Association between plasma lipid levels during acute coronary syndrome and long-term malignancy risk. The ABC-4* study on heart disease

Background Emerging evidence suggests that patients with coronary artery disease carry an increased risk of developing malignancy, with deleterious effects on long-term prognosis. Our aim was to ascertain whether baseline plasma lipid levels during acute coronary syndrome (ACS) are associated with malignancy in long-term. Methods This study included 589 patients admitted with ACS to three centers and discharged alive. Plasma lipid levels were assessed on the first morning after admission. Patients were followed for 17 years or until death. Results Five hundred seventy-one patients were free from malignancy at enrollment, of them 99 (17.3%) developed the disease during follow-up and 75 (13.1%) died due to it. Compared to patients without malignancy, those with malignancy showed lower plasma levels of total cholesterol (TC), low-density lipoprotein (LDL), and triglycerides (TG). The groups showed similar statin use rates at any time in follow-up. The incidence rate of neoplasia and neoplastic mortality was higher in patients with baseline TC or LDL values ≤ median; they showed 85 and 72% increased incidence rate of developing malignancy and 133 and 122% increased incidence rate of neoplastic death respectively. No differences were observed relative to HDL and TG levels. In survival analysis using Cox regression with parsimonious models, patients with baseline TC or LDL values > median, respectively, showed risks of 0.6(95% CI 0.4–0.9; p = 0.01) and 0.6(95%CI 0.4–0.9; p = 0.02) for malignancy onset, and 0.5(95% CI 0.3–0.8; p = 0.005) and 0.5(95% CI 0.3–0.8; p = 0.004) for neoplastic death. Similar results were obtained using competitive risk analysis with parsimonious models. Conclusions This long-term prospective study of an unselected real-world patient sample showed that neoplasia onset and mortality are independently associated with low plasma TC and LDL levels at admission for ACS.

For the present study, the following data were re- residuals were used to test the proportionality assumption 154 with 95% confidence intervals (CI). All hazard ratios (HR) 155 estimated in survival analysis were based on analysis of 156 dichotomous variables, using the 50th percentile for 157 continuous variables, and absence/presence of a feature 158 for categorical variables. The same models were also 159 assessed using the continuous baseline variables, and the 160 strength of association expressed as Z values (the ratio of 161 the HR and SE). The International System of Units is 162 used throughout the text. Unless otherwise indicated, 163 two-tailed P values of < 0.05 were considered signifi-164 cant. Statistical analyses were performed using STATA 165 14 (College Station, Texas, USA). 166 Results 167 All enrolled patients completed the follow-up unless 168 pre-empted by death-except three patients for whom 169 survival time was censored before 17 years (two withdrew 170 consent and one moved overseas). Among the589 patients 171 who were discharged alive, 18patients had previously 172 diagnosed malignancy at the time of enrollment and were 173 excluded from the present analysis. Ninety-nine patients 174 developed the disease during the follow-up (Fig. 1). Table   T1 1 175 presents the patients' baseline clinical characteristics ac-176 cording to the development of neoplasia during follow-up.

177
The two groups did not differ in age at enrollment, history 178 of hypertension or alcohol use. The prevalence of neopla-  Plasma HDL levels did not differ between groups.

189
Comparing patients who developed neoplasia to those 190 who did not, there were no differences in the rate of  The incidence rate of new malignancy throughout 195 follow-up after ACS was approximately18 cases/1000 196 person-years. Unexpectedly, this incidence rate was 197 markedly higher (23 cases/1000 person-years)among 198 patients with baseline TC ≤ median value of 208 mg/dL, 199 and the estimated rate ratio was significantly below 1 200 (Table   T2 2). A similar rate ratio was observed for LDL. In 201 contrast, the rate ratio was closer to 1and non-significant 202 for HDL and TG.

203
At the end of follow-up, 75 Figure   F2 2 presents the cumulative incidence of malignancy 224 onset and neoplastic death throughout the follow-up in 225 patients with plasma TC and LDL values of > or ≤ median 226 values, revealing significant differences between these 227 groups (Table 2). There were no significant differences 228 relative to HDL and TG (Fig.   F3 3).

229
Univariable Cox survival analysis demonstrated that the 230 hazard of malignancy onset and neoplastic mortality 231 throughout follow-up after ACS were higher among 232 patients with baseline TC or LDL values ≤ median values 233 (Table   T3 3). The proportional hazards assumption was 234 verified for all variables concerning plasma lipid levels 235 (p ≥ 0.10).

236
The higher hazard remained significant even after 237 accounting for clinical confounders in the fully adjusted 238 models and the parsimonious models (Table 3). Fully      The relationship between plasma cholesterol con-313 centration and mortality is complex. Although plasma 314 concentration is positively correlated with CAD-related 315 mortality, it shows a negative relationship with death 316 from cancer. These two relationships could reflect causal 317 mechanisms that are reversible by changes in plasma TC 318 concentration. In this scenario, the benefits of lipid 319 reduction for heart disease might be partly offset by 320 increased cancer-related mortality [31].

321
In concordance with the medical knowledge, we found 322 association between malignancy risk and other impor-     (1995)(1996)(1997)(1998), 364 and steadily increased from the 1st to the 17th year of 365 follow-up, in accordance with guideline revisions over the 366 time period. However, our statistical analysis results 367 suggested that lipid-lowering treatment did not influence 368 the association of plasma lipid levels with cancer onset 369 and mortality. Yet is to be considered that risk factors of 370 occurrence of cancer vary by type of cancer, and it is of 371 clinical relevance. However, this issue is beyond the scope 372 of the present study, which aimed to assess the relation-373 ship between lipid and cancer incidence and death after 374 ACS. One more limitation is that only baseline plasma 375 lipid measurements were considered in the present study, 376 while changes in lipid profile are to be expected through 377 such a long time of follow up, mainly due to lifestyle and 378 treatment changes. Nevertheless, the associations we 379 observed seem to be clinically consistent, and the assess-380 ment of lipid profile at admission for ACS can be a sort 381 key point in the patient's life. Finally, since the patients in 382 this study were all Caucasians, we cannot generalize the 383 present findings to other populations and ethnic groups.