Table 3 Summary of primary prevention studies in high risk populations (7 studies)

Agewall 2002 [52] (n = 118)

Study design: Prospective cohort study

Follow-up: 3.0 ± 0.6 yrs

Population description: Males 56 to 77 yrs with treated hypertension

Overall risk of bias ^a : High risk

Funding: NR

Model: Cox proportional hazards

Variables: Age and other variables; LDL-C not included

CVD Outcomes: Non-fatal MI or CD

Lp(a) assay: Isoform independence - NR; NR if fresh or frozen samples

Lp(a) comparison type: Continuous

Two effects sizes reported, each using a different model for the relationship between log per Lp(a) increase and non-fatal MI or CD:

HR 2.84, 95 % CI: 1.06 to 7.63 (adjusted for age, BP, smoking, cholesterol, diabetes)

HR 2.97, 95 % CI: 1.03 to 8.37 (adjusted for CD at entry)

Both were statistically significant showing that Lp(a) is a significant and independent predictor for major coronary events

CHOICE [53] (n = 833)

Study design: Prospective cohort study

Follow-up: Median 27.4 mths

Population description: Mixed gender adults 17 yrs + on dialysis

Overall risk of bias ^a : Moderate risk

Funding: Public/government

Model: Cox proportional hazards

Variables: Age, gender and other variables; LDL-C not included

CVD Outcomes: ASCVD

Lp(a) assay: Isoform independent;

frozen samples

Lp(a) comparison type: Categorical

Ten effect sizes reported from five models of two categorical comparisons. Nine out of ten showed a statistically significant, positive association (same direction) for Lp(a) with respect to ASCVD.

Maximum effect size reported was for Lp(a) ≥ 206 nmol/L (ref) vs. Lp(a) < 206 nmol/L (HR 1.89, 95 % CI: 1.3 to 2.75).

One effect size was NS: Lp(a) ≥ 52.5 nmol/L (ref) vs. Lp(a) <52.5 nmol/L (HR 1.25, 95 % CI: 0.99 to 1.58)

The authors concluded that ASCVD was significantly and independently associated with high Lp(a) (>123 nmol/L) and low molecular weight (LMW) apo(a) isoforms, though a stronger relationship was found for ASCVD and low molecular weight isoform size. This is in ESRD patients and there was a high transplantation rate (17.3 %) which may have biased the Lp(a) results

Cleveland Clinic Hemodialysis Cohort [54] (n = 129)

Study design: Prospective cohort study

Follow-up: 4 yrs

Population description: Mixed gender adults ≥ 18 yrs on haemodialysis

Risk of bias assessment overall ^a : Moderate risk

Funding: Public/government

Model: Multiple regression (OR) and Cox proportional hazards (HR)

Variables: Gender and other variables; includes LDL-C

CVD Outcomes: Atherosclerotic events including stroke and MI

Lp(a) assay: Isoform independence -NR/unclear; frozen samples

Lp(a) comparison type: Continuous

Two effect sizes reported for Lp(a) with respect to atherosclerotic events:

OR 1.02, 95 % CI: 1.01 to 1.04 (multiple regression)

HR 1.603, 95 % CI: 1.08 to 2.38 (Cox proportional)

A 1-mg/dL or 10-mg/dlL increment in baseline Lp(a) concentration was associated with a 1.02 or 1.26 increase, respectively, in the relative risk of sustaining an event (p = 0.001)

Both results suggested that baseline Lp(a) is a significant and independent risk factor for clinical events

Diamant Alpin Collaborative Dialysis Cohort [55] (n = 279)

Study design: Prospective cohort study

Follow-up:2 yrs

Population description:

Mixed gender adults 22 to 92 yrs with and without type 2 diabetes

Risk of bias assessment overall ^a : High risk

Funding: Pharma

Model: Cox proportional hazards regression

Variables: Age, gender and other variables; LDL-C not included

CVD Outcomes: CVD events (MI, de novo angina pectoris or coronary revascularization, ischemic stroke, or PAD) and CV death (due to cardiac arrhythmia, MI, or HF)

Lp(a) assay: NR/unclear; Immunoturbidimetric assay; Isoform dependence - NR/unclear; NR if fresh or frozen samples

Lp(a) comparison type: Categorical

One effect size reported which showed a statistically significant positive association (same direction) for Lp(a) > 300 mg/L vs ≤ 300 with CVD events and CV deaths: HR 1.67, 95 % CI: 1.04 to 2.63

This result suggested that Lp(a) is an independent and significant predictor of CV events.

JDCS [56] (n = 1304)

Study design: RCT

Follow-up: Median 7.8 yrs

Population description: Mixed gender Japanese adults 40 to 70 yrs with Type 2 diabetes

Risk of bias assessment overall ^a : High risk

Funding: Public/government

Model: Cox proportional hazards

Variables: Age, gender and other variables; LDL-C not included

CVD Outcomes: Stroke (ischemic, hemorrhagic or TIA), CHD

Lp(a) assay: Isoform independence -NR/unclear; frozen samples

Lp(a) comparison type: Continuous

One effect size reported which showed a statistically significant positive association (same direction) of Lp(a) (per 1 μmol/l increase) with an increased risk of stroke (ischemic, hemorrhagic or TIA): HR 1.16, 95 % CI: 1.03 to 1.31

Suggests that increasing Lp(a) is and independent and significant risk factor for stroke

Koda 1999 [57] (n = 390)

Study design: Prospective cohort study

Follow-up:2.3 yrs

Population description:

Mixed gender Japanese adults ≥ 18 yrs with or without type 2 diabetes receiving haemodialysis

Risk of bias assessment overall ^a : High risk

Funding: NR/unclear

Model: Multiple logistic regression model

Variables: Age, gender, albumin, Lp(a), diabetic state; LDL-C not included

CVD Outcomes: Death and CV death

Lp(a) assay: NR/unclear; Immunoturbidimetric assay; Isoform dependence - NR/unclear; NR if fresh or frozen samples

Lp(a) comparison type: Categorical

One effect size reported for Lp(a) showed a statistically significant positive association (same direction) with respect to CV death, comparing High Lp(a) [≥ 30 mg/dL] vs Low Lp(a) [< 30 mg/dL]: OR 3.93, 95 % CI: NR. This association was statistically significant.

One effect size reported for Lp(a) with respect to overall death, comparing High Lp(a) [≥ 30 mg/dL] vs Low Lp(a) [< 30 mg/dL]: OR 1.97 95 % CI: NR. This association was not statistically significant.

Suggests that high Lp(a) [≥ 30 mg/dL] is an independent and significant risk factor for atherosclerotic CV death, but not overall death.

Zimmermann 1999 [58] (n = 440)

Study design: Prospective cohort study

Follow-up: 12 and 24mths

Population description: White mixed gender adults 20 to 88 yrs on chronic haemodialysis

Risk of bias assessment overall ^a : Low risk

Funding: Public/government

Model: Cox proportional hazards

Variables: Age, gender and other variables; LDL-C not included

CVD Outcomes: All deaths, stroke, HF, MI

Lp(a) assay: Isoform independence - NR; fresh samples

Lp(a) comparison type: Categorical

Lp(a) was significantly associated with risk of all-cause and cardiovascular mortality in univariate Cox regression analysis, but was not significant in the multivariable Cox regression analysis.

This study is in haemodialysis patients, in such patients Lp(a) reacts as an acute phase protein in combination with other factors such as fibrinogen, HDL-C and Apo A-I, changing the atherogenic risk profile. When Lp(a) is added to the multivariable model with these other variables Lp(a) no longer remains significant as an independent factor.