated in a CRC screening programme. We also evaluated the association of those factors with endoscopic findings at follow-up surveillance colonoscopy.

2. Materials and methods An iFOBT-based CRC screening programme was implemented in the Italian region of Lombardy (∼10 million residents) since 2005. Eligible subjects are men and women aged 50–69 years, who are invited to undergo 1-day iFOBT every two years, with no dietary or medical restriction. In the case of positive iFOBT, subjects are invited to undergo a total colonoscopy. A quantitative iFOBT with a cut-off faecal haemoglobin concentration of 100 ng/mL were used. In this observational cohort study, individuals were identified retrospectively using the database of the division of endoscopy of the European Institute of Oncology (IEO) in Milan where colonoscopy was performed. We identified and selected all consecutive asymptomatic participants (N = 870) in the CRC screening programme of Lombardy region, who underwent a colonoscopy after positive iFOBT at the institute in 2007–2009. Patients with one first-degree relative diagnosed with CRC before the age of 60, and patients who had undergone colorectal endoscopic examination before the initial colonoscopy were excluded, as well as patients with hereditary CRC syndromes (e.g. Lynch and familial adenomatous polyposis syndromes), chronic inflammatory bowel disease, history of colorectal polyps or cancer, or previous bowel resection. At the time of first colonoscopy, participants gave written informed consent to the use of their personal data for epidemiological research purposes and to be contacted by telephone. The study was formally notified to the IEO Institutional Review Board on July 15th, 2010. During 2012 an ad hoc telephone questionnaire was administered by two clinicians to retrieve information on body mass index (BMI), smoking and alcohol habits, physical activity, fruits and vegetables consumption, and use of low-dose aspirin at the time of index colonoscopy. We categorized variables according to median values, tertiles or quartiles, as appropriate. Age was dichotomized according to the median value. Number of pack-years was calculated as the average number of cigarettes smoked per day multiplied by the years of smoking divided by 20. Pack-years = 0 was kept as a separate category. Pack-years > 0 were categorized using quartiles, as well as fruit and vegetable courses. Individuals were categorized as never and ever drinkers and the number of drinks per day was calculated as the sum of any type of alcoholic beverages consumed in a week divided by 7. Drinks = 0 was kept as a separate category. Drinks > 0 were categorized using tertiles. Physical activity was categorized as low (sedentary typical day and no regular sport), moderate (moderately active day and no regular sport/sedentary day and regular sport) and high (very active day with or without sport/moderately active day and regular sport). Among regular users of low-dose aspirin, duration was categorized according to the approximate median value of 5 years. Individuals were interviewed during 2012 and all questions referred to the time of initial colonoscopy. If a person did not answer the first phone call, a second attempt was made after 21 days and, if necessary, a third attempt after another 21 days. The 2006 guideline on post-polypectomy surveillance of the United States Multi-Society Task Force on CRC were used to distinguish two main types of adenomas: low-risk adenomas, defined as 1–2 tubular adenomas < 10 mm, and high-risk adenomas, defined as adenoma with villous histology, high-grade dysplasia, ≥10 mm,

or 3 or more adenomas [13]. Invasive neoplasia and high-risk adenoma were grouped to form the advanced neoplasia (advanced neoplasia) category. 2.1. Statistical methods The Chi-square test and the Chi-square test for trend were used to explore the associations between endoscopic findings and individuals’ characteristics. Multivariable logistic regression was used to identify factors independently associated with colorectal neoplasia at initial colonoscopy. Variables that were statistically significant or borderline significant (P-value ≤ 0.05) at univariate analysis were included in the multivariable model. An individual risk score was built based on the same variables: one point was added to the score for each of the risk category of all included variable. If any of the factors used to generate the lifestyle score was missing, the lifestyle score was coded as missing. Classification accuracy of the final model was evaluated internally with respect to discrimination (i.e. the ability of the model to classify a patient with the outcome from a patient without the outcome) and calibration (i.e. the agreement between the outcome frequencies observed in the data and the predicted probabilities of the model). Discrimination was measured by the area under the ROC curve (AUC) [14]; calibration was tested using the Hosmer and Lemeshow goodness-of-fit test [15]. A 10-folds internal crossvalidation was used to address model over-fit, in order to assess how our score would generalize to a new patient cohort [16]. Then, among patients with adenoma at the first colonoscopy, a multivariable logistic model for the prediction of advanced neoplasia at surveillance colonoscopy was built including severity of adenoma at first colonoscopy and individual patient risk score as independent variables. The surveillance colonoscopy date was decided at the time of initial colonoscopy and based on the severity of outcome of the initial colonoscopy, according the so-called “doctor’s

n=870 Colonoscopies after positive iFOBT in 20072009

n=120 Excluded (n=19 previous colonoscopy; n=20 family history; n=70 not reachable; n=5 refused; n=6 died)

n=750 Telephone interviews

n=280 No adenoma

n=470 Adenoma

n=98 Lost to follow-up

n=372 Underwent surveillance colonoscopy

Fig. 1. Study flowchart. iFOBT, immunochemical faecal occult blood test.

E. Botteri et al. / Digestive and Liver Disease 48 (2016) 321–326

care scheme”. This scheme of examination is non-informative with regard to the outcome of interest [17]. SAS (SAS Institute, Cary, NC) and R (http://cran.r-project.org/) software was used. All tests were two-sided. 3. Results One-hundred and twenty eligible patients out of the total 870 (13.8%) were excluded (see Fig. 1). The study population was 750 subjects (51.2% males, median age 64 years); 67 subjects (8.9%) were older than 69 years because they performed their first iFOBT after the second or third invitation. Endoscopic findings are reported in Table 1: 186 cases had no polyps (24.8%), 35 had non-adenomatous polyps (4.7%); 130 cases had low-risk adenomas (17.3%) and 340 had high-risk adenomas (45.3%). Finally, 59 cases had invasive neoplasia (7.9%): 41 were adenocarcinoma (5.5%; stage I, n = 10; stage II, n = 10; stage III, n = 19; stage IV, n = 1; 1 missing); 16 were adenocarcinoma in a polyp (2.1%), 1 was a neuroendocrine tumour (0.1%), and 1 was squamous cell carcinoma (0.1%). The positive predictive value (PPV) of iFOBT was 53.2% for advanced neoplasia and 7.6% for adenocarcinoma. The detection rates of advanced neoplasia in association with participants’ characteristics are reported in Table 2. The detection rate of advanced neoplasia was higher in males compared to females, it increased with increasing age, pack-years of smoking and alcohol consumption, while it decreased with increasing

323

Table 1 Most severe finding at initial colonoscopy. Outcome

All N (%)

Men N (%)

WomenN (%)

Negative colonoscopy Non-oncological conditions Non-adenomatous polyp Low-risk adenomaa High-risk adenomab Adenocarcinoma Squamous cell carcinoma Neuroendocrine tumour Total

43 (5.7) 143 (19.1) 35 (4.7) 130 (17.3) 340 (45.3) 57 (7.6) 1 (0.1) 1 (0.1) 750

19 (4.9) 55 (14.3) 17 (4.4) 54 (14.1) 205 (53.4) 34 (8.9) 0 (0.0) 0 (0.0) 384 (51.2)

24 (6.6) 88 (24.0) 18 (4.9) 76 (20.8) 135 (36.9) 23 (6.3) 1 (0.3) 1 (0.3) 366 (48.9)

a One or two adenomas < 10 mm with no villous component and no evidence of high-grade dysplasia; b At least one of the following: adenoma with villous histology, high-grade dysplasia, diameter ≥ 10 mm, or ≥3 adenomas.

physical activity and consumption of fruit and vegetables. Finally, the detection rate of advanced neoplasia was lower in long-term users of low-dose aspirin (>5 years) compared to non-users and short-term users. Multivariable analysis confirmed the association between detection of advanced neoplasia and all factors but age and physical activity (P = 0.15 and 0.06, respectively). However, in multivariable analysis, age and physical activity were associated with detection rate of adenocarcinoma (N = 57), together with gender and alcohol consumption (data not shown). The predicted probability of detection of advanced neoplasia for a typical male and female subject (all factors in the model were set to their mean values) at

Table 2 Characteristics and association with advanced neoplasia, defined as high-risk adenoma and invasive neoplasia, at initial colonoscopy. Variable

Categories

All individuals

Univariate analysis

Multivariable analysis

Frequency No. (col %)

Advanced eoplasia No. (row %)

Pc

Comparison

OR (95% CI)d

Pd

Predictors of advanced colorectal neoplasia at initial and surveillance colonoscopy after positive screening immunochemical faecal occult blood test.

Characteristics such as gender and lifestyle are not taken in account in colorectal cancer screening and surveillance recommendations...
562B Sizes 0 Downloads 6 Views