• 沒有找到結果。

In contrast to previous prediction models that incorporate only baseline f-Hb concentration and personal characteristics in the model, the proposed risk prediction model for CRC in this thesis is specific and considers the FIT results with time-varying characteristics. By combining information on both of baseline and updated f-Hb concentrations from population-based biennial FIT screening with time-dependent analysis that provides a feasible way for identifying the effect of f-Hb on the risk of CRC at pre-clinical and clinical phases. The advent of population-based screening with the FIT provides an opportunity to fully assess the usage of an incremental category of not only baseline but also subsequent measurements of f-Hb concentration on the risk of colorectal neoplasia using a longitudinal follow-up on those who

potentially at risk to develop the colorectal neoplasia.

The role of baseline quantitative f-Hb concentration on the prediction of colorectal neoplasia.

The present study firstly demonstrated that a dose-response relationship between baseline f-Hb and the risk of developing colorectal neoplasia. We have shown that a quantitative baseline f-Hb concentration is an independent predictive factor of colorectal neoplasia risk, including both invasive colorectal cancer and

adenoma. The AUC was 68.1% for baseline f-Hb using time-independent model. The incremental f-Hb concentration in association with the risk for colorectal cancer is consistent with the findings in the previous literatures (L.-S. Chen et al., 2011) (Yen et al., 2014) The possible causal effect of an incremental increase in fecal hemoglobin on advanced CRC incidence was also assessed by our study. The accuracy of predicting incident advanced CRC was assessed by using time-invariant receive operating characteristics curve analysis. The 69.8% of AUC implies that quantitative f-Hb alone is the major factor for predicting risk of advanced colorectal cancer.

The odds ratios for colorectal cancer increased with f-Hb concentration from 6-9 ug/g to 450 ug/g or above, dropped a bit for those with 100-149 ug/g, and then rebounded for those with 150-249 ug/g or above. The drop of odds ratios for

colorectal cancer could be possibly compensated by the early detection of adenoma.

However, it is interesting to see that the hazard ratios for colorectal cancer increased with f-Hb concentration from 6-9 ug/g to 450 ug/g or above by Cox proportional hazards regression model once taking follow-up time into account. Additionally, the odds ratios increased with f-Hb concentrations and are generally lower for each category of f-Hb concentration in advanced colorectal cancer compared with colorectal cancer.

Dynamic change of f-HB concentration for prediction of colorectal cancer

When the prevalent CRC cases were excluded from analysis, we found that the effect sizes of each category of f-Hb concentration on the incident CRCs after follow-up on those who test negative at prevalent screen or with a negative finding on colonoscopy were smaller than combined prevalent and incident cases. The predictive ability should be different for baseline and subsequent f-Hb concentration. A time-dependent analysis was conducted to include the repeated measurements of f-Hb concentration taken during the follow-up period at repeated screening. The findings were similar to those already described, but with larger HRs for f-Hb higher than 20-49 ug/g or above.

It should be noted that our dynamic logistic regression model with Markov process revealed the role of f-Hb not only in the initiation of PCDP but also in disease progress from PCDP to CP. A dose-response manner of f-Hb concentration on each process was also found. Moreover, the high predictive ability of this model was demonstrated by 88% of AUC in ROC analysis. These findings have significant implications for the use of f-Hb concentrations to distinguish the different risk groups according to the health status which provides opportunities for individually tailored screening strategy including the age-and sex- specific inter-screening interval by using the full information on f-Hb concentration measurements. The higher the risk

predicted by the proposed model, the shorter inter-screening interval, and the more advanced detection methods should be considered. Such a prediction model might be useful for alerting someone even having lower f-Hb but with elevated f-Hb at

repeated screen or improving patients’ awareness for attending the repeated screen and thereby detecting CRC earlier. An application is like that, the individual time-dependent risk score can be generated from both of first screening and subsequent screenings by the proposed risk prediction model which can then be applied in an individual-risk-guided invitation to a large population-based screening program.

Strengths and Limitations

The strength of our risk prediction model is that it was developed by using large population-based screening data. Large population-based study has also gained sufficient statistical power for building a prediction model for CRC. There are two limitations in the current thesis. The developed prediction models did not validate by other datasets as one of our limitation. The cross-validation and external validation are further required. The predictive abilities of FIT by time-dependent Cox

proportional hazard regression model and dynamic logistic regression model with Markov process are limited to the colorectal cancer. The study on the performance of FIT for colorectal neoplasia should be carried out in the future.

In conclusion, the results of this large population-based prospective cohort study demonstrated not only a dose-response relationship between f-Hb and incident CRC but also the dose-response manner in f-Hb concentration which plays the role on both of initiator and promoter of CRC by the novel dynamic logistic regression

model. These findings have significant implications for the better use of f-Hb concentrations on the individual-tailor CRC screening strategy.

Table 1 Number of participants, colorectal cancer (CRC) case, adenoma (non-advanced and advanced) cases in a CRC screening program

Characteristics Participants Adenoma % Advanced

Adenoma % Colorectal

Table 2 The demographic characteristics (site and stage) in Colorectal cancer cases

Table 3 Numbers of repeated visit and follow-up time of nationwide colorectal cancer screening

from 2004 to 2014

Free of CRC CRC

N / mean (% / std) N / mean (% / std)

Number of visit 1 215629 97.7% 5075 2.3%

2 247310 98.9% 2791 1.1%

3 259876 99.5% 1282 0.5%

4 138341 99.7% 455 0.3%

5 39300 99.7% 112 0.3%

6 8949 99.9% 13 0.1%

7 1734 99.8% 3 0.2%

8 76 100.0% 0 0.0%

Follow up time 7.76 1.54 4.13 2.79

Table 4 Crude ORs and adjusted ORs (aOR) for Adenoma

Adenoma

Univariate analysis Multivariate analysis

Variables OR 95% CI aOR 95% CI

Table 5 Crude ORs and adjusted ORs (aOR) for Colorectal cancer

Colorectal cancer

Univariate analysis Multivariate analysis

Variables OR 95% CI aOR 95% CI

Table 6 Crude ORs and adjusted ORs (aOR) for Neoplasia

Neoplasia

Univariate analysis Multivariate analysis

Variables OR 95% CI aOR 95% CI

Table 7 Crude ORs and adjusted ORs (aOR) for Advanced Adenoma

Advanced Adenoma

Univariate analysis Multivariate analysis

Variables OR 95% CI aOR 95% CI

Table 8 Crude ORs and adjusted ORs (aOR) for Advanced Neoplasia (AA + CRC)

Advanced Neoplasia

Univariate analysis Multivariate analysis

Variables OR 95% CI aOR 95% CI

Table 9 Crude ORs and adjusted ORs (aOR) for Advanced CRC (above stage2)

Advanced CRC

Univariate analysis Multivariate analysis

Variables OR 95% CI aOR 95% CI

Table 10 AUC of univariate and multivariate logistic regression

Adenoma Colorectal cancer Neoplasia Advanced Adenoma Advanced Neoplasia Advanced colorectal cancer

AUC 95% CI AUC 95% CI AUC 95% CI AUC 95% CI AUC 95% CI AUC 95% CI

Age 0.5497 0.5469 0.5525 0.6015 0.5961 0.6070 0.5522 0.5497 0.5548 0.5481 0.5424 0.5538 0.5692 0.5653 0.5731 0.6098 0.6021 0.6176 Gender 0.5755 0.5729 0.5780 0.5552 0.5502 0.5602 0.5726 0.5703 0.5749 0.5955 0.5904 0.6006 0.5754 0.5718 0.5790 0.5484 0.5413 0.5556 f-Hb 0.6773 0.6742 0.6803 0.6998 0.6940 0.7056 0.6851 0.6824 0.6878 0.7330 0.7269 0.7390 0.7180 0.7138 0.7222 0.6968 0.6885 0.7050 Combined 0.7173 0.7144 0.7202 0.7355 0.7299 0.7410 0.7180 0.7154 0.7207 0.7652 0.7595 0.7709 0.7449 0.7408 0.7490 0.7359 0.7280 0.7438

Table 11 Estimated effect of baseline f-Hb on colorectal cancer incidence using Cox proportional hazards regression model

Univariate analysis Multivariate analysis

Variables HR 95% CI aHR 95% CI

Table 12 Estimated effect of baseline f-Hb on colorectal cancer incidence using Cox proportional hazards regression model (exclude prevalent CRC case)

Univariate Multivariate

Table 13 Crude HRs and adjusted HRs (aHR) for Advanced Colorectal cancer

Advanced CRC

Univariate analysis Multivariate analysis

Variables HR 95% CI aHR 95% CI

Table 14 Estimated effect of f-Hb as time-varying covariate on colorectal cancer incidence using Poisson regression model

Time-dependent Time-dependent

Table 15 Estimated results on the risk of CRC evolution based on three-state Markov model with compound rate for interval cancer

Normal to PCDP PCDP to CP

Regression coefficient aHR Regression coefficient aHR

Estimate 95 CI Estimate 95 CI Estimate 95 CI Estimate 95 CI

Figure 1 Work flow of the Taiwanese nationwide colorectal cancer screening program.

FIT indicates fecal immunochemical testing

Figure 2 Receiver operating characteristic (ROC) curve of FIT for Adenoma

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

Sensitivity

1-Specificity

Age+ Gender + fHBC AUC=0.7173 Age AUC = 0.5497

Gender AUC = 0.5755 fHbC AUC = 0.6773

Figure 3 Receiver operating characteristic (ROC) curve of FIT for colorectal cancer

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

Sensitivity

1-Specificity

Age+ Gender + fHBC AUC=0.7355 Age AUC = 0.6015

Gender AUC = 0.5552 fHbC AUC = 0.6998

Figure 4 Receiver operating characteristic (ROC) curve of FIT for neoplasia

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

Sensitivity

1-Specificity

Age+ Gender + fHBC AUC=0.7180 Age AUC = 0.5522

Gender AUC = 0.5720 fHbC AUC = 0.6851

Figure 5 Receiver operating characteristic (ROC) curve of FIT for advanced adenoma

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

Sensitivity

1-Specificity

Age+ Gender + fHBC AUC=0.7652 Age AUC = 0.5481

Gender AUC = 0.5955 fHbC AUC = 0.7330

Figure 6 Receiver operating characteristic (ROC) curve of FIT for advanced Neoplasia

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

Sensitivity

1-Specificity

Age+ Gender + fHBC AUC=0.7449 Age AUC = 0.5692

Gender AUC = 0.5754 fHbC AUC = 0.7180

Figure 7 Receiver operating characteristic (ROC) curve of FIT for advanced colorectal cancer

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

Sensitivity

1-Specificity

Age+ Gender + fHBC AUC=0.7355 Age AUC = 0.6096

Gender AUC = 0.5484 fHbC AUC = 0.6964

Figure 8 Receiver operating characteristic (ROC) curve of predicting colorectal cancer incidence by age, gender, and time-varying f-Hb in 2 years, 4 years, and 6 years

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Sensitivity

1-Spec

2 year AUC: 91.2 (95%CI: 91.0-91.3)

4 year AUC: 87.9 (95%CI:82.3-93.4)

6 year AUC: 83.8 (95%CI: 79.4-88.4)

Figure 9 Receiver operating characteristic curve (ROC) of FIT considering age and gender for CRC with Markov process of dynamic logistic regression approach

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Sensitivity

1-Specificity

AUC:88.0%(87.3%-88.6%)

REFERENCE

Auge, J. M., Pellise, M., Escudero, J. M., Hernandez, C., Andreu, M., Grau, J., . . . Group, P. (2014). Risk stratification for advanced colorectal neoplasia according to fecal hemoglobin concentration in a colorectal cancer screening program.

Gastroenterology, 147(3), 628-636 e621. doi:10.1053/j.gastro.2014.06.008 Brenner, H., Kloor, M., & Pox, C. P. (2014). Colorectal cancer. Lancet, 383(9927),

1490-1502. doi:10.1016/S0140-6736(13)61649-9

Chen, L.-S., Yen, A. M.-F., Chiu, S. Y.-H., Liao, C.-S., & Chen, H.-H. (2011). Baseline faecal occult blood concentration as a predictor of incident colorectal neoplasia:

longitudinal follow-up of a Taiwanese population-based colorectal cancer screening cohort. The Lancet Oncology, 12(6), 551-558. doi:10.1016/s1470-2045(11)70101-2

Chen, L. S., Yen, M. F., Wu, H. M., Liao, C. S., Liou, D. M., Kuo, H. S., & Chen, T. H.

(2005). Predictive survival model with time-dependent prognostic factors:

development of computer-aided SAS Macro program. J Eval Clin Pract, 11(2), 181-193. doi:10.1111/j.1365-2753.2005.00519.x

Chiu, S. Y., Chuang, S. L., Chen, S. L., Yen, A. M., Fann, J. C., Chang, D. C., . . . Chiu, H. M. (2017). Faecal haemoglobin concentration influences risk prediction of interval cancers resulting from inadequate colonoscopy quality: analysis of the Taiwanese Nationwide Colorectal Cancer Screening Program. Gut, 66(2), 293-300. doi:10.1136/gutjnl-2015-310256

Cooper, J. A., Parsons, N., Stinton, C., Mathews, C., Smith, S., Halloran, S. P., . . .

FIT and routine screening data: development of a risk prediction model. Br J Cancer, 118(2), 285-293. doi:10.1038/bjc.2017.375

de Wijkerslooth, T. R., Stoop, E. M., Bossuyt, P. M., Meijer, G. A., van Ballegooijen, M., van Roon, A. H., . . . Kuipers, E. J. (2012). Immunochemical fecal occult blood testing is equally sensitive for proximal and distal advanced neoplasia. Am J Gastroenterol, 107(10), 1570-1578. doi:10.1038/ajg.2012.249

Digby, J., Fraser, C. G., Carey, F. A., McDonald, P. J., Strachan, J. A., Diament, R.

H., . . . Steele, R. J. C. (2013). Faecal haemoglobin concentration is related to severity of colorectal neoplasia. Journal of Clinical Pathology, 66(5), 415-419.

doi:10.1136/jclinpath-2013-201445

Guittet, L., Guillaume, E., Levillain, R., Beley, P., Tichet, J., Lantieri, O., & Launoy, G.

(2011). Analytical comparison of three quantitative immunochemical fecal occult blood tests for colorectal cancer screening. Cancer Epidemiol Biomarkers Prev, 20(7), 1492-1501. doi:10.1158/1055-9965.EPI-10-0594

Hardcastle, J. D., Chamberlain, J. O., Robinson, M. H. E., Moss, S. M., Amar, S. S., Balfour, T. W., . . . Mangham, C. M. (1996). Randomised controlled trial of faecal-occult-blood screening for colorectal cancer. The Lancet, 348(9040), 1472-1477. doi:10.1016/s0140-6736(96)03386-7

Haug, U., Kuntz, K. M., Knudsen, A. B., Hundt, S., & Brenner, H. (2011). Sensitivity of immunochemical faecal occult blood testing for detecting left- vs right-sided colorectal neoplasia. Br J Cancer, 104(11), 1779-1785.

doi:10.1038/bjc.2011.160

Hewitson, P., Glasziou, P., Watson, E., Towler, B., & Irwig, L. (2008). Cochrane

systematic review of colorectal cancer screening using the fecal occult blood test

(hemoccult): an update. Am J Gastroenterol, 103(6), 1541-1549.

doi:10.1111/j.1572-0241.2008.01875.x

Hol, L., van Leerdam, M. E., van Ballegooijen, M., van Vuuren, A. J., van Dekken, H., Reijerink, J. C., . . . Kuipers, E. J. (2010). Screening for colorectal cancer:

randomised trial comparing guaiac-based and immunochemical faecal occult blood testing and flexible sigmoidoscopy. Gut, 59(1), 62-68.

doi:10.1136/gut.2009.177089

Hol, L., Wilschut, J. A., van Ballegooijen, M., van Vuuren, A. J., van der Valk, H., Reijerink, J. C., . . . van Leerdam, M. E. (2009). Screening for colorectal cancer:

random comparison of guaiac and immunochemical faecal occult blood testing at different cut-off levels. Br J Cancer, 100(7), 1103-1110.

doi:10.1038/sj.bjc.6604961

Ju, J. H., Chang, S. C., Wang, H. S., Yang, S. H., Jiang, J. K., Chen, W. C., . . . Lin, J. K.

(2007). Changes in disease pattern and treatment outcome of colorectal cancer: a review of 5,474 cases in 20 years. Int J Colorectal Dis, 22(8), 855-862.

doi:10.1007/s00384-007-0293-z

Kronborg, O., Fenger, C., Olsen, J., Jørgensen, O. D., & Søndergaard, O. (1996).

Randomised study of screening for colorectal cancer with faecal-occult-blood test. The Lancet, 348(9040), 1467-1471. doi:10.1016/s0140-6736(96)03430-7 Li, W., Zhao, L. Z., Ma, D. W., Wang, D. Z., Shi, L., Wang, H. L., . . . Chen, S. L.

(2018). Predicting the risk for colorectal cancer with personal characteristics and fecal immunochemical test. Medicine (Baltimore), 97(18), e0529.

doi:10.1097/MD.0000000000010529

Liao, C. S., Lin, Y. M., Chang, H. C., Chen, Y. H., Chong, L. W., Chen, C. H., . . . Shih,

concentration for risk prediction of colorectal neoplasia. World J Gastroenterol, 19(45), 8366-8372. doi:10.3748/wjg.v19.i45.8366

Loupakis, F., Yang, D., Yau, L., Feng, S., Cremolini, C., Zhang, W., . . . Lenz, H. J.

(2015). Primary tumor location as a prognostic factor in metastatic colorectal cancer. J Natl Cancer Inst, 107(3). doi:10.1093/jnci/dju427

Mandel, J. S., Bond, J. H., Church, T. R., Snover, D. C., Bradley, G. M., Schuman, L.

M., & Ederer, F. (1993). Reducing Mortality from Colorectal Cancer by

Screening for Fecal Occult Blood. New England Journal of Medicine, 328(19), 1365-1371. doi:10.1056/nejm199305133281901

Meguid, R. A., Slidell, M. B., Wolfgang, C. L., Chang, D. C., & Ahuja, N. (2008). Is there a difference in survival between right- versus left-sided colon cancers? Ann Surg Oncol, 15(9), 2388-2394. doi:10.1245/s10434-008-0015-y

Morikawa, T., Kato, J., Yamaji, Y., Wada, R., Mitsushima, T., & Shiratori, Y. (2005). A comparison of the immunochemical fecal occult blood test and total

colonoscopy in the asymptomatic population. Gastroenterology, 129(2), 422-428. doi:10.1016/j.gastro.2005.05.056

Park, C. H., Jung, Y. S., Kim, N. H., Park, J. H., Park, D. I., & Sohn, C. I. (2019).

Usefulness of risk stratification models for colorectal cancer based on fecal hemoglobin concentration and clinical risk factors. Gastrointest Endosc, 89(6), 1204-1211 e1201. doi:10.1016/j.gie.2019.02.023

Parra-Blanco, A., Gimeno-Garcia, A. Z., Quintero, E., Nicolas, D., Moreno, S. G., Jimenez, A., . . . Lopez-Bastida, J. (2010). Diagnostic accuracy of

immunochemical versus guaiac faecal occult blood tests for colorectal cancer screening. J Gastroenterol, 45(7), 703-712. doi:10.1007/s00535-010-0214-8

Price, T. J., Beeke, C., Ullah, S., Padbury, R., Maddern, G., Roder, D., . . . Karapetis, C.

(2015). Does the primary site of colorectal cancer impact outcomes for patients with metastatic disease? Cancer, 121(6), 830-835. doi:10.1002/cncr.29129 Smith, A., Young, G. P., Cole, S. R., & Bampton, P. (2006). Comparison of a

brush-sampling fecal immunochemical test for hemoglobin with a sensitive guaiac-based fecal occult blood test in detection of colorectal neoplasia. Cancer, 107(9), 2152-2159. doi:10.1002/cncr.22230

Stegeman, I., de Wijkerslooth, T. R., Stoop, E. M., van Leerdam, M. E., Dekker, E., van Ballegooijen, M., . . . Bossuyt, P. M. (2014). Combining risk factors with faecal immunochemical test outcome for selecting CRC screenees for colonoscopy.

Gut, 63(3), 466-471. doi:10.1136/gutjnl-2013-305013

Taylor, D. P., Burt, R. W., Williams, M. S., Haug, P. J., & Cannon-Albright, L. A.

(2010). Population-based family history-specific risks for colorectal cancer: a constellation approach. Gastroenterology, 138(3), 877-885.

doi:10.1053/j.gastro.2009.11.044

van Rossum, L. G., van Rijn, A. F., Laheij, R. J., van Oijen, M. G., Fockens, P., van Krieken, H. H., . . . Dekker, E. (2008). Random comparison of guaiac and immunochemical fecal occult blood tests for colorectal cancer in a screening population. Gastroenterology, 135(1), 82-90. doi:10.1053/j.gastro.2008.03.040 Weiss, J. M., Pfau, P. R., O'Connor, E. S., King, J., LoConte, N., Kennedy, G., & Smith,

M. A. (2011). Mortality by stage for right- versus left-sided colon cancer:

analysis of surveillance, epidemiology, and end results--Medicare data. J Clin Oncol, 29(33), 4401-4409. doi:10.1200/JCO.2011.36.4414

Weitz, J., Koch, M., Debus, J., Hohler, T., Galle, P. R., & Buchler, M. W. (2005).

Colorectal cancer. Lancet, 365(9454), 153-165. doi:10.1016/S0140-6736(05)17706-X

Win, A. K., Macinnis, R. J., Hopper, J. L., & Jenkins, M. A. (2012). Risk prediction models for colorectal cancer: a review. Cancer Epidemiol Biomarkers Prev, 21(3), 398-410. doi:10.1158/1055-9965.EPI-11-0771

Wong, R. (2010). Proximal tumors are associated with greater mortality in colon cancer.

J Gen Intern Med, 25(11), 1157-1163. doi:10.1007/s11606-010-1460-4 Yang, J., Du, X. L., Li, S. T., Wang, B. Y., Wu, Y. Y., Chen, Z. L., . . . Yang, J. (2016).

Characteristics of Differently Located Colorectal Cancers Support Proximal and Distal Classification: A Population-Based Study of 57,847 Patients. PLoS One, 11(12), e0167540. doi:10.1371/journal.pone.0167540

Yen, A. M., Chen, S. L., Chiu, S. Y., Fann, J. C., Wang, P. E., Lin, S. C., . . . Chen, H. H.

(2014). A new insight into fecal hemoglobin concentration-dependent predictor for colorectal neoplasia. Int J Cancer, 135(5), 1203-1212. doi:10.1002/ijc.28748

相關文件