Low-density lipoprotein (LDL) cholesterol and triglyceride-rich lipoproteins constitute non-high-density lipoprotein (non-HDL) cholesterol. These are atherogenic lipoproteins and non-HDL cholesterol is a secondary target of treatment beyond LDL cholesterol in patients with hypertriglyceridemia. Some investigators favor total apolipoprotein B over non-HDL cholesterol as the secondary target of treatment. This is based on publications suggesting that total apolipoprotein B is more predictive of cardiovascular events than non-HDL cholesterol. Several methods are available for estimating total apolipoprotein B. This study compared total apolipoprotein estimated by immunonephelometric assay (INA), vertical auto profile (VAP), nuclear magnetic resonance (NMR), and non-HDL cholesterol levels in patients with hypertriglyceridemia from the previously reported Simvastatin plus Fenofibrate for Combined Hyperlipidemia (SAFARI) trial. Total apolipoprotein B levels were found to be highest by INA, intermediate by NMR and non-HDL cholesterol, and lowest by VAP. Concordance for non-HDL cholesterol levels among the INA, VAP, and NMR methods was better than that for total apolipoprotein B levels; the correlation between non-HDL cholesterol and apolipoprotein B by INA was strongest (0.929). In patients with a low triglyceride/HDL cholesterol ratio (<3.5), total apolipoprotein B determined by INA was higher than that estimated from non-HDL cholesterol levels, whereas in patients with a high triglyceride/HDL C ratio (≥3.5), apolipoprotein B predicted using non-HDL cholesterol was in better agreement with INA-determined apolipoprotein B levels. Similar trends were observed with VAP using equations specific for LDL particle size. In conclusion, more work is needed to improve agreement of apolipoprotein B measurements among methods employed clinically. Non-HDL cholesterol is also useful to predict total apolipoprotein B and some improvement may be attained by taking into account the ratio of triglyceride/HDL cholesterol as a measurement of LDL particle size.

Non-high-density lipoprotein (non-HDL) cholesterol is a measurement of low-density lipoprotein (LDL) cholesterol and atherogenic triglyceride-rich lipoproteins. It is recommended as a secondary target of lipid-lowering therapy in patients with hypertriglyceridemia. Apolipoprotein B, a proposed alternative goal of treatment in these patients, represents non-HDL particle number. These are considered better predictors of cardiovascular disease events than LDL cholesterol. Some investigators have claimed that apolipoprotein B is a more robust cardiovascular disease predictor than non-HDL cholesterol, whereas others have reported that non-HDL cholesterol predicts cardiovascular disease as effectively as apolipoprotein B. Non-HDL cholesterol has the advantages of simplicity, lower cost, and wide availability. It remains to be proved whether apolipoprotein B is superior to non-HDL cholesterol as a predictor of cardiovascular disease in routine clinical practice. At present, 1 limitation of the use of apolipoprotein B is a lack of consensus on the accuracy of its measurement; nonetheless, some clinicians use apolipoprotein B as a target of therapy. There are ≥3 commercial techniques available for apolipoprotein B measurement. The most widely used method is immunoassay, which uses 3 variants: immunoturbidimetric assay (INTA), immunonephelometric assay (INA), and radial immunodiffusion (RID). Accurate apolipoprotein B measurements can be achieved using any of these 3 methods. A second measurement isolates lipoproteins by ultracentrifugation (vertical auto profile [VAP]) and provides cholesterol concentrations in subfractions of apolipoprotein B–containing lipoproteins. A third method measures the number of apolipoprotein B–containing lipoprotein particles by nuclear magnetic resonance (NMR). This number can be converted to apolipoprotein B concentrations. Because non-HDL cholesterol is highly correlated with apolipoprotein B, non-HDL cholesterol levels can be converted into apolipoprotein B from predetermined ratios of non-HDL cholesterol to apolipoprotein B. The purpose of the present study was to compare these 4 methods for estimating apolipoprotein B in hypertriglyceridemic patients participating in the previously reported Simvastatin plus Fenofibrate for Combined Hyperlipidemia (SAFARI) study.

## Methods

The previously reported SAFARI study ( NCT00092157 ) compared the effect of simvastatin (20 mg/day) to simvastatin (20 mg/day) plus fenofibrate (160 mg/day) on percent decrease of triglyceride during 12 weeks in patients with combined hyperlipidemia and hypertriglyceridemia. Relations between non-HDL cholesterol and apolipoprotein B in the SAFARI trial have also been published recently. Subsequently, apolipoprotein B was estimated in plasma samples of these patients by immunoassay, VAP, NMR, and non-HDL cholesterol at week 0 (before treatment) and week 12 (on treatment) of the study. Approval from institutional review boards was obtained for the study.

Apolipoprotein B was measured by INA in the Medical Research Laboratory International, Inc. (Highland Heights, Kentucky). Total plasma lipids and lipoprotein cholesterol were estimated by the same laboratory. Non-HDL cholesterol was calculated as the difference between total cholesterol and HDL cholesterol. The latter was measured after polyanion precipitation of apolipoprotein B–containing lipoproteins and/or β quantification when plasma triglycerides were >400 mg/dl. The immunoassay of apolipoprotein B was carried out using a primary standard for apolipoprotein B, a liquid-stabilized reference material containing LDL, named SP3-07, selected by the International Federation of Clinical Chemistry as described previously and accepted by the World Health Organization and the Centers for Disease Control and Prevention as the universal apolipoprotein B reference standard. Apolipoprotein B content of SP3-07 was determined by the Lowry method using bovine serum albumin as a standard. Commercial immunoassay kits are generally calibrated with SP3-07.

Total apolipoprotein B was estimated from non-HDL cholesterol using a method previously described in a mostly metabolic syndrome population (n = 274) from the Veterans Affairs Medical Center at Dallas, in which apolipoprotein B was measured chemically in isolated VLDL plus intermediate-density lipoprotein and LDL fractions using the Markwell modification of the Lowry procedure with bovine serum albumin from the National Bureau of Standards as a reference standard. A chromogenicity factor of 1:1 between apolipoprotein B and bovine serum albumin was assigned. Apolipoprotein B was estimated using 2 equations developed for the estimation of apolipoprotein B in this cohort. The first equation was derived by linear regression analysis with only non-HDL cholesterol and apolipoprotein B as variables: apolipoprotein B = (non-HDL cholesterol/1.62) + 15.108. The second equation, a modest refinement of the first equation that took into account the effects of triglyceride and HDL cholesterol on apolipoprotein B levels, was derived by multiple regression analysis in which triglycerides and HDL cholesterol were entered: apolipoprotein B = (non-HDL cholesterol/1.69) + 9.695 + 4.146 ln triglyceride − 0.252 HDL cholesterol, where “ln” is natural log.

In the present study, the size of the Dallas cohort was increased to 488 patients; this cohort was a mixture of patients with and without metabolic syndrome. This cohort had considerable variability in triglyceride levels and LDL composition, and as a result the correlation coefficient between apolipoprotein B and non-HDL cholesterol was 0.863. McLaughlin et al reported that a triglyceride/HDL cholesterol ratio <3.5 predicts LDL pattern A (predominantly large, buoyant LDL) with a relatively high non-HDL cholesterol/apolipoprotein B ratio, and a triglyceride/HDL cholesterol ratio ≥3.5 is associated strongly with pattern B (small, dense LDL) with a relatively low non-HDL cholesterol/apolipoprotein B ratio. In our 488 subjects, 203 subjects had a triglyceride/HDL cholesterol ratio <3.5 and had an average non-HDL cholesterol/apolipoprotein B ratio of 1.39. In 284 subjects with triglyceride/HDL cholesterol ratio ≥3.5, the average non-HDL cholesterol/apolipoprotein B ratio was 1.28. These ratios were used to estimate apolipoprotein B levels as follows: for triglyceride/HDL cholesterol ratio <3.5, apolipoprotein B = non-HDL cholesterol/1.39 (equation 1A); for triglyceride/HDL cholesterol ratio ≥3.5, apolipoprotein B = non-HDL cholesterol/1.28 (equation 2A).

An alternative approach is to determine regression equations relating apolipoprotein B to non-HDL cholesterol for subjects with triglyceride/HDL cholesterol ratios split at 3.5. The following equations were used: for triglyceride/HDL cholesterol ratio <3.5, apolipoprotein B = (non-HDL cholesterol/1.73) + 23.126 mg/dl (equation 1B); for triglyceride/HDL cholesterol ratio ≥3.5, apolipoprotein B = (non-HDL cholesterol/1.51) + 17.952 mg/dl (equation 2B).

The VAP apolipoprotein B test is commercially available through Atherotech (Birmingham, Alabama). The VAP method separates plasma lipoprotein subclasses based on buoyancy by a single ultracentrifugal spin through a salt gradient and quantifies the cholesterol in the lipoproteins by continuous-flow enzymatic analysis. Concentrations of individual lipoproteins (VLDL, intermediate-density lipoprotein, LDL, lipoprotein[a], and HDL) and their subclasses and LDL patterns (A, B, A/B) are determined by deconvolution of the main continuous cholesterol curve using in-house developed software. Recently, Atherotech has used a variation of the non-HDL cholesterol method described in equations 1 and 2 to estimate VAP apolipoprotein B. Separate equations were developed to describe relations between non-HDL cholesterol and apolipoprotein B in 1,797 patients with LDL patterns A, A/B, and B. Apolipoprotein B was measured by immunoturbidimetric assay provided by Abbott Diagnostics (Princeton, New Jersey), which had been calibrated with the World Health Organization/International Federation of Clinical Chemistry/Centers for Disease Control and Prevention reference material for apolipoprotein B. These equations are provided in a patent application. Since this application, Atherotech has slightly modified these equations; the latter were used to calculate apolipoprotein B levels in VAP LDL patterns in the present study.

NMR is used to determine lipoprotein particle concentrations in LDL and VLDL plus intermediate-density lipoprotein (LipoScience, Inc., Raleigh, North Carolina). This method uses curve fitting (spectral deconvolution) of the methyl lipid resonance envelope in plasma; amplitude and shape of the resulting curve depend on amplitudes of methyl resonances of lipoprotein components. A curve-fitting algorithm is used to determine concentrations of lipoproteins from the plasma spectrum. To estimate apolipoprotein B by NMR methods, LDL plus intermediate-density lipoprotein plus VLDL particle concentrations by NMR are converted to apolipoprotein B by multiplying observed particle concentrations by 0.055. This conversion is based on the fact that the molecular weight of apolipoprotein B is 550,000 Da, and all lipoproteins in density ranges of LDL, intermediate-density lipoprotein, and VLDL have 1 apolipoprotein B molecule per particle. Non-HDL cholesterol from NMR analysis was calculated as the sum of total cholesterol reported for VLDL, intermediate-density lipoprotein, LDL, and lipoprotein(a) subfractions.

Data are summarized as mean ± SD. Pearson correlation coefficients were calculated for linear regression analyses. Repeated-measurements analysis of variance was used for comparisons of means of total apolipoprotein B, HDL cholesterol, and non-HDL cholesterol among methods. An alpha value of 0.05 was accepted as significantly different for the comparisons.

## Results

Figure 1 shows a plot of apolipoprotein B versus non-HDL cholesterol levels for 1,204 samples (602 at baseline, 401 on simvastatin + fenofibrate therapy, and 201 on simvastatin alone) from patients participating in the SAFARI trial. The 2 measurements were performed in Medical Research Laboratory International; apolipoprotein B was measured by INA and non-HDL cholesterol by standard chemical methods. Overall, for the vast majority of subjects, non-HDL cholesterol correlated strongly and positively with total apolipoprotein B (r = 0.929, SEE 13.2). A few subjects had relatively high levels of non-HDL cholesterol with relatively low levels of total apolipoprotein B, and 4 subjects had relatively high levels of total apolipoprotein B with relatively low non-HDL cholesterol levels. Distribution of ratios of non-HDL cholesterol to apolipoprotein B was unimodal and generally symmetrical around the mean ratio of 1.27 ± 0.139 mg of non-HDL cholesterol per milligram of total apolipoprotein B ( Figure 1 ).

Apolipoprotein B measurements were made by INA and VAP in 1,095 paired samples obtained from 548 patients during baseline, 367 during combined drug therapy, and 181 during statin monotherapy who were participating in the SAFARI trial. Table 1 lists total apolipoprotein B levels and other lipid measurements for these patients including those with triglyceride/HDL cholesterol ratios <3.5 (390 samples) and ≥3.5 (705 samples). Patients with triglyceride/HDL cholesterol <3.5 had lower levels of non-HDL cholesterol and triglyceride plus higher non-HDL cholesterol/apolipoprotein B ratios and HDL cholesterol levels than those with triglyceride/HDL cholesterol ratios ≥3.5. Total apolipoprotein B levels derived from non-HDL cholesterol/apolipoprotein B ratios alone (equations 1A and 2A) and by regression analysis (equations 1B and 2B) were compared to apolipoprotein B determined by INA (Medical Research Laboratory International). For the subgroup with triglyceride/HDL cholesterol <3.5, mean apolipoprotein B level estimated by regression analysis (equation 1B) was essentially the same as when apolipoprotein B was estimated by the non-HDL cholesterol/apolipoprotein B ratio (equation 1A) and the 2 methods were lower than apolipoprotein B by INA. In the subgroup with triglyceride/HDL cholesterol ≥3.5, equations 2A and 2B produced results similar to apolipoprotein B obtained by INA. Therefore, there is no advantage in using equations 1B and 2B. Average apolipoprotein B levels by non-HDL cholesterol/apolipoprotein B ratios alone and by regression analysis were somewhat lower than that obtained by INA. In all categories non-HDL cholesterol/apolipoprotein B ratios were lower by Medical Research Laboratory International analysis.

Subgroup | Equation ^{⁎ } |
Apolipoprotein B (mg/dl) | r (p Value) ^{† } |
Non-HDL Cholesterol (mg/dl) | Total Apolipoprotein B/Non-HDL Cholesterol Ratio | Triglycerides (mg/dl) | HDL cholesterol (mg/dl) |
---|---|---|---|---|---|---|---|

Triglyceride/high-density lipoprotein cholesterol <3.5 (n = 390) | 1A | 106 ± 30 | 0.97 | 148 ± 42 | 1.39 ± 0 | 127 ± 36 | 55 ± 11 |

1B | 105 ± 23 | 0.98 | 1.38 ± 0 | ||||

INA ^{‡ } |
118 ± 31 ^{‡ } |
— | 1.25 ± 0.09 | ||||

Triglyceride/high-density lipoprotein cholesterol ≥3.5 (n = 705) | 2A | 151 ± 33 | 0.94 | 198 ± 43 | 1.39 ± 0 | 256 ± 87 | 41 ± 8 |

2B | 150 ± 28 | 0.94 | 1.32 ± 0 | ||||

INA ^{‡ } |
155 ± 31 ^{‡ } |
— | 1.28 ± 0.16 | ||||

Total apolipoprotein B (n = 1,095) | 1A + 2A | 130 ± 36 | 0.97 | 180 ± 50 | 1.39 ± 0 | 210 ± 96 | 46 ± 11 |

1B + 2B | 137 ± 37 | 0.97 | 1.32 ± 0 | ||||

INA ^{‡ } |
142 ± 36 ^{‡ } |
— | 1.27 ± 0.14 |

⁎ For triglyceride/high-density lipoprotein cholesterol ratio <3.5, apolipoprotein B (milligrams per deciliter) = Non-HDL cholesterol (milligrams per deciliter)/1.39 (equation 1A). For triglyceride/high-density lipoprotein cholesterol ratio >3.5, apolipoprotein B (milligrams per deciliter) = Non-HDL cholesterol/1.32 (equation 2A). An alternative approach is to determine equations relating apolipoprotein B to Non-HDL cholesterol for subjects with triglyceride/high-density lipoprotein cholesterol ratios split at 3.5: for triglyceride/high-density lipoprotein cholesterol ratio <3.5, apolipoprotein B = Non-HDL cholesterol/1.73 + 23.126 mg/dl (equation 1B); for triglyceride/high-density lipoprotein cholesterol ratio >3.5, apolipoprotein B = Non-HDL cholesterol/1.51 + 17.952 mg/dl (equation 2B).

† Pearson correlation coefficient (p <0.0001).

‡ Total apolipoprotein B measured by immunonephelometry at Medical Research Laboratory International.

Apolipoprotein B was estimated by VAP making use of the 3 VAP equations for different LDL patterns ( Table 2 ). The SAFARI cohort had 409 samples with VAP patterns A and A/B and 686 samples with pattern B at baseline. The difference between INA and VAP estimations of average apolipoprotein B level was 17%, with a Pearson correlation coefficient of 0.952 between the 2 measurements (p = 0.001). VAP-estimated apolipoprotein B levels for the 2 patterns (A plus A/B and B) were lower than those obtained by INA. This can be explained by lower non-HDL cholesterol/apolipoprotein B ratios observed by INA. That is, INA results suggest that non-HDL particles are relatively depleted in cholesterol compared to those determined by VAP.

Subgroup | Method | Apolipoprotein B (mg/dl) | Non-HDL Cholesterol (mg/dl) | Non-HDL Cholesterol/Total Apolipoprotein B | Triglycerides (mg/dl) | HDL Cholesterol (mg/dl) |
---|---|---|---|---|---|---|

Vertical auto profile pattern A + A/B (n = 409) | VAP | 103 ± 24 ^{⁎ } |
155 ± 44 | 1.50 ± 0.07 ^{⁎ } |
198 ± 105 | 50 ± 11 |

INA | 123 ± 34 | 159 ± 49 | 1.29 ± 0.19 | 160 ± 77 | 52 ± 12 | |

r ^{† } |
0.95 | 0.93 | — | 0.85 | 0.95 | |

Vertical auto profile pattern B (n = 686) | VAP | 127 ± 23 ^{⁎ } |
185 ± 41 | 1.44 ± 0.07 ^{⁎ } |
279 ± 98 ^{⁎ } |
42 ± 8 |

INA | 153 ± 32 | 193 ± 44 | 126 ± 0.09 | 240 ± 94 | 42 ± 9 | |

r ^{† } |
0.93 | 0.94 | — | 0.79 | 0.91 | |

Total group (n = 1,095) | VAP | 118 ± 24 ^{⁎ } |
174 ± 44 | 1.50 ± 0.08 | 249 ± 108 | 45 ± 9 |

INA | 142 ± 34 | 180 ± 47 | 1.26 ± 0.08 | 212 ± 95 | 45 ± 11 | |

r ^{† }(SEE) ^{‡ } |
0.94 (10.7) | 0.95 (13.2) | — | 0.83 (60.3) | 0.95 (3.6) |