Plasma triglycerides predict ten-years all-cause mortality in outpatients with type 2 diabetes mellitus: a longitudinal observational study

Statistical analysis

Continuous variables were expressed as mean value ± standard deviation (SD) or as median value (interquartile range), when appropriate. Categorical variables were expressed as percentages. Differences between groups were assessed by the Student’s t test for continuous variables, the Kruskal-Wallis test for continuous variables without normal distribution, the x ² test for the categorical data. Multivariate Cox proportional hazard analysis was used to examine the impact of different metabolic parameters and treatments on the Hazard Ratio (HR) for total mortality (method: stepwise backward). The assumption of proportionality of all variables introduced in the models was assessed through the analysis of Schoenfeld residuals. The Cox models were adjusted for other factors associated with mortality. Two different models were tested:

Model 1 included: age and gender; tertiles of BMI; mean values of fasting blood glucose, HbA_1c, systolic blood pressure, serum LDL-C, HDL-C, and triglycerides; antihypertensive drugs (yes/no), lipid-lowering drugs (yes/no), and antidiabetic treatment. For the latter, five categories were compared with diet (reference group), including 1. biguanides, 2. sulfonylureas, 3. biguanides + sulfonylureas, 4. insulin, and 5. insulin + oral hypoglycemic agents.

Analyses were performed by SPSS for Windows statistical package, version 13.0.

Our research is in compliance with the Helsinki Declaration. No informed consent was required because this study is observational and did not affect patient care in any way.

Population characteristics

Cox analysis

Serum Triglycerides and 10 years mortality

The contribution of triglycerides to CVD risk has been much debated in the past, with many important prospective studies observing an association between elevated triglycerides levels and CVD risk (in particular coronary risk) [18]-[20], but also risk for microangiopathy [21]; nevertheless, more recent studies gave different results, suggesting that this association might be weakened when adjustment for other risk factors was made [22]. The pathogenetic mechanism undergoing these findings remains uncertain, and could be mainly related to triglycerides induced endothelial dysfunction through oxidative stress. By Cox analysis (Model 1) we demonstrated a direct association between long-term mortality risk and triglycerides levels; the association was strong and significant even after multivariate adjustment for traditional CVD risk factors including BMI, HbA_1c, LDL-C, and medication use. This independent association with long term all-cause mortality support the idea that serum triglycerides could play a role in type 2 diabetic patients mortality risk [23].

Treating residual risk by lowering triglycerides is still debated at present time [24]; infact, it is not clear whether or not a pharmacological intervention targeted to reducing triglycerides levels would contribute to a cardiovascular risk reduction [25]. Dedicated trials failed to find a significant reduction in CVD outcomes in a diabetic population treated with fenofibrates, [26] while meta-analyses conducted upon pre-specified subgroups have confirmed the clinical benefits of fibrates on major CVD events [27],[28]. A randomized clinical trail with statin therapy plus extended release niacin including over 3,000 patients (one-third with diabetes) with established cardiovascular disease, low levels of HDL cholesterol, and triglycerides levels of 150-400 mg/dL, was halted early due to lack of efficacy on the primary cardiovascular disease outcome, and a possible increase in ischemic stroke in those on combination therapy [29]. Hence, combination lipid-lowering therapy cannot be broadly recommended. However, ADA guidelines state that triglycerides concentrations below 150 mg/dL are desirable. According to this, in our population, the satisfaction of this  desirable  triglycerides level was associated with a lower 10 years total mortality. Thereby, we suggest that more attention should be given to cardiovascular risk management in type 2 diabetic patients with high triglycerides levels; specifically, a more strict management of the other modifiable risk factors (e.g. diet, physical exercise, blood pressure, LDL-C goal) could be indicated.

As a matter of fact, the second Cox analysis was modelled upon accomplishment of ADA criteria (Model 2), and confirmed the previous observation; among the criteria established for CVD prevention in patients with type 2 diabetes, only triglycerides desirable level criteria (<150 mg/dl) was significantly associated with a lower mortality risk in our population.

More interesting, diabetic patients belonging to the “partially satisfying criteria” group (reached triglycerides target for mean values, but failure in one or more follow-up visit) had a higher HR for mortality (HR: 2.02; 95% CI: 1.11-3.48) compared to the “satisfying criteria" reference group (always reached target); besides, their HR for total mortality was very similar to the “not satisfying criteria” group (HR: 1.83; 95%IC: 1.08-2.97). We are not sure about the nature of this finding in the “partially satisfying criteria” group, and there are at least two possible explanations: 1. the atherogenic process might be activated by hypertriglyceridemia, but might successively continue independently from the persistence of high fasting triglycerides; 2. in this group, unmeasured fluctuations of triglycerides levels might be present, and might be as atherogenic and dangerous as high fasting triglycerides levels, thus predisposing them to a similar increase in the risk [30].

Lipid-lowering drugs and 10 years mortality

CHD is the most important cause of morbidity and mortality among subjects with diabetes mellitus. Statins mega-trial have demonstrated that LDL-C reduction is effective in reducing cardiovascular mortality in diabetic subjects; [31] therefore, LDL-C reduction is one of the primary therapeutic goals in reducing cardiovascular risk in diabetes [32]. According to these findings, our data show that the lipid lowering treatment is significantly associated with a reduced mortality risk, and this phenomenon is independent from other variables included into the model (HbA_1c, blood pressure, LDL-C, HDL-C, and triglycerides values). The vast majority of subjects taking lipid lowering drugs were assuming statins (90%); thus, this class of drugs is most likely to be responsible of the reduced mortality, independent from the known effect on serum cholesterol levels. Once more, there might be evidence of the “very discussed” pleiotropic effects of statins [33]; these effects may include reduced endothelial inflammation, nitric oxide production enhancement, and improved insulin sensitivity [34]. Important clinical trials such as WOSCOPS [35] and CARE [36] indicated that, despite comparable serum cholesterol levels, patients treated with statin had lower risk of CVD events compared with the placebo groups. Most of these effects might be mediated by the inhibition of isoprenoid synthesis, which affects multiple signalling pathways.

Biguanides and 10 years mortality

Metformin is considered the first line pharmacological therapy in type 2 diabetes [37],[38]. Its main effect is enhancement of cell insulin sensitivity [39]. Important studies, such as the UK Prospective Diabetes Study (UKPDS), have demonstrated a significant effect of metformin therapy on overall mortality risk reduction, and particularly on CVD and myocardial infarction outcomes [40]. Other clinical and experimental studies have shown that metformin treatment might be associated with improved outcomes, supporting the conclusions from UKPDS. In addition, a well-designed retrospective analysis has shown significantly lower mortality rates in patients receiving metformin compared with patients treated with sulphonylurea monotherapy [41]. The ADA and EASD consensus guidelines on the management of hyperglycemia in type 2 diabetes explicitly state that metformin should be used as first-line therapy, in addition to lifestyle interventions. Our data are consistent with this indication, showing that metformin treatment is associated with a reduced mortality risk compared to diet and other antidiabetic agents. Nevertheless, we have to make some considerations upon this result. In particular, patients treated with metformin had, on average, lower creatinine levels compared with controls (0.96 ± 0.31 vs 1.12 ± 0.63). Thus, although this finding is not surprising (metformin cannot be prescribed in subjects with chronic kidney disease), the difference in renal function might contribute to the results.

Finally, we have to acknowledge some important limitations of the study. First, our laboratory data consisted of only minimum, maximum, and mean values, while the single laboratory determinations were unavailable. Second, we disposed of data concerning presence/absence of micro/macro vascular complications, but we didn’t have precise additional information upon the onset of cardiovascular events. Third, we only disposed of all-cause mortality without any information concerning the possible cause of death. We would also underline the strength of our clinical study; indeed, our results are based on a large sample of older adults type 2 diabetic outpatients (1917 individuals, 40% females) with a long-term follow-up (on average ten years).