| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Medicine and Therapeutics2 Hong Kong Institute of Diabetes and Obesity3 Li Ka Shing Institute of Health Sciences, The Chinese University of Hong Kong, The Prince of Wales Hospital, Shatin, Hong Kong SAR, China4 Tianjin Cancer Hospital and Institute, Tianjin, China
(Correspondence should be addressed to J C N Chan; Email: jchan{at}cuhk.edu.hk)
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
27.6 kg/m2), triglyceride (0.81 to <1.41 mmol/l), high-density lipoprotein cholesterol (<0.9 or 1.8 mmol/l), total cholesterol (<4.3 mmol/l) and white blood cell (WBC) count (<5.8x109 count per litre) were associated with increased cancer risks and exhibited non-linear relationships. We further linear transformed these terms for selection using backward Cox regression (P<0.05 for stay) in the training dataset. In the test dataset, calibration was checked using Hosmer–Lemeshow test and discrimination checked using area under receiver operating characteristic curve. In addition to age and current smoking, only linear-transformed total cholesterol and WBC count were selected. The risk score was 0.0488xage (years)–0.5810xtotal cholesterol (mmol/l, coded to 4.3 if >4.3)–0.3596xWBC count (109 counts/l, 5.8 if >5.8)+0.6390xcurrent smoking status (1 if yes). The 5-year probability of cancer was 1–0.9590EXP(0.9382x(RISK SCORE+1.5903)).The predicted cancer probability was not significantly different from the observed cancer probability during the 5-year follow-up. The adjusted area under receiver operating characteristic curve was 0.712. In conclusion, BMI, lipids and WBC count have predicting values for cancer.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The Framingham Study developed a risk score for predicting coronary heart disease in general population (Anderson et al. 1991). Using the Hong Kong Diabetes Registry, our group has developed risk scores to predict end-stage renal disease (Yang et al. 2006b, 2007c), stroke (Yang et al. 2007b) and coronary heart disease (Yang et al. 2008a) in Chinese T2DM patients. More recently, our group has developed a hazard ratio (HR) curve technique to identify the full-range risk relationships between common risk factors and all-cause death (So et al. 2008) and coronary heart disease (Yang et al. 2007a). Based on the identified curving relationships between risk factors and all-cause death, we also developed a risk score to predict all-cause death in T2DM (Yang et al. 2008b). In this analysis, we applied similar methods to delineate the relationships between risk factors and all-site cancer followed by validation of a risk score to predict all-site cancer in T2DM.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The Prince of Wales Hospital is the teaching hospital of the Chinese University of Hong Kong and serves a population of over 1.2 million. The Hong Kong Diabetes Registry was established in 1995 as part of a quality improvement programme and enrols 30–50 ambulatory diabetic patients on a weekly basis. The referral sources included general practitioners, community clinics and other specialty clinics and patients discharged from the Prince of Wales Hospital or other regional hospitals. Enrolled patients with hospital admissions within 6–8 weeks prior to assessment accounted for <10% of all referrals. The 4-h assessment of complications and risk factors was performed on an outpatient basis, modified from the European DiabCare protocol (Piwernetz et al. 1993). Once a diabetic subject had undergone this comprehensive assessment, he/she was considered to have entered this study cohort and would be followed up till the time of death. Ethical approval was obtained from the Chinese University of Hong Kong Clinical Research Ethics Committee. The Declaration of Helsinki was adhered and informed consent obtained from every patient for data analysis and research purpose at the time of assessment.
For this analysis, the clinical endpoints including discharge diagnoses of hospital admissions and mortality on or before 30 July 2005 were recorded or otherwise censored on 30 July 2005. Details of all medical admissions of the cohort by that date were retrieved from the Hong Kong Hospital Authority Central Computer System that records admissions to all public hospitals in Hong Kong. The latter provide 95% of the total hospital bed-days in Hong Kong due to its heavily subsidized health care system. Mortality data from the Hong Kong Death Registry was also retrieved and all causes of death were further ascertained by review of case notes by an endocrinologist. These databases were matched by a unique identification number, the Hong Kong Identity Card number, which is compulsory for all residents in Hong Kong.
Hospital discharge principal diagnoses coded by the International Classification of Diseases, Ninth Revision (ICD-9) were used to identify first cancer events. The endpoint of this study was defined as having a first incident cancer (codes 140–208). Upper digestive tract cancer was defined as malignant neoplasm of esophagus and stomach (codes 150–151) and lower digestive tract cancer defined as malignant neoplasm of colon, rectum, rectosigmoid junction and anus (codes 153–154).
From 1995 to 2005, 7920 diabetic patients were enrolled in this Registry. Among them, 332 with type 1 diabetes defined as acute presentation with diabetic ketoacidosis, heavy ketonuria (>3+) or continuous requirement of insulin within 1 year of diagnosis (Laakso & Pyorala 1985), and 5 with uncertain type 1 diabetes status, were excluded from the analysis. Among them, 49 with non-Chinese or unknown nationality were excluded. Another 160 patients were excluded due to past history of cancer. A total of 7374 Chinese T2DM patients without cancer history were included in the analysis.
Clinical measurements and laboratory assays
Details of assessment methods and definitions have been described previously (Yang et al. 2006b). On the day of assessment, patients attended the centre after an 8-h fasting and underwent anthropometric measurements and laboratory investigations. This study used the abbreviated MDRD formula recalibrated for the Chinese (Ma et al. 2006) to estimate glomerular filtration rate expressed in ml/min 1.73/m2: eGFR=186x(SCRx0.011)–1.154x(age)–0.203x(0.742 if female)x1.233, where SCR is serum creatinine expressed as µmol/l (original mg/dl converted to µmol/l) and 1.233 is the adjusting coefficient for the Chinese. Body weight and height were measured with subjects wearing light clothing and without shoes. Blood pressure (BP) was measured using a Dinamap PRO-100 device (Critikon, Milwaukee, WI, USA).
Complete blood picture was assayed using Beckman Coulter counter. Plasma glucose was measured by a hexokinase method (Hitachi 911 automated analyser, Boerhringer Mannheim). Total cholesterol, triglyceride and high-density lipoprotein cholesterol (HDL-C) were measured by enzymatic methods on a Hitachi 911 automated analyser (Boehringer Mannheim) using the reagent kits supplied by the manufacturer of the analyser. The inter- and intra-assay coefficients of variation have been described previously (Yang et al. 2006b).
Statistical analyses
The Statistical Analysis System (Release 9.10) was used to perform the statistical analysis (SAS Institute Inc., Cary, NC, USA). Restricted cubic splines (RCS) are piecewise cubic polynomials connected across different intervals of a continuous variable (Harrell 2001). It can fit sharply curving shapes with an additional advantage that only k–1 parameters must be estimated (k is the number of knots) (Harrell 2001). With the application of RCS in Cox models, we had developed a HR curve method to examine full-range associations between risk factors and endpoints such as death (So et al. 2008) and coronary heart disease (Yang et al. 2007a). As before, we chose four knots at quantiles 0.05, 0.35, 0.65 and 0.95, which has been suggested to offer adequate fit of the model and is a good compromise between flexibility and loss of precision caused by overfitting a small sample (Harrell 2001). HR between two points of a continuous variable can be estimated by EXP (Y2–Y1), where Y1 and Y2 are the corresponding spline function values of the two points. If we select a proper point Y1 as the referent, EXP (Y2–Y1) stands for the HR of point 2 versus point 1. Thus, we obtained the HR curves by plotting the HRs of all other points versus the referent point.
In this analysis, a forward stepwise algorithm (P<0.05 for entry and stay) was used to identify the risk factors of all-site cancer. The detail on how to use stepwise algorithm in the spline Cox model has been described (So et al. 2008). Based on the shapes of these curved relationships between risk factors and all-site cancer, these continuous variables were categorized into two or three levels and then traditional Cox analysis was performed to confirm the findings in the risk curve analysis. We also attempted to devise simple formulae to mimic the curved relationships between identified risk factors and all-site cancer to develop a cancer risk score.
Split-half validation was used in developing and validating the risk score (Yang et al. 2008b). In the training subsample, Cox proportional hazard regression with the backward algorithm (P<0.05 for stay) was used to obtain a subset of baseline predictors for all-site cancer, which were identified in the above step. Proportional hazards assumption and functional form were checked using the Supremum test (Lin et al. 1993). P<0.05 for proportional hazards was considered to violate the assumption while P<0.05 for functional form suggests that further improvement in the linear transformation remains possible. The construction of a risk score and a t-year probability equation have been described in detail elsewhere (Yang et al. 2008a). Let LR denote the likelihood ratio 
2 and P the number of predictors in the final model, then the estimated shrinkage is (LR-p)/LR and a shrinkage below 0.85 raises concern of overfitting (Harrell 2001).
Validation of the risk score was performed using the test dataset. Calibration was checked using the Hosmer–Lemeshow test (Yang et al. 2008b). P<0.05 indicates significant difference between the predicted and observed rates, suggesting poor calibration. We used Pencina & D'Agostino (2004) method to calculate C index (or c statistics) for overall discrimination and Chambless & Diao (2006) method to calculate the aROC, sensitivity and specificity for the 5-year specific discrimination and selection of cut-off points. Kaplan–Meier estimator was also used to calculate the cumulative incidence of all-site cancer in patients with a risk score at or above and below the cut-off point.
In order to identify the contributions of cancer subtypes to the overall predicting ability of the risk score, a series of Cox regression analyses using the whole cohort were performed, with the risk score as the sole predictor and major subtypes of cancer in the cohort as endpoints. P<0.05 is considered to be statistically significant.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The median age of the cohort was 57 (interquartile range (IQR) 47–67) years. At the entry of the study, they had a median duration of 6 years of diabetes (IQR 1–11 years). 4.95% (365 patients) of the cohort developed cancer during a median follow-up period of 5.45 (IQR 2.92–7.78) years and the annual incidence rate was 9.20 (95% CI 8.26–10.14) per 1000 person-years. There were 732 deaths, including 161 deaths with cancer as the principal diagnosis. Comparisons of baseline variables between the cancer cases and their non-cancer counterparts are listed in Table 1. The training dataset and the test dataset contained 3701 (total cancer 187) and 3673 (total cancer 178) patients, respectively. Baseline variables of the two groups are listed in Table 1.
|
Age, body mass index (BMI), current smoker status, total cholesterol, HDL-C, triglyceride and white blood cell (WBC) were selected as predictors of incident cancer by the stepwise spline Cox regression model. Except for age that was linearly associated with cancer, other continuous risk factors were associated with all-site cancer in a non-linear manner (Fig. 1). BMI was associated with cancer in a linear manner in univariate analysis but in a U-shaped relationship in multivariate analysis. HDL-C was associated with cancer in a U-shaped manner with the nadir at 1.22 mmol/l. Total cholesterol was negatively associated with rapid increase in cancer risk at levels <4.3 mmol/l. From 0.2 to 2.5 mmol/l, triglyceride was associated with cancer in an A-shaped manner with the zenith at 1.17 mmol/l. WBC count exhibited a negative risk association with all-site cancer as the count fell below 5.8x109 counts per litre.
|
|
In the training database, age, current smoking and formula linear-transformed total cholesterol and WBC count were selected to predict incident cancer while ex-smoking and the formula linear-transformed BMI, HDL-C and triglyceride were not selected into the predicting model by the backward algorithms (Table 3). The overall C index of the predicting model in the training dataset was 0.716 (95% CI 0.685–0.747). The shrinkage of the predicting model was, if any, small (shrinkage=0.9382) and the formulae linear-transformed TC and WBC were adequate (P>0.05 for functional form). The constructed risk score was 0.0488xage (years)–0.5810xtotal cholesterol (mmol/l, coded to 4.3 if >4.3)–0.3596xWBC (109 counts/l, 5.8 if >5.8)+0.6390xcurrent smoking status (1 if yes). The 5-year probability of cancer was 1–0.9590EXP(0.9382x(RISK SCORE+1.5903)).
|
In the test database, the predicted cancer probability was not significantly different from the observed probability of cancer over 5 years of follow-up (Fig. 2). The overall C index was 0.695 (95% CI 0.667–0.724). With follow-up time and censoring status considered, the adjusted aROC was 0.712 for a 5-year observation period. Using a cut-off point of –1.2133, the sensitivity was 56.6% and the specificity was 71.4%. Sensitivity and specificity at other selected cut-off points are listed in Table 4. Cumulative incidences of all-site cancer in patients with a risk score at or above and below the cut-off point are shown in Fig. 3.
|
|
|
In the combined database, the HR of patients with the risk scores at or above the suggested cut-off point of –1.2133 compared with those with risk score less than the cut-off point were as follows:
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The nature of associations between cancer and lipids remains controversial. Large epidemiological studies of general population suggest an inverse total cholesterol–cancer relationship (Schatzkin et al. 1987, Schuit et al. 1993, Steenland et al. 1995). In support of these findings, we further identified the threshold value of total cholesterol <4.30 mmol/l for the inverse total cholesterol–cancer relationship in T2DM. Schuit et al. (1993) proposed several hypotheses for this inverse relationship: 1) low total cholesterol may enhance carcinogenesis by affecting the integrity of cellular membranes; 2) both low total cholesterol and predisposition to cancers are expressions of an underlying, presumably genetic factor; 3) total cholesterol may exert a conditional influence (acting indirectly via hormonal or immunological mechanisms) or promote the action of an already initiated carcinogenic process. Other authors speculated that low total cholesterol might be attributable to cholesterol-lowering action of preclinical cancer (International Collaborative Group 1982, Sherwin et al. 1987). In our cohort, the significance of total cholesterol <4.3 mmol/l for cancer persisted after excluding patients with a follow-up period of <2.5 years. Large cohort studies with long follow-up (10 years) showed that the inverse total cholesterol–cancer relationship remained significant for measurements taken 5 or more years before diagnosis of cancer (Schatzkin et al. 1987, Schuit et al. 1993). Taken together, our findings suggest that the inverse total cholesterol–cancer relationship in general population is also applicable to T2DM with a threshold value around 4.3 mmol/l.
Findings relating to associations of HDL-C and triglyceride with cancer have been inconsistent. In a study of 38 823 Norwegian women aged 17–54 years with a median follow-up period of 17.2 years, women with HDL-C above 1.64 mmol/l were at decreased risk of breast cancer than their counterparts with a HDL level of 1.20 mmol/l and below (Furberg et al. 2004). In another study of 5207 Danish women followed for 26 years, inverse association between HDL-C and breast cancer had also been reported (Hoyer & Engholm 1992). Other investigators reported that breast cancer incidence covaried with triglyceride levels in ex-smokers (Manjer et al. 2001), although others failed to detect a significant overall negative relationship between serum triglycerides and breast cancer risk (Vatten & Foss 1990) or cancer mortality (Cowan et al. 1990, Bahl et al. 2005). Our study suggests that both high and low HDL-C were associated with cancer, especially high HDL-C for longer term cancer risk. In all the above studies, the HR curves of lipids for cancer were not examined. The non-linear nature of lipid association with cancer risk may explain previous negative reports where lipids were analysed either as continuous variables or categorical ones. Despite the significant associations on risk curve analysis, HDL-C and triglyceride were not selected into the cancer risk score possibly due to unsatisfactory data transformation, weak effects or small sample size.
In the Second National Health and Nutrition Examination Survey (Erlinger et al. 2004), increased WBC was significantly associated with high total cancer mortality. On the other hand, in a Chinese population, low lymphocyte count was associated with total cancer mortality (Huang et al. 2003). In this study, there was a tendency for WBC count to exhibit a U-shaped risk association with rapid increase in cancer risk as WBC count fell below <5.8x109 counts per litre. There was a trend for high WBC above 10x109 counts per litre to be associated with increased risk of cancer, albeit short of significance.
Obesity was considered to be a risk factor for cancers including colon, liver, breast, pancreatic and kidney (Calle & Kaaks 2004, Fryzek et al. 2005, Frezza et al. 2006). Conversely, low BMI was associated with increased risk of esophageal (Engeland et al. 2004), oral cavity and oropharyngeal cancer (Nieto et al. 2003). In this study, we found that BMI<24 kg/m2 and 27.6 kg/m2 were both associated with increased risk of cancer compared with patients with a BMI level 24 to <27.6 kg/m2.
The nature of the close links between T2DM and increased cancer risk remains unclear, although it has long been recognized that lipoproteins exhibit regulatory effects on immune responses, immune mediation pathways, cellular metabolism of the immune system and antigen-nonspecific host resistance mediated by the lymphoid and reticuloendothelial systems. Other supporting evidence includes differences in lipoproteins in cancer patients (Edgington & Curtiss 1981). In the early 1980s, Vitale & Broitman (1981) hypothesized that, ‘If tumour cells or transformed cells (premalignant) are no longer regulated in terms of cholesterol, then changes in the concentration of lipids may enhance either the growth or transformation of aberrant cells to malignant cells by further deregulating cholesterol metabolism’. Against this background, the predictive values of low total cholesterol and WBC count for cancer in our analysis support the possible causal role of abnormal lipid metabolism and immune responses, frequently observed in T2DM, in the development of cancer. While experimental studies are needed to test these hypotheses, further investigations are also needed to examine whether these lipid–WBC–cancer relationships can also be generalized to non-diabetic populations.
In this analysis, we further confirmed the utility of these risk relationships by developing and validating an all-site cancer risk score with moderate discrimination and calibration. The predicted probability was similar to the actual probability in the test dataset, suggesting an adequate calibration of the risk score. The Framingham coronary heart disease risk score achieved a c statistics of 0.79 for men and 0.83 for women in their original population, 0.63–0.75 for men and 0.66–0.83 for women in separate USA populations (D'Agostino et al. 2001) and 0.71 for men and 0.74 for women in Chinese populations (Liu et al. 2004). Using the Hong Kong Diabetes Registry, we have developed various risk scores to predict stroke (aROC=0.77) (Yang et al. 2007b), end-stage renal disease risk (aROC>0.94) (Yang et al. 2006b, 2007c), coronary heart disease (aROC=0.74) (Yang et al. 2008a), heart failure (aROC=0.88) (Yang et al. 2006a) and all-cause death (aROC=0.85) (Yang et al. 2008b). Our current all-cancer risk score has an aROC of 0.71, which is higher than that of 0.66 for a breast cancer risk scores using the Gail model (Barlow et al. 2006, Chen et al. 2006). However, compared with other risk scores, cancer remains the least predictable endpoint in T2DM.
Our subgroup analysis suggests that cancer risk score may have different predicting abilities for the subtypes of cancer. For some of these cancers, the risk scores may be used in combination with other validated screening tests to help clinicians make informed decisions. Besides, there were large differences in the HR of the risk score for cancer subtypes, suggesting that population differences in cancer patterns and aetiologies may further influence the performance of the risk score. Thus, the risk score may not be readily extrapolated from one population to another, and recalibration of the probability and/or adjustment of coefficients might be needed.
This study has several limitations. First, the current risk score only aims to predict all types of cancer. Site-specific cancer risk scores may have better performance, although due to sample size limitations, we did not develop such risk scores. Second, BMI, HDL-C and triglyceride which exhibited significant curved relationships were not selected into the predicting model. Their non-selection was likely due to our inability to devise adequate formulae to mimic their sharply curving shapes. Third, the overall C index or aROC is not optimal and that additional biomarkers need to be discovered for inclusion to improve its discrimination.
In conclusion, based on a relatively large prospective cohort of Hong Kong Chinese patients with T2DM, we detected that low total cholesterol (<4.3 mmol/l), high and low HDL-C (<0.9 or 1.8 mmol/l), triglyceride range of 0.81–1.41 mmol/l and low WBC (<5.8x109 count per litre) were associated with increased risk of all-site cancer. Using total cholesterol and WBC, we were able to develop an all-site cancer risk score with moderate performance in the test dataset. In the light of the increased risk for multiple-site cancer in diabetic patients and continuing progress in the screening strategies and treatment of cancers, especially if detected early, the external validation of the risk score in other diabetic populations will be of public health importance.
|
Acknowledgements |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Anderson KM, Odell PM, Wilson PW & Kannel WB 1991 Cardiovascular disease risk profiles. American Heart Journal 121 293–298.[CrossRef][Web of Science][Medline]
Bahl M, Ennis M, Tannock IF, Hux JE, Pritchard KI, Koo J & Goodwin PJ 2005 Serum lipids and outcome of early-stage breast cancer: Results of a Prospective Cohort Study. Breast Cancer Research and Treatment 94 135–144.[CrossRef][Web of Science][Medline]
Barlow WE, White E, Ballard-Barbash R, Vacek PM, Titus-Ernstoff L, Carney PA, Tice JA, Buist DS, Geller BM, Rosenberg R et al. 2006 Prospective breast cancer risk prediction model for women undergoing screening mammography. Journal of National Cancer Institute 98 1204–1214.
Calle EE & Kaaks R 2004 Overweight, obesity and cancer: epidemiological evidence and proposed mechanisms. Nature Review Cancer 4 579–591.[CrossRef]
Chambless LE & Diao G 2006 Estimation of time-dependent area under the ROC curve for long-term risk prediction. Statistics in Medicine 25 3474–3486.[CrossRef][Web of Science][Medline]
Chen J, Pee D, Ayyagari R, Graubard B, Schairer C, Byrne C, Benichou J & Gail MH 2006 Projecting absolute invasive breast cancer risk in white women with a model that includes mammographic density. Journal of National Cancer Institute 98 1215–1226.
Cowan LD, O'Connell DL, Criqui MH, Barrett-Connor E, Bush TL & Wallace RB 1990 Cancer mortality and lipid and lipoprotein levels. Lipid Research Clinics Program Mortality Follow-up Study. American Journal of Epidemiology 131 468–482.
D'Agostino RB Sr, Grundy S, Sullivan LM & Wilson P 2001 Validation of the Framingham coronary heart disease prediction scores: results of a multiple ethnic groups investigation. JAMA 286 180–187.
Edgington TS & Curtiss LK 1981 Plasma lipoproteins with bioregulatory properties including the capacity to regulate lymphocyte function and the immune response. Cancer Research 41 3786–3788.
Engeland A, Tretli S & Bjorge T 2004 Height and body mass index in relation to esophageal cancer; 23-year follow-up of two million Norwegian men and women. Cancer Causes and Control 15 837–843.[CrossRef][Web of Science][Medline]
Erlinger TP, Muntner P & Helzlsouer KJ 2004 WBC count and the risk of cancer mortality in a national sample of US adults: results from the Second National Health and Nutrition Examination Survey mortality study. Cancer Epidemiology, Biomarkers and Prevention 13 1052–1056.
Frezza EE, Wachtel MS & Chiriva-Internati M 2006 Influence of obesity on the risk of developing colon cancer. Gut 55 285–291.
Friberg E, Orsini N, Mantzoros CS & Wolk A 2007 Diabetes mellitus and risk of endometrial cancer: a meta-analysis. Diabetologia 50 1365–1374.[CrossRef][Medline]
Fryzek JP, Schenk M, Kinnard M, Greenson JK & Garabrant DH 2005 The association of body mass index and pancreatic cancer in residents of southeastern Michigan 1996–1999. American Journal of Epidemiology 162 222–228.
Furberg AS, Veierod MB, Wilsgaard T, Bernstein L & Thune I 2004 Serum high-density lipoprotein cholesterol, metabolic profile, and breast cancer risk. Journal of National Cancer Institute 96 1152–1160.
Harrell F Regression Modelling Strategies with applications to Linear Models, Logistic Regression, and Survival Analysis. 2001 Spinger-Varlag New York, Inc New York:
Hoyer AP & Engholm G 1992 Serum lipids and breast cancer risk: a cohort study of 5,207 Danish women. Cancer Causes and Control 3 403–408.[CrossRef][Web of Science][Medline]
Huang ZS, Chien KL, Yang CY, Wang CH, Chang TC & Chen CJ 2003 Peripheral differential leukocyte counts and subsequent mortality from all diseases, cancers, and cardiovascular diseases in Taiwanese. Journal of the Formosan Medical Association 102 775–781.[Web of Science][Medline]
Huxley R, Ansary-Moghaddam A, Berrington de Gonzalez A, Barzi F & Woodward M 2005 Type-II diabetes and pancreatic cancer: a meta-analysis of 36 studies. British Journal of Cancer 92 2076–2083.[CrossRef][Web of Science][Medline]
International Collaborative Group Circulating cholesterol level and risk of death from cancer in men aged 40 to 69 years. Experience of an international collaborative group JAMA 248 1982 2853–2859.
Laakso M & Pyorala K 1985 Age of onset and type of diabetes. Diabetes Care 8 114–117.[Abstract]
Lin DY, Wei LJ & Ying Z 1993 Checking the cox model with cumulative sums of martingale-based residuals. Biometrika 80 557–572.
Lipscombe LL, Goodwin PJ, Zinman B, McLaughlin JR & Hux JE 2006 Diabetes mellitus and breast cancer: a retrospective population-based cohort study. Breast Cancer Research 98 349–356.[CrossRef][Web of Science]
Liu J, Hong Y, D'Agostino RB Sr, Wu Z, Wang W, Sun J, Wilson PW, Kannel WB & Zhao D 2004 Predictive value for the Chinese population of the Framingham CHD risk assessment tool compared with the Chinese Multi-Provincial Cohort Study. JAMA 291 2591–2599.
Ma YC, Zuo L, Chen JH, Luo Q, Yu XQ, Li Y, Xu JS, Huang SM, Wang LN, Huang W et al. 2006 Modified glomerular filtration rate estimating equation for Chinese patients with chronic kidney disease. Journal of the American Society of Nephrology 17 2937–2944.
Manjer J, Malina J, Berglund G, Bondeson L, Garne JP & Janzon L 2001 Breast cancer incidence in ex-smokers in relation to body mass index, weight gain and blood lipid levels. European Journal of Cancer Prevention 10 281–287.[CrossRef][Web of Science][Medline]
Manson J, Colditz G, Stampfer M, Willett W, Krolewski A, Rosner BA, Arky RA, Speizer F & Hennekens C 1991 A prospective study of maturity-onset diabetes mellitus and risk of coronary heart disease and stroke in women. Archives of Internal Medicine 151 1141–1147.
Nieto A, Sanchez MJ, Martinez C, Castellsague X, Quintana MJ, Bosch X, Conde M, Munoz N, Herrero R & Franceschi S 2003 Lifetime body mass index and risk of oral cavity and oropharyngeal cancer by smoking and drinking habits. British Journal of Cancer 89 1667–1671.[CrossRef][Web of Science][Medline]
Pencina MJ & D'Agostino RB 2004 Overall C as a measure of discrimination in survival analysis: model specific population value and confidence interval estimation. Statistics in Medicine 23 2109–2123.[CrossRef][Web of Science][Medline]
Piwernetz K, Home PD, Snorgaard O, Antsiferov M, Staehr-Johansen K & Krans M 1993 For the DiabCare Monitoring Group of the St. Vincent Declaration Steering Committee. Monitoring the targets of the St. Vincent declaration and the implementation of quality management in diabetes care: the DiabCare initiative. Diabetic Medicine 10 371–377.[Web of Science][Medline]
Rodriguez C, Patel AV, Mondul AM, Jacobs EJ, Thun MJ & Calle EE 2005 Diabetes and risk of prostate cancer in a prospective cohort of US men. American Journal of Epidemiology 161 147–152.
Rousseau MC, Parent ME, Pollak MN & Siemiatycki J 2006 Diabetes mellitus and cancer risk in a population-based case–control study among men from Montreal, Canada. International Journal of Cancer 118 2105–2109.[CrossRef][Web of Science][Medline]
Schatzkin A, Hoover RN, Taylor PR, Ziegler RG, Carter CL, Larson DB & Licitra LM 1987 Serum cholesterol and cancer in the NHANES I epidemiologic followup study. National Health and Nutrition Examination Survey. Lancet 2 298–301.[Web of Science][Medline]
Schuit AJ, Van Dijk CE, Dekker JM, Schouten EG & Kok FJ 1993 Inverse association between serum total cholesterol and cancer mortality in Dutch civil servants. American Journal of Epidemiology 137 966–976.
Seow A, Yuan JM, Koh WP, Lee HP & Yu MC 2006 Diabetes mellitus and risk of colorectal cancer in the Singapore Chinese Health Study. Journal of National Cancer Institute 98 135–138.
Sherwin RW, Wentworth DN, Cutler JA, Hulley SB, Kuller LH & Stamler J 1987 Serum cholesterol levels and cancer mortality in 361,662 men screened for the multiple risk factor intervention trial. JAMA 257 943–948.
So WY, Yang X, Ma RC, Kong AP, Lam CW, Ho CS, Cockram CS, Ko GT, Chow CC & Wong V et al. 2008 Risk factors in V-shaped risk associations with all-cause mortality in type 2 diabetes – The Hong Kong Diabetes Registry. Diabetes/Metabolism Research and Reviews 24 238–246.
Steenland K, Nowlin S & Palu S 1995 Cancer incidence in the National Health and Nutrition Survey I. Follow-up data: diabetes, cholesterol, pulse and physical activity. Cancer Epidemiology, Biomarkers and Prevention 4 807–811.[Abstract]
Vatten LJ & Foss OP 1990 Total serum cholesterol and triglycerides and risk of breast cancer: a prospective study of 24,329 Norwegian women. Cancer Research 50 2341–2346.
Vitale JJ & Broitman SA 1981 Lipids and immune function. Cancer Research 41 3706–3710.
Wild S, Roglic G, Green A, Sicree R & King H 2004 Global prevalence of diabetes: estimates for the year 2000 and projections for 2030. Diabetes Care 27 1047–1053.
Wu AY, Kong NC, de Leon FA, Pan CY, Tai TY, Yeung VT, Yoo SJ, Rouillon A & Weir MR 2005 An alarmingly high prevalence of diabetic nephropathy in Asian type 2 diabetic patients: the MicroAlbuminuria Prevalence (MAP) Study. Diabetologia 48 17–26.[CrossRef][Web of Science][Medline]
Yang XL, Kong PS, So WY, Ma RC, Cockram CS, Chow CC, Tong PC & Chan JC 2006a Risk scores for predicting hard coronary heart disease and nonfatal heart failure in Hong Kong Chinese type 2 diabetic patients – The Hong Kong Diabetes Registry. The 19th World Diabetes Congress. Cape Town South Africa:
Yang XL, So WY, Kong AP, Clarke P, Ho CS, Lam CW, Ng MH, Lyu RR, Yin DD, Chow CC et al. 2006b End-stage renal disease risk equations for Hong Kong Chinese patients with type 2 diabetes: Hong Kong Diabetes Registry. Diabetologia 49 2299–2308.
Yang X, Ma RC, So WY, Ko GT, Kong AP, Lam CW, Ho CS, Cockram CS, Wong VC, Tong PC et al. 2007a Impacts of chronic kidney disease and albuminuria on associations between coronary heart disease and its traditional risk factors in type 2 diabetic patients – the Hong Kong diabetes registry. Cardiovascular Diabetology 6 37[CrossRef][Medline]
Yang XL, So WY, Kong AP, Ho CS, Lam CW, Stevens RJ, Lyu RR, Yin DD, Cockram CS, Tong PC et al. 2007b Development and validation of stroke risk equation for Hong Kong Chinese patients with type 2 diabetes: the Hong Kong Diabetes Registry. Diabetes Care 30 65–70.
Yang XL, So WY, Kong AP, Ho CS, Lam CW, Ng MH, Lyu RR, Yin DD, Chow CC, Cockram CS et al. 2007c Modified end-stage renal disease risk score for Chinese type 2 diabetic patients – the Hong Kong Diabetes Registry. Diabetologia 50 1348–1350.
Yang X, So WY, Kong AP, Ma RC, Ko GT, HO CS, Lam CW, Cockram CS, Chan JC & Tong PC 2008a Development and validation of a total coronary heart disease risk score in type 2 diabetes mellitus. American Journal of Cardiology 101 596–601.[CrossRef][Web of Science][Medline]
Yang X, So WY, Tong PC, Ma RC, Kong AP, Lam CW, Ho CS, Cockram CS, Ko GT, Chow CC et al. 2008b Development and validation of an all-cause mortality risk score in type 2 diabetes – The Hong Kong Diabetes Registry. Archives of Internal Medicine 168.
This article has been cited by other articles:
|
|
 |
|
  X. Yang, W. Y. So, R. C.W. Ma, G. T.C. Ko, A. P.S. Kong, H. Zhao, A. O.Y. Luk, C. W.K. Lam, C. S. Ho, P. C.Y. Tong, et al. Low LDL Cholesterol, Albuminuria, and Statins for the Risk of Cancer in Type 2 Diabetes: The Hong Kong Diabetes Registry Diabetes Care, October 1, 2009; 32(10): 1826 - 1832. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 X. Yang, H. Zhao, Y. Sui, R. C.W. Ma, W. Y. So, G. T.C. Ko, A. P.S. Kong, R. Ozaki, C. Y. Yeung, G. Xu, et al. Additive Interaction Between the Renin-Angiotensin System and Lipid Metabolism for Cancer in Type 2 Diabetes Diabetes, July 1, 2009; 58(7): 1518 - 1525. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
