Lp-PLA2 activity and mass and CRP are associated with incident symptomatic peripheral arterial disease

Long follow up is needed in prospective cohort study evaluation of plasma biomarkers for incident peripheral arterial disease (PAD) Middle-aged PAD-free individuals from the cardiovascular cohort of the Malmö Diet and Cancer study (n = 5550; 1991–94) were followed prospectively for a median time of 23.4 years. The plasma biomarkers lipoprotein-associated phospholipase A2 (Lp-PLA2) activity and mass, proneurotensin, and CRP, were studied in relation to incidence of PAD until December 31st, 2016. The diagnosis of PAD could be validated and confirmed in 98%. Cox regression was used to calculate hazard ratios (HR) per 1 standard deviation increment of each respective log transformed plasma biomarker. Cumulative incidence of PAD was 4.4% (men 5.9%, women 3.3%). Adjusting for age, gender, smoking, body mass index, hypertension, diabetes mellitus, Lp-PLA2 activity (HR 1.33; 95% CI 1.17–1.52), Lp-PLA2 mass (HR 1.20; 95% CI 1.05–1.37) and CRP (HR 1.55; 95% CI 1.36–1.76) remained independently associated with incident PAD. The plasma biomarkers Lp-PLA2 activity and mass, and CRP were markers of PAD risk, implying that they might be useful biomarkers for subclinical atherosclerosis and atherosclerotic disease.

Endpoint ascertainment. Individuals from the MDCS with a first registered diagnosis of PAD were identified from Swedish national registers 16 (the Inpatient 17 and Outpatient Register and the Cause of Death Register) by linkage of the ten-digit personal identification number unique to each Swedish resident. Follow-up extended until December 31 st , 2016. In both the Inpatient and Cause of Death registers, diagnoses are coded using a Swedish revision of the International Classification of Disease (ICD), version 8  Validation of PAD diagnosis. One hundred patients with diagnosis of PAD were randomly selected for the validation procedure using patient record data. Among 100 patients, 69 had chronic limb-threatening limb ischemia, 13 had acute limb ischemia, 15 had claudication, and one had asymptomatic PAD. Among the 13 patients with acute limb ischemia, twelve had acute thrombotic occlusion and one had an embolic occlusion. Two patients had venous insufficiency and were misdiagnosed. PAD could therefore be confirmed in 98% of cases and symptomatic PAD in 97% of cases.
Laboratory measurements. Plasma biomarkers were measured from fasting plasma samples that had been frozen at −80 °C immediately after collection 15 . Proneurotensin was measured using a chemiluminometric sandwich immunoassay to detect a proneurotensin fragment 10 . CRP was measured by a high-sensitivity Tina-quant ® latex assay (Roche Diagnostics, Basel, Switzerland). The average coefficient of variation (CV) was 4.59% 18 . Lp-PLA 2 was expressed as enzymatic activity and mass (quantity) 19 . Lp-PLA 2 activity was measured in www.nature.com/scientificreports www.nature.com/scientificreports/ duplicate using [3 H]-platelet activating factor as a substrate 19 . The range of detection was 8-150 nmol/min/ml. Samples were retested if the replicate CV was >20%. The average CV was 5.78% 18 . Lp-PLA 2 mass measurements were performed using the second generation PLAQ TM test (diaDexus Inc., South San Francisco, CA, USA) commercially available enzyme-linked immunosorbent assay (ELISA) kit 19 . All samples were analyzed in duplicate, and if a duplicate showed a CV of more than 20%, the sample was reanalyzed. The average CV was 4.62% on random of 50 first participants in the MDCS 18 . Plasma-EDTA samples are stable for Lp-PLA 2 activity and mass measurements within 7 days of collection for refrigerated samples and for more than 10 years from collection when stored at −70 °C 19 .
Statistical analysis. Quantitative normal and skewed distributed variables are presented as mean with standard deviation and median with interquartile range (IQR), respectively. Dichotomous variables are presented as count and proportion. Individuals with a diagnosis of PAD at baseline were excluded from the current study and prospective analyses included only incident PAD. Plasma biomarkers and confounders for incident PAD were assessed using Cox regression models, and hazard ratios (HRs) were expressed per one standard deviation (SD) increment of each respective log transformed plasma biomarker (skewed distributed) in the Cox regression models. Cumulative incidence of PAD was analyzed using Kaplan-Meier method. Log-rank test was used in the comparison of quartiles for Lp-PLA 2 activity. Analyses were performed using SPSS for Windows, version 23.0 (SPSS Inc, Chicago, IL). A p-value less than 0.05 was considered significant.

Results
Baseline conventional risk factor assessment. Fifteen individuals with known PAD at baseline were excluded. The cumulative incidence of PAD was 4.4% (244/5550), 5.9% (137/2307) for men versus 3.3% (107/3243) for women (p < 0.001), during a median follow up period of 23.4 years (IQR 19.4-24.3). Baseline risk factor characteristics for individuals with or without PAD in the cohort are shown in Table 1. When including the conventional risk markers listed in Table 1 into a Cox regression analysis, age at baseline (p < 0.001), male gender (p < 0.001), current smoking (p < 0.001), diabetes mellitus (p < 0.001) and hypertension (p = 0.001) were independently associated with incident PAD, whereas BMI (p = 0.74) were not.

Discussion
The present longitudinal prospective study suggests that elevation of plasma Lp-PLA 2 (activity), Lp-PLA 2 (mass), and CRP, are all markers of subclinical disease susceptibility long time before diagnosis of PAD. This finding suggests that these biomarkers indicate subclinical PAD prior to onset of symptoms due to PAD. The association between Lp-PLA 2 and incident PAD has previously been evaluated in three different cohorts, the Cardiovascular   21 and Atherosclerosis Risk in Communities (ARIC) study 22 . The Cardiovascular Health and ARIC study showed an association between Lp-PLA 2 and incident PAD, whereas the MESA study showed no association. The negative findings in the MESA study 21 were probably related to the fact that PAD was exclusively defined with repeated measurements of ankle-brachial index (ABI), resulting in inclusion of mainly asymptomatic PAD patients, whereas the other cohort studies used clinical endpoint data retrieved from hospital registries. Hence, it is likely that individuals that later develop symptomatic PAD as opposed to asymptomatic PAD may have been exposed to larger amounts of pathophysiological stimulators such as Lp-PLA 2 , promoting development of PAD. Plasma Lp-PLA 2 (activity) and Lp-PLA 2 (mass) were both strong predictors for incident PAD in the present study, since these biomarkers remained independently associated with incident PAD after adjusting for confounders in the extended multivariable analysis including the well established risk marker CRP 6,7 . The elevated Lp-PLA 2 levels themselves, however, seem unlikely to be a primary causal factor for atherosclerotic disease according to phase III randomized controlled trials of inhibitors of Lp-PLA 2 , varespladib and darapladib 23 , and also to genetic studies using Mendelian randomization 24 . Instead, Lp-PLA 2 levels appears to reflect an ongoing inflammatory process.
CRP is an established marker of systemic inflammation, vascular disease, and increased risk in PAD patients 6,7 and the present study data showed that CRP, after adjusting for relevant confounders, was elevated already at baseline in patients developing symptomatic PAD, probably reflecting presence of subclinical atherosclerotic disease. Results from an interventional trial evaluating rosuvastatin (JUPITER) 25 suggest that CRP may assist in risk stratifying healthy participants without hyperlipidemia for rosuvastatin treatment in primary prevention of atherosclerosis.
Increased plasma proneurotensin levels has been shown to be strongly associated with obesity 26 and a predictor of cardiovascular disease in both genders and diabetes mellitus in women in two separate prospective cohorts 10,27 . The absence of association between proneurotensin and incident PAD, may be attributed to the finding that body mass index not was associated with incident PAD in the present study.  Table 2. Multi-variable adjusted hazards ratios for incident PAD in relation to plasma biomarkers. The following variables were entered in the multivariable analysis besides each respective plasma biomarker: Age at study entry, sex, body mass index, current smoking, diabetes mellitus, hypertension. *Hazard ratios (HR) were expressed per one SD increment of each respective log transformed plasma biomarker. PAD; peripheral artery disease, CRP; C-reactive protein, Lp-PLA 2 (activity and mass); lipoprotein-associated phospholipase A2, HR; Hazard Ratio, CI; Confidence interval.
www.nature.com/scientificreports www.nature.com/scientificreports/ The MDCS cohort was originally designed to investigate the effects of diet on cancer risk 28 . A limitation of the study was that ABI not was measured at baseline to classify and exclude asymptomatic PAD individuals. Re-invitation of study participant survivors for detection of asymptomatic PAD by measurement of ABI would have been interesting to increase the total number of incident PAD individuals, perform subgroup analysis of factors associated with symptomatic and asymptomatic incident PAD, and increase power in the overall statistical analysis. A major strength of the present longitudinal study design, on the other hand, is the inclusion of healthy middle-aged individuals followed up for a median time of 23.4 years. Re-invitation would also have captured changes in risk factor status among study participants during this long period. During the last decade, the declining prevalence of smoking and the improved pharmacological treatment among individuals with cardiovascular disease might well have impacted the cumulative incidence of, above all, symptomatic PAD 29 . Another strength of this study was the high validity of symptomatic PAD, we could reliably confirm the PAD diagnosis in 98% of cases.
Lp-PLA 2 mass and activity, but not CRP, have previously been found to be predictors of incident abdominal aortic aneurysm (AAA) 30 in the same cohort during shorter follow up. CRP was the strongest predictor for incident PAD in the present study, and this discrepancy compared to the findings concerning AAA prediction may be interpreted as suggesting that PAD is a more inflammation-driven disease than AAA. Whereas the two diseases shared some similar risk factors in the two cohort studies, diabetes mellitus and hypertension were associated with incident PAD, but not with incident AAA. Actually, diabetes mellitus is a protective factor and reduces the incidence of AAA 31 . These differences in premorbid plasma biomarker profile highlight that PAD and AAA are different diseases with different pathophysiologies which should be analyzed and discussed separately.
In conclusion, plasma biomarkers Lp-PLA 2 activity and mass were found to be useful markers of PAD risk during long-term follow-up together with the established biomarker CRP, implying that they might also be used as predictors for subclinical atherosclerosis and atherosclerotic disease.