SEARCH

SEARCH BY CITATION

Keywords:

  • KEYWORDS;
  • triglyceride;
  • cancer;
  • statin;
  • type 2 diabetes mellitus;
  • biological interaction

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

BACKGROUND:

Patients with type 2 diabetes mellitus (T2DM) have increased cancer risks. The authors reported nonlinear associations of cancer with triglyceride and other lipids in T2DM. Crosstalk between lipid metabolism and the renin-angiotensin system may increase cancer risk via activation of insulin-like growth factor-1 pathway in T2DM. In this analysis, the authors explored associations of cancer risk with high/low triglyceride in T2DM and possible modifying effects of statins on this risk association, if any.

METHODS:

A consecutive cohort of 5166 Chinese patients with T2DM, free of cancer at enrollment and not using statins at or before enrollment, was analyzed using Cox models. Biological interactions were estimated using relative excess risk because of interaction, attributable proportion because of interaction, and synergy index. Relative excess risk because of interaction >0, attributable proportion because of interaction >0, or synergy index >1 indicates biological interaction.

RESULTS:

During 5.25 years of follow-up (median), 4.7% (n = 243) patients developed cancer. Triglyceride <1.70 mmol/L was associated with increased cancer risk in the entire cohort and in statin nonusers, but not in statin users. Patients with triglyceride <1.70 mmol/L plus nonuse of statins during follow-up had 2.74-fold increased cancer risk compared with their counterparts with either triglyceride ≥1.70 mmol/L or use of statins or both. There was significant interaction between triglyceride <1.70 mmol/L and nonuse of statins (relative excess risk because of interaction, 0.99; 95% confidence interval [CI], 0.07-1.90 and attributable proportion because of interaction, 0.36; 95% CI, 0.02-0.70).

CONCLUSIONS:

In Chinese T2DM patients, triglyceride <1.70 mmol/L might be associated with increased cancer risk, which was attenuated in the presence of use of statins. Cancer 2011. © 2010 American Cancer Society.

Diabetes is among the most prevalent diseases in the world, with a rapid increase in Asian populations in recent years.1 In addition to cardiovascular disease, diabetes also predisposes to a variety of cancers.2 In Chinese patients with type 2 diabetes mellitus (T2DM), the incidence of cancer was ⅓ higher than that in the general population.3 In a series of investigations, our group reported the nonlinear risk associations of cancer with lipids exemplified by a V-shaped relationship for low-density lipoprotein (LDL) cholesterol and high-density lipoprotein (HDL) cholesterol3, 4 and an A-shaped relationship with triglyceride with a zenith at 1.17 mmol/L.4 On the basis of these findings, we further observed that the cancer risk association of low LDL cholesterol depended on the presence of albuminuria, which was attenuated by use of statins.5 By using clinical and experimental studies, we reported crosstalk between lipid metabolism and the renin-angiotensin system. This interaction may increase cancer risk via activation of insulin-like growth factor (IGF)-1, Akt, and cholesterol biosynthesis pathways in T2DM.6 Given the intimate relationships between cholesterol and free fatty acid (triglyceride) metabolism, we used triglyceride ≥1.70 mmol/L, 1 of the diagnostic criteria for metabolic syndrome, as a cutoff value7 and asked the questions: 1) whether high and low triglyceride levels increased risk of cancer as in the case of LDL cholesterol3; and 2) if so, whether use of lipid-lowering drugs might modify these risks.5

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

Patients

The study used data of the Hong Kong Diabetes Registry, which was established at the Prince of Wales Hospital, the teaching hospital of the Chinese University of Hong Kong. The hospital serves a population of >1.2 million. The referral sources of the cohort included general practitioners, community clinics, other specialty clinics, the Prince of Wales Hospital itself, and other hospitals. The 1998 World Health Organization criteria (the 1985 criteria were used before July 1998) were used to diagnose diabetes.8 Enrolled patients with hospital admissions within 6 to 8 weeks before assessment accounted for <10% of all referrals. A 4-hour assessment of complications and risk factors was performed on an outpatient basis, modified from the European DiabCare protocol.9 Once a patient had undergone this comprehensive assessment, he/she was considered to have entered this study cohort and would be followed until 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 to and written informed consent was obtained from all patients, at the time of assessment, for research purposes.

For this analysis, the clinical endpoints, including discharge diagnoses and mortality from enrollment to July 30, 2005 were recorded, or otherwise censored on July 30, 2005. Details of all medical admissions of the cohort by that date were retrieved from the Hong Kong Hospital Authority Central Computer System, which recorded admissions to all public hospitals in Hong Kong. Collectively, these hospitals provide 95% of the total hospital bed-days in Hong Kong.10 Mortality data from the Hong Kong Death Registry was also retrieved and cross-checked with hospital discharge status. Drug use data were extracted from the Hospital Authority computer system that recorded all drug dispensary data in public hospitals including the start dates and end dates for each of the drugs of interest. In Hong Kong, all medications are dispensed onsite in both inpatient and outpatient settings. These databases were matched by a unique identification number, the Hong Kong Identity Card number, which is compulsory for all residents in Hong Kong.

From 1996 to 2005, 7387 diabetic patients were enrolled in the registry. After sequential exclusion of 328 patients with type 1 diabetes mellitus (defined as acute presentation with diabetic ketoacidosis, heavy ketonuria [>3+], or continuous requirement of insulin within 1 year of diagnosis) or missing data on types of diabetes, 45 with non-Chinese or unknown nationality, 175 with a known history of cancer or receiving cancer treatment at enrollment, and 736 with missing values on any variables used in the analysis (see Table 1 for a list of variables), a total of 5166 patients were used in the analysis, after further exclusion of 937 patients who had ever used statins at or before enrollment to remove potential prevalent user bias (Fig. 1).

thumbnail image

Figure 1. Patient flow chart is shown.

Download figure to PowerPoint

Table 1. Clinical and Biochemical Characteristics of the Study Cohort of Type 2 Diabetic Patients Stratified According to Occurrence of Cancer During Follow-Up Period
 Noncancer (n=4923)Cancer (n=243) 
 Median (25th-75th) or n (%)Median (25th-75th) or n (%)P value
  • LDL-C indicates low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol; BP, blood pressure; ACR, spot urine albumin:creatinine ratio; eGFR, estimated glomerular filtration rate; ACEIs, angiotensin-converting enzyme inhibitors; ARBs, angiotensin II receptor blockers.

  • a

    Derived from Wilcoxon Two-Sample test.

  • b

    Derived from Chi-square test.

  • c

    From enrolment to the earliest date of cancer, death or censoring.

  • d, a

    Fisher exact test.

Baseline variables   
Age, years56 (45-67)66 (57-72)<.0001a
Male gender2290 (46.5)1230 (53.5).0323b
Smoking status  <.0001b
 Ex-smoker718 (14.6)51 (21.0) 
 Current smoker764 (15.5)59 (24.3) 
Alcohol intake  <.0001b
 Ex-drinker582 (11.8)52 (21.4) 
 Current drinker373 (7.6)20 (8.2) 
Body mass index, kg/m224.7 (22.4-27.3)24.3 (22.4-27.0).2316a
Duration of diabetes, years6 (2-11)7 (2-12).0293a
Systolic BP, mmHg134 (121-147)136 (126-150).0013a
Diastolic BP, mmHg75 (68-82)75 (68-83).7085a
Glycated hemoglobin, %7.2 (6.3-8.4)7.3 (6.4-8.6).5292a
LDL-C, mmol/L3.13 (2.60-3.79)3.10 (2.50-3.80).3458a
HDL-C, mmol/L1.25 (1.06-1.51)1.25 (1.01-1.54).5511a
Triglyceride, mmol/L1.30 (0.91-1.91)1.21 (0.87-1.64).0118a
 Triglyceride<1.70 mmol/L3339 (67.8)186 (76.5).0044b
Total cholesterol, mmol/L5.17 (4.50-5.80)5.00 (4.38-5.80).1485a
ACR (mg/mmol)1.9 (0.8-9.3)3.2 (0.9-16.2).0003a
 Micro-albuminuria1298 (26.4)81 (33.3).0129b
 Macro-albuminuria749 (15.2)43 (17.9) 
eGFR, ml min−1 1.73 m−2106 (86.4-127.2)99 (79-117).0006a
 eGFR<60 ml min−1 1.73 m−21454 (29.5)101 (41.1).0001b
Medications at enrolment   
Antihypertensive drugs other than ACEIs or ARBs1698 (34.5)110 (45.3).0006b
Events and medications during follow-upc   
Death (all-cause)308 (6.3)124 (51.0)<.0001b
ACEIs or ARBs2686 (54.6)123 (50.6).2283b
Statins1381 (28.1)34 (14.0)<.0001b
Fibrates493 (10.0)14 (5.7).0296b
Acarbose567 (11.5)25 (10.3).5570b
Glibenclamide1390 (28.2)62 (25.5).3571b
Gliclazide2238 (45.5)89 (36.6).0069b
Glimepiride66 (1.3)2 (0.8).7710d
Glipizide571 (11.6)46 (18.9).0006b
Metformin3690 (75.0)150 (61.7)<.0001b
Pioglitazone53 (1.1)0% (0).1787d
Rosiglitazone267 (5.4)2 (0.8).0016b
Tolbutamide43 (0.9)1 (0.4).7222d

Clinical and Laboratory Measurements

Details of assessment methods and definitions have been described.3 Patients attended the center after 8 hours of fasting to undergo clinical assessments and laboratory investigations. A sterile, random spot urinary sample was used to measure albumin to creatinine ratio (ACR). Microalbuminuria was defined as ACR ≥2.5 mg/mmol in men and ≥3.5 mg/mmol in women. The abbreviated Modification of Diet in Renal Disease Study formula recalibrated for Chinese11 was used to estimate glomerular filtration rate (GFR) expressed in mL/min/1.73 m2: estimated GFR = 186 × (SCR × 0.011)−1.154 × (age)−0.203 × (0.742 if female) × 1.233, where SCR is serum creatinine expressed as μmol/L (original mg/dL converted to μmol/L), and 1.233 is the adjusting coefficient for Chinese. Lipids (total cholesterol, triglyceride, and HDL cholesterol) were measured by enzymatic methods on a Hitachi (Hitachi Instrument Division, Ibaraki-ken, Japan) 911 automated analyzer, using reagent kits supplied by the manufacturer of the analyzer. LDL cholesterol was calculated using Friedewald's equation.12 The precision performance of these assays was within the manufacturer's specifications.

Definition of Cancer

A trained team at the Hospital Authority of Hong Kong coded all hospital admissions. Hospital discharge principle diagnoses, coded by the International Classification of Diseases, ninth revision, were used to identify cancer events. The endpoint of this study was defined as having incident cancer (either fatal or nonfatal; codes 140-208) during the follow-up period.

Statistical Analyses

The Statistical Analysis System (release 9.10) was used to perform all statistical analysis (SAS Institute, Cary, NC). Follow-up time was calculated as the period in years from the first enrollment since December 1, 1996 to the date of the first cancer event, death, or censoring, whichever came first. Cox proportional hazard regression was used to obtain hazard ratios (HRs) and 95% confidence intervals (CIs) of triglyceride <1.70 mmol/L versus ≥1.70 mmol/L in the entire cohort and in statin users and nonusers, separately. A structured adjustment scheme was used to adjust for covariates. First, we adjusted for age, smoking status, HbA1c, body mass index (BMI), LDL cholesterol-related risk indictors (LDL cholesterol <2.80 mmol/L plus albuminuria and LDL cholesterol ≥3.80 mmol/L),5 and HDL cholesterol. As BMI and HDL cholesterol were associated with cancer in nonlinear manners, we used restricted cubic splines with 4 knots at the 5th, 35th, 65th, and 95th percentiles of BMI and HDL cholesterol to adjust for their confounding effects.4 Second, we adjusted for covariates with P values <.10 (see Table 2 for the list of covariates, including alcohol intake and duration of diabetes).

Table 2. Clinical and Biochemical Characteristics of the Study Cohort of Type 2 Diabetic Patients Stratified by Triglyceride < 1.70 mmol/L and Use of Statins During Follow-Up Period
 Nonstatin UsersStatin Users  
 TG≥1.70 mmol/La (n=1014)TG<1.70 mmol/La (n=2737)PbTG≥1.70 mmol/La (n=627)TG<1.70 mmol/La (n=788)PbPb,cPbd
  • TG indicates triglyceride; LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol; BP, blood pressure; ACR, spot urine albumin:creatinine ratio; eGFR, estimated glomerular filtration rate; ACEIs, angiotensin-converting enzyme inhibitors; ARBs, angiotensin II receptor blockers.

  • a

    Median (interquartile range) or n (%).

  • b

    Derived from Wilcoxon Two-Sample test, Chi-square test, or Fisher exact test.

  • c

    P for comparison between statin users and non-users among those with triglyceride ≥ 1.70 mmol/L.

  • d

    P for comparison between statin users and non-users among those with triglyceride < 1.70 mmol/L.

  • e

    From enrolment to the earliest date of cancer, death or censoring.

Baseline variables        
Age, years56 (23)56 (22).314657 (20)59 (17).0615.4974<.0001
Male gender449 (44.3)1325 (48.3).0244277 (44.2)369 (46.8).3204.9679.4331
Smoking status  .0353  .1427.4738.6923
Ex-smoker125 (12.3)428 (15.6) 83 (13.2)133 (16.9)   
Current smoker159 (15.7)430 (15.7) 110 (17.5)124 (15.7)   
Alcohol intake  .4711  .7121.6529.5102
Ex-drinker113 (11.1)340 (12.4) 76 (12.1)105 (13.3)   
Current drinker85 (8.4)210 (7.7) 46 (7.3)52 (6.6)   
Body mass index, kg/m225.8 (5.2)24.1 (4.6)<.000125.8 (5.0)24.4 (4.8)<.0001.4018.2525
Duration of diabetes, years5 (9)6 (8).02007 (10)8 (10).0189<.0001<.0001
Systolic BP, mmHg135 (26)131 (25)<.0001139 (26)137 (29).0017<.0001<.0001
Diastolic BP,mmHg77 (14)73 (13)<.000178 (15)75 (14)<.0001.0028.0055
Glycated hemoglobin, %7.3 (1.9)7.0 (2.0)<.00017.8 (2.1)7.6 (2.4).0999<.0001<.0001
LDL-C, mmol/L3.00 (1.16)2.90 (1.04).00173.73 (1.20)3.71 (1.10).6805<.0001<.0001
HDL-C, mmol/L1.10 (0.35)1.35 (0.48)<.00011.12 (0.35)1.31 (0.46)<.0001.0606.0659
Total cholesterol, mmol/L5.25 (1.22)4.80 (1.12)<.00016.07 (1.34)5.61 (1.47)<.0001<.0001<.0001
Urinary ACR, mg/mmol2.6 (10.9)1.4 (4.8)<.00015.7 (41.0)2.9 (15.3)<.0001<.0001<.0001
 Micro-albuminuria307 (30.3)681 (24.9)<.0001260 (41.5)401 (50.9)<.0001<.0001<.0001
 Macro-albuminuria167 (16.5)262 (9.6) 170 (27.1)221 (28.1)   
eGFR, ml min-1 1.73 m-2104 (42)109 (40)<.000195 (45)103 (40).0002<.0001<.0001
eGFR<60325 (32.1)694 (25.4)<.0001268 (42.7)268 (34.0).0008<.0001<.0001
Medications at enrolment        
Antihypertensive drugs other than ACEIs or ARBs392 (38.7)858 (31.4)<.0001267 (42.6)291 (36.9).0306.1150.0032
Events during follow-up        
Cancer43 (4.2)166 (6.1).030514 (2.2)20 (2.5).7096.0309<.0001
Death (all-cause)84 (8.3)250 (9.1).416937 (5.9)61 (7.7).1757.0727.2244
Medications during follow-upe        
ACEIs or ARBs544 (53.7)1212 (44.3)<.0001490 (78.2)563 (71.5).0041<.0001<.0001
Fibrates161 (15.9)109 (4.0)<.0001161 (25.7)76 (9.6)<.0001<.0001<.0001
Acarbose106 (10.5)247 (9.0).1831117 (18.7)122 (15.5).1130<.0001<.0001
Glibenclamide294 (29.0)747 (27.3).3013178 (28.4)233 (29.6).6274.7925.2089
Gliclazide442 (43.6)1168 (42.7).6150334 (53.3)383 (48.6).0812<.0001.0031
Glimepiride12 (1.2)18 (.7).108414 (2.2)24 (3.1).3475.0981<.0001
Glipizide114 (11.2)294 (10.7).661793 (14.8)116 (14.7).9531.0333.0021
Metformin777 (76.6)1893 (69.2)<.0001546 (87.1)624 (79.2)<.0001<.0001<.0001
Pioglitazone11 (1.1)14 (0.5).055313 (2.1)15 (1.9).8198.1050.0001
Rosiglitazone35 (3.5)71 (2.6).159278 (12.4)85 (10.8).3332<.0001<.0001
Tolbutamide8 (0.8)23 (0.8).87735 (0.8)8 (1.0).6697.9850.6431
Time using statins, years   2.17 (3.18)1.90 (3.04).1592  

To ascertain whether use of statins conferred clinical benefits in patients with triglyceride <1.70 mmol/L, we examined biological interaction between triglyceride <1.70 mmol/L and nonuse of statins after enrollment, to the earliest date of incident cancer, death, or 30 July 30, 2005, whichever came first.13, 14 We used 3 measures to estimate biological interaction: 1) relative excess risk because of interaction; 2) attributable proportion because of interaction; and 3) synergy index. The relative excess risk because of interaction is the excess risk because of interaction relative to the risk without exposure. Attributable proportion because of interaction refers to the attributable proportion of disease that is because of interaction in persons with both exposures. Synergy index is the excess risk from both exposures when there is a biological interaction, relative to the risk from both exposures without interaction.15 Relative excess risk because of interaction >0, attributable proportion because of interaction >0, or synergy index >1 indicates biological interaction. In Cox models, relative excess risk because of interaction is the best among the 3 measures.16 In addition, Kaplan-Meier was used to examine the cumulative incidences of cancer in patients stratified by triglyceride levels (ie, <1.70 mmol/L vs ≥1.70 mmol/L) and statin use status (ie, use vs nonuse) (Statistical Package for the Social Sciences for Windows, Release 13.0, SPSS Inc, Chicago, Ill).

The same 2-step adjustment scheme was used to control for confounding effects of other covariates. In addition, we calculated the propensity score (probability) of initiation of statins during the follow-up period using a stepwise logistic procedure (P = .30).17 BMI, LDL cholesterol, triglyceride, HbA1c, natural logarithm (ACR + 1), duration of diabetes, retinopathy, prior myocardial infarction, and prior stroke were selected to calculate the propensity score (c statistic = 0.80). We then used stratified Cox models on deciles of the score to adjust for probability of initiation of statins during follow-up, which was automatically exported from the SAS procedure of logistic regression. Proportionality was checked using either the Supremum test18 or plots of log negative log (survival distribution function) versus log (follow-up time), depending on whether the covariates were continuous or categorical. Correlations between pairs of baseline covariates were checked using Pearson correlation test, and none of the pairs was highly correlated (correlation coefficient, <0.60).19 A 2-sided P value <.05 was considered to be significant.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

Characteristics of the Patients

At the time of enrollment, the cohort had a median of 56 (25th to 75th percentiles, 46-67) years of age and 6 (25th to 75th percentiles, 2-11) years of duration of disease. During a median of 5.25 (25th to 75th percentiles, 3.02-7.19) years of follow-up, 4.7% (n = 243) patients developed cancer, and the incidence of cancer was 9.34 (95% CI, 8.18-10.51) per 1000 person-years. Compared with those without cancer, patients who developed cancer were older, more likely to be smokers or alcohol drinkers, had longer duration of diabetes, higher blood pressure, and lower triglyceride. They were more likely to have albuminuria and renal dysfunction, but less likely to use statins, fibrates, gliclazide, metformin, and rosiglitazone (Table 1). Among statin nonusers, crude cancer rate was higher in patients with triglyceride <1.70 mmol/L than in those with triglyceride ≥1.70 mmol/L. Among statin users, the crude cancer rates were similar between these 2 groups. When stratified by triglyceride level of 1.70 mmol/L, the crude cancer rates were higher in nonstatin users than statin users irrespective of the stratum, with a greater difference in patients with triglyceride <1.70 mmol/L (Table 2).

Associations Between Low Triglyceride and Cancer

Triglyceride <1.70 mmol/L was associated with increased cancer risk after adjusting for age, smoking status, alcohol intake, HbA1c, BMI, LDL cholesterol-related risk indicators, and HDL cholesterol in the entire cohort. The statistical significance persisted after further adjusting for covariates with P value <.10. The magnitude of HR of low triglyceride was accentuated in nonstatin users (P = .0156) but attenuated in statin users (P = .3887) (Table 3).

Table 3. Hazard Ratios for the Risk of Cancer in Relation to Triglyceride and Statin Use in Type 2 Diabetes
ExposuresNumber at riskHR95% CIP value
  • HR indicates hazard ratio; CI, 95% confidence interval.

  • a

    Adjusted for age, smoking status, alcohol intake, HbA1c, body mass index (BMI), low-density lipoprotein cholesterol-related risk indictors and high-density lipoprotein (HDL) cholesterol, in which spline functions of BMI and HDL cholesterol were used to adjust for nonlinear associations with cancer.

  • b

    Further adjusted for use of statins (for the triglyceride model in the whole cohort), triglyceride (for the statin model in the whole cohort), use of angiotensin-converting enzyme inhibitors/angiotensin II receptor blockers, antihypertensive drugs, gliclazide, metformin and rosiglitazone (P<.10).

  • c

    Cox models stratified on deciles of probability of initiation of statin therapy during follow-up were used in all interactive models.

Triglyceride <1.70 mmol/L as a risk factor for cancer    
 Among the entire cohort    
 Model Onea35251.621.17-2.24.0038
 Model Twob35251.501.08-2.08.0159
 Among statin non-users    
 Model Onea27371.591.10-2.29.0135
 Model Twob27371.581.09-2.28.0156
 Among statin users    
 Model Onea7881.320.61-2.81.4795
 Model Twob7881.410.65-3.07.3887
Use versus non-use of statins for cancerc    
 Among the entire cohort    
 Model Onea14150.370.25-0.55<.0001
 Model Twob14150.450.30-0.67<.0001
 Among triglyceride ≥ 1.70 mmol/L    
 Model Onea6270.400.21-0.77.0060
 Model Twob6270.440.22-0.87.0178
 Among triglyceride < 1.70 mmol/L    
 Model Onea7880.350.21-0.58<.0001
 Model Twob7880.420.25-0.71.0010
Interaction models of triglyceride <1.70 mmol/L and non-use of statinsc    
 Model Onea    
 Triglyceride <1.70 mmol/L and use of statins vs. others7881.150.57-2.33.7017
 Triglyceride ≥1.70 mmol/L and non-use of statins vs. others10142.161.16-4.04.0156
 Triglyceride <1.70 mmol/L and non-use of statins vs. others27373.511.89-6.50<.0001
 Model Twob    
 Triglyceride <1.70 mmol/L and use of statins vs. others7881.060.52-2.15.8671
 Triglyceride ≥1.70 mmol/L and non-use of statins vs. others10141.690.90-3.18.1031
 Triglyceride <1.70 mmol/L and non-use of statins vs. others27372.741.47-5.11.0015

Associations Between Use of Statins and Cancer

In univariate and multivariate Cox model analysis with further adjustment for probability of use of statin during the follow-up period, use of statins was associated with a significantly lower risk of cancer among patients with triglyceride <1.70 mmol/L. To a less extent, use of statins was also associated with lower cancer risk among patients with triglyceride ≥ 1.70 mmol/L in univariable and multivariable Cox model analyses (Table 3).

Interaction of Low Triglyceride and Nonuse of Statins on Cancer Risk

The Kaplan-Meier plot showed that statin nonusers with triglyceride <1.70 mmol/L had the highest cumulative incidence of cancer, followed by statin nonusers with triglyceride ≥1.70 mmol/L. Statin users, regardless of triglyceride levels, had the lowest cumulative incidence of cancer during the 7-year follow-up period (Fig. 2). After adjusting for covariates, patients with triglyceride <1.70 mmol/L plus nonuse of statins had 2.74-fold increased cancer risk compared with their counterparts with either triglyceride ≥1.70 mmol/L or use of statins or both. In multivariate analysis, nonuse of statins alone and triglyceride <1.70 mmol/L alone were not associated with increased cancer risks (Table 3), but there was a significant biological interaction between triglyceride <1.70 mmol/L and nonuse of statins. The multivariate relative excess risk because of interaction and attributable proportion because of interaction were, respectively, 0.99 (95% CI, 0.07-1.90) and 0.36 (95% CI, 0.02-0.70), although synergy index was not significant (Table 4).

thumbnail image

Figure 2. A Kaplan-Meier plot of cancer cumulative incidence is stratified by whether triglyceride (TG) was below 1.70 mmol/L and whether statin therapy was initiated during follow-up (log rank test, P = .0002): (A) statin nonusers whose TG was <1.70 mmol/L (the first curve from the top); (B) statin users whose TG was <1.70 mmol/L (the last curve from the top); (C) statin users whose TG was ≥1.70 mmol/L (the third curve from the top); and (D) statin nonusers whose TG was ≥1.70 mmol/L (the second curve from the top).

Download figure to PowerPoint

Table 4. Biological Interactions of Low Triglyceride and Nonuse of Statins for the Risk of Cancer in Type 2 Diabetes Mellitus
Measures of Biological InteractionEstimate95% CI
  • CI indicates confidence interval; RERI, relative excess risk due to interaction; AP, attributable proportion due to interaction; S, synergy index.

  • a

    Adjusted for age, smoking status, alcohol intake, HbA1c, body mass index (BMI), low-density lipoprotein cholesterol-related risk indictors and high-density lipoprotein (HDL) cholesterol, in which spline functions of BMI and HDL cholesterol were used to adjust for non-linear associations with cancer.

  • b

    Cox models stratified on deciles of probability of initiation of statin therapy during follow-up were used in all interactive models.

  • c

    Statistically significant with RERI >0, AP >0 or S >1 indicating biological interaction.

  • d

    Further adjusted for use of statins (for the triglyceride model in the whole cohort), triglyceride (for the statin model in the whole cohort), use of angiotensin-converting enzyme inhibitors/angiotensin II receptor blockers, antihypertensive drugs, gliclazide, metformin and rosiglitazone (P<.10).

Model Onea,b, ¶  
 RERI1.200.05-2.34c
 AP0.340.04-0.65c
 S1.910.78-4.70
Model Twob,d  
 RERI0.990.07-1.90c
 AP0.360.02-0.70c
 S2.310.54-9.93

Copresence of Low Triglyceride and Nonuse of Statins for the Risk of Site-Specific Cancers

Copresence of triglyceride <1.7 mmol/L and nonuse of statins was associated with increased risk of cancers in digestive organs and peritoneum as well as cancers in sites other than digestive organs and peritoneum. Other site-specific cancers also showed a similar trend, albeit short of significance with a small sample size (Table 5).

Table 5. Univariable Hazard Ratios of Copresence of Low Triglyceride and Nonuse of Statins for Site-Specific Cancers in Type 2 Diabetes
Cancer subtypesaNo. of cancerbHazard Ratio (HR, 95% CI)P value
  • CI indicates confidence interval.

  • a

    Classification was based on the International Classification of Disease code, 9th revision; Stratified Cox models on deciles of use of statins were used in all the analyses.

  • b

    Site-specific cancers were defined as the first event of the respective site-specific cancer during follow-up period, and the follow-up time was calculated as the period from enrolment for the first site-specific cancers, death or censoring date whichever came first. The sum exceeds 243 due to multiple sites of cancers in some patients.

Lip, oral cavity, and pharynx9Not estimated due to small sample 
Digestive organs and peritoneum1192.38 (1.49-3.82).0003
Respiratory and intrathoracic organs311.30 (0.58-2.91).6211
Bone, connective tissue, skin, and breast362.58 (1.20-5.53).0152
Genitourinary organs381.92 (0.89-4.14).0956
Lymphatic and hematopoietic tissue9Not estimated due to small sample 
Other and unspecified sites352.57 (1.09-6.08).0317
Cancers other than digestive organs and peritoneum cancer1412.10 (1.41-3.10).0003

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

In this analysis, we have made novel discoveries regarding the risk association of cancer with low triglyceride, defined as <1.70 mmol/L (but not high triglyceride), and that this risk association was attenuated in the presence of statin therapy.

The association between triglyceride and cancer remains inconsistent in general populations and largely unexplored in patients with T2DM despite their increased cancer risks.2 Some studies reported positive associations between triglyceride and certain cancers such as lung, rectal, thyroid,20 breast,21 and endometrial cancer.22 Conversely, other workers have reported negative associations of triglyceride with non-Hodgkin lymphoma and prostate cancer in men.20 Recent studies also failed to detect any significant relationships between triglyceride and breast cancer risk23, 24 or cancer mortality.25

By using a prospective diabetes registry with detailed documentation of risk factors, complications, clinical outcomes, and drug use, our group has systematically examined the risk relationships between T2DM and cancer in Chinese populations. We first reported an A-shaped association between triglyceride levels and cancer, with a zenith at 1.17 mmol/L and a long right tail covering values >2.5 mmol/L.4 In this analysis, we further confirmed that copresence of low triglyceride, defined as <1.70 mmol/L, and nonuse of statins was associated with increased risk of cancer in T2DM. Although we detected increased risks of all-site cancer in patients with low triglyceride plus nonuse of statins, the small sample size prevented us from testing associations for site-specific cancers. Notwithstanding the consistent overall trend for site-specific cancers, we acknowledged that other studies have reported a negative association between diabetes and other site-specific cancers, such as lung cancer.26

Although the nature of these findings is not immediately obvious, they were in agreement with our previous findings. In a series of investigations, we reported a V-shaped cancer risk association of LDL cholesterol, with a nadir at 3.28 mmol/L. LDL cholesterol both ≥3.80 mmol/L and <2.8 mmol/L was associated with elevated risks of cancers at any sites in T2DM patients naive to statins.3 Copresence of LDL cholesterol <2.80 mmol/L and albuminuria dramatically increased the risk, which was more than their summation. Similar to findings in the present analysis, the increased cancer risk associated with copresence of low LDL cholesterol and albuminuria was attenuated in the presence of statin therapy.5 In a uninephrectomized rat model that developed glucose intolerance, renal impairment, and renal cancer 9 months after the procedure, we observed up-regulation of hydroxymethylglutaryl-coenzyme A reductase, and IGF-1 and Akt pathways. In animals treated with a renin-angiotensin system inhibitor, none developed renal cancer with normalization of the above pathways.6 In this context, during cholesterol biosynthesis, intermediate nonsterol products such as farnesyl diphosphate, geranylgeranyl diphosphate, and dolichyl phosphate are produced. These molecules have been implicated in the post-translational modification and activity of many small proteins essential for cellular growth.27 On the basis of these prior knowledge and our findings, we hypothesized that copresence of LDL cholesterol and albuminuria might be associated with up-regulation of cholesterol biosynthesis and intracellular accumulation of nonsterol by-products to increase risk of cancer in T2DM.5

Furthermore, lipid biosynthesis is tightly regulated by 3 sterol regulatory element-binding proteins (SREBPs) with multiple feedback mechanisms.28 Among the 3 SREBPs, SREBP-1c regulates biosynthesis of free fatty acid or triglyceride, whereas SREBP-1a and SREBP-2 regulate cholesterol and free fatty acid biosynthesis.28 Insulin selectively increases expression of SREBP-1c,29 whereas IGF-1 induces SREBP-1a and SREBP-1c expression via activation of phosphoinositide 3-kinase/Akt pathway.30

Thus, using similar arguments and given the importance of insulin action in lipid storage, we postulate that insufficient insulin secretion or its action may up-regulate the IGF-1 and cholesterol biosynthesis pathway, which can lead to intracellular accumulation of nonsterol products, which nonetheless can be inhibited by use of statins. In support of this hypothesis, we recently reported a linear association between HbA1c and cancer risk. Importantly, use of insulin was independently associated with 80% risk reduction in cancer risks in these patients with T2DM.31 Put another way, insulin treatment in insulin-insufficient patients may restore synthesis of free fatty acid and prevent up-regulation of the IGF-1 and cholesterol synthesis pathways, with attenuated cancer risk (See Fig. 3). Of note, this hypothesis only explains in part the link between T2DM and cancer in patients with insufficient insulin action; that between cancer and hyperinsulinemia requires further investigations.

thumbnail image

Figure 3. Hypothesized pathways for increased risk of cancer in Type 2 diabetes mellitus (T2DM) are shown. Insufficient insulin secretion or action may cause down-regulation of SREBP-1c and its downstream signals, leading to a low triglyceride level, which may up-regulate the IGF-1, SREBP-1a and cholesterol synthesis pathways. The up-regulation of these pathways will then lead to increased production of intermediate molecules, which may lead to carcinogenesis. HMG-CoA indicates hydroxymethylglutaryl-coenzyme A; FFA, free fatty acid; [UPWARDS ARROW], activation/up-regulation; [DOWNWARDS ARROW], inhibition/down-regulation.

Download figure to PowerPoint

Our study has several limitations. First, triglyceride levels during follow-up were not systematically collected and therefore not available for this analysis. The triglyceride status in some patients might also have changed during follow-up period. Second, the small number of fibrate users does not allow us to perform a meaningful analysis to examine interaction between low triglyceride and fibrate use. Third, hospital admission and death registry data were used to retrieve cancer events during the follow-up period, and a small number of cancer cases might have been missed. Fourth, this was a mainly clinic-based cohort, although major referral bias was unlikely, as evidenced by the low rates of complications at enrollment and an annual mortality rate of 1% to 2%, comparable to most Caucasian community-based cohorts or clinical trials.32 Fifth, this study was largely exploratory, and requires independent replication.

In conclusion, in a prospective cohort of Chinese patients with T2DM with detailed documentation of drug use during the follow-up period, low triglyceride level was associated with increased cancer risk, which was attenuated in the presence of statin therapy. Our findings emphasize the critical importance of maintaining energy balance and internal milieu as reflected by the narrow range of biomarkers including lipid values, such as LDL cholesterol and HDL cholesterol.3, 4 Taking all data into consideration, we hypothesize that in T2DM, insufficient intracellular lipid synthesis because of lack of insulin action may cause up-regulation of other pathways, resulting in production of intermediate carcinogenic molecules, which nonetheless can be inhibited by statin therapy. Given the high risk of T2DM patients for cancer, our findings suggest that statins may have beneficial effects beyond cardiovascular protection, and that lipid values may help identify high-risk subjects for further intervention.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

We thank all medical and nursing staff of the Diabetes and Endocrine Center of the Prince of Wales Hospital for recruiting and managing these patients.

CONFLICT OF INTEREST DISCLOSURES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

This study was supported by a Research Grant Council Direct Allocation (Reference No. 2008.1.043). Other support included the Hong Kong Foundation for Research and Development in Diabetes and Lioa Wun Yuk Diabetes Memorial Fund, established under the auspices of the Chinese University of Hong Kong. J.C.N.C. has received research grant and/or honoraria for consultancy or giving lectures from Merck, Pfizer, Astra Zeneca, and/or BMS, all of which have been donated to the Chinese University of Hong Kong to support ongoing research and development. P.C.Y.T. has been a member of the Asian Advisory Board of Eli-Lilly and has received honoraria from Sanofi-Aventis (Symposia).

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES