Study population
We included 340 subjects with confirmed FCHL diagnosis and available VLDL-C measures.
The median age of patients at diagnosis was 47.0 (35.0–58.0) years, 65% were women, 12.0% were under statin treatment and 19.11% had type 2 diabetes (T2D). Overall, we identified 137 (40.3%) subjects who satisfied the diagnosis of isolated hypercholesterolemia and 203 (59.7%) who belonged to the mixed dyslipidemia phenotype, we did not observe subjects who completed criteria for isolated hypertriglyceridemia. On comparing differences across FCHL phenotypes, in the mixed dyslipidemia phenotype the age at diagnosis was highest, fewer patients were women, more often had T2D and more patients were under statin treatment compared to the isolated hypercholesterolemia phenotype (p < 0.010). As expected, subjects with mixed dyslipidemia had higher values of apolipoprotein B, non-HDL cholesterol, LDL-C and VLDL-C (p < 0.001, Table 1).
Table 1
Biochemical and clinical characteristics of patients with FCHL in the overall population and stratified by FCHL dyslipidemia phenotype.
Variable
|
Overall
n = 340
|
Isolated hypercholesterolemia
n = 137
|
Mixed dyslipidemia
n = 203
|
p
|
Sex (female)
|
221 (65.0)
|
105 (76.6)
|
116 (57.1)
|
< 0.001
|
Age (years)
|
47.0 (35.0–58.0)
|
43.0 (32.0–57.0)
|
48.0 (37.0–58.0)
|
0.019
|
Type 2 Diabetes (%)
|
65 (19.1)
|
10 (7.3)
|
55 (27.1)
|
< 0.001
|
Hypertension (%)
|
70 (20.6)
|
21 (15.3)
|
49 (24.3)
|
0.046
|
Total cholesterol (mg/dL)
|
209.0 (179.0-241.5)
|
179.0 (160.0-198.8)
|
226.5 (206.0-266.8)
|
< 0.001
|
HDL cholesterol (mg/dL)
|
42.0 (35.0-48.8)
|
47.0 (41.0–54.0)
|
38.0 (33.0–44.0)
|
< 0.001
|
Non-HDL cholesterol (mg/dL)
|
168.0 (133.0-198.0)
|
129.5 (112.0-154.8)
|
188.0 (168.3–227.0)
|
< 0.001
|
Triglycerides (mg/dL)
|
182.5 (107.3-310.3)
|
99.0 (73.0-122.3)
|
271.0 (205.5-394.8)
|
< 0.001
|
Apolipoprotein B (mg/dL)
|
116.0 (90.0-136.8)
|
87.0 (72.9-103.8)
|
128.5 (114.3-148.8)
|
< 0.001
|
VLDL-Triglycerides (mg/dL)
|
120.8 (61.3-240.1)
|
54.7 (34.0-73.9)
|
211.0 (144.2-329.9)
|
< 0.001
|
VLDL-Cholesterol (mg/dL)
|
32.4 (16.5–52.5)
|
14.7 (9.1–19.7)
|
49.0 (36.1–67.4)
|
< 0.001
|
VLDL-Cholesterol Martin (mg/dL)
|
32.9 (21.3–46.0)
|
19.7 (16.0-22.9)
|
43.5 (34.5–56.4)
|
< 0.001
|
VLDL-Cholesterol Sampson (mg/dL)
|
35.1 (18.6–54.4)
|
16.5 (12.0-20.9)
|
51.6 (37.8–69.7)
|
< 0.001
|
VLDL-Cholesterol Friedewald (mg/dL)
|
36.5 (21.5–62.1)
|
19.8 (14.6–24.5)
|
54.2 (41.1–79.0)
|
< 0.001
|
LDL-Cholesterol (mg/dL)*
|
127.7 (106.4-151.6)
|
114.2 (98.4-136.9)
|
139.0 (119.4-161.7)
|
< 0.001
|
LDL-cholesterol Martin (mg/dL)
|
130.1 (106.2-151.4)
|
109.4 (95.0-131.4)
|
142.9 (121.5-162.3)
|
< 0.001
|
LDL-Cholesterol Sampson (mg/dL)
|
127.8 (101.1-145.8)
|
111.8 (95.6-133.1)
|
134.6 (110.7-157.5)
|
< 0.001
|
LDL-Cholesterol Friedewald (mg/dL)
|
122.6 (97.2-142.4)
|
109.4 (94.4-130.9)
|
128.8 (102.1-153.5)
|
< 0.001
|
Statin treatment (%)
|
41 (12.0)
|
5 (3.6)
|
36 (17.7)
|
< 0.001
|
VLDL-C comparative assessment
For VLDL-C measured by ultracentrifugation, we observed the highest correlation for VLDL-C estimated by Sampsom’s formula (ρ = 0.937, 95%CI 0.921–0.948), followed by Martin’s (ρ = 0.935, 95%CI 0.921–0.948) and Friedewad’s (ρ = 0.933, 95%CI 0.917–0.945) formulas. VLDL-C estimation errors (RMSE) were also comparatively lower for Sampson’s formula, followed by Martin’s and Friedewald’s and were further reduced when only analyzing individuals with triglycerides < 800 mg/dL (Fig. 1A-C). Bland-Altman analyses showed smaller bias for Martin’s formula (d = 1.87, 95%CI 0.46,3.30) followed by Sampson’s (d=-2.09, 95%CI -3.29-0.90) and Friedewald’s formulas (d=-6.20, 95%CI 0.45–3.30, Figs. 1D-F) compared to LDL-C measured by ultracentrifugation.
LDL-C comparative assessment
We observed the highest correlation for Sampson’s formula, which also displayed the lowest RMSE and highest R2 despite having slightly higher bias compared to Martin’s (Table 2, Fig. 2); similarly, Sampson’s formula had lower bias compared to Martin’s and Friedewald’s formulas for LDL-C estimation. When assessing performance according to dyslipidemia phenotypes, Friedewald’s and Sampson’s formulas had similar RMSE and linear correlation, which were higher than Martin’s for isolated hypercholesterolemia. However, performance of Sampson’s formula drastically improved in mixed dyslipidemia compared to other methods. Comparing correlation strength across triglyceride levels, Sampson’s showed consistently improved correlations compared to Martin’s and Friedewald’s formulas for triglyceride categories < 400mg/dL [LDL-S: ρ = 0.962, 95% CI 0.952–0.970; LDL-M: ρ = 0.956, 95%CI 0.945–0.965; LDL-F: ρ = 0.955, 95%CI 0.944–0.964] and even > 400mgdL [LDL-S: ρ = 0.642, 95%CI 0.445–0.779; LDL-M: ρ = 0.508, 95%CI 0.270–0.687; LDL-F: ρ = 0.577, 95%CI0.359-0.736]. Nevertheless, Sampson’s formula had slightly higher bias compared to Martin’s when compared using Bland-Altman analyses.
Table 2. Performance metrics for all three formulas compared to LDL-C estimated using VLDL-C measured by ultracentrifugation in the overall population and stratified by FCHL dyslipidemia phenotype.
Metric
|
LDL-F
|
LDL-M
|
LDL-S
|
Isolated Hypercholesterolemia
|
Mixed dyslipidemia
|
LDL-F
|
LDL-M
|
LDL-S
|
LDL-F
|
LDL-M
|
LDL-S
|
ρ (95%CI)
|
0.895
(0.872-0.915)
|
0.899
(0.876-0.917)
|
0.917
(0.899-0.932)
|
0.962
(0.947-0.973)
|
0.957
(0.941-0.969)
|
0.961
(0.9460.972)
|
0.855
(0.814-0.889)
|
0.871
(0.834-0.901)
|
0.875
(0.838-0.904)
|
ρ with ApoB (95%CI)
|
0.644
(0.577-0.702)
|
0.788
(0.744-0.825)
|
0.704
(0.646-0.754)
|
0.856
(0.825-0.882)
|
0.868
(0.8400.892)
|
0.862
(0.832-0.887)
|
0.628
(0.558-0.688)
|
0.729
(0.675-0.775)
|
0.662
(0.598-0.718)
|
R2
|
0.802
|
0.807
|
0.840
|
0.645
|
0.606
|
0.614
|
0.731
|
0.769
|
0.782
|
RMSE
|
44.96
|
30.22
|
19.99
|
10.74
|
10.98
|
10.19
|
44.44
|
29.41
|
18.41
|
Bias (95%CI)
|
12.33
(7.71,16.95)
|
1.12
(-2.10,4.35)
|
4.59
(2.51,6.67)
|
2.85
(2.16,5.56)
|
4.14
(2.41,5.86)
|
2.29
(0.59,4.00)
|
18.05
(10.47,25.63)
|
-0.91
(-6.19,4.37)
|
6.14
(2.86,9.42)
|
Abbreviations= RMSE: Root of Mean Squared Error; 95%CI: 95% Confidence Interval; LDL-F: LDL-C estimated by the Friedewald equation; LDL-M: LDL-C estimated by Martin’s formula; LDL-S: LDL-C estimated by Sampson’s formula.
Apo B comparative assessment
When we compared the correlations between Apo B and LDL-C estimated by the three equations, we observed the highest correlation for Martin’s formula overall and in isolated hypercholesterolemia, but Sampson’s had slightly higher correlation in mixed dyslipidemia compared to Martin’s equation (Table 2). However, comparing correlations strength across triglyceride levels, Martin’s showed consistently improved correlations compared to Sampson’s and Friedewald’s formulas for triglyceride categories < 400mg/dL [LDL-M: ρ = 0.853, 95%CI0.818-0.881; LDL-S: ρ = 0.806, 95%CI 0.761–0.843; LDL-F: ρ = 0.772, 95%CI 0.721–0.815] and 400-800mg/dL (LDL-M: ρ = 0.853, 95%CI 0.727–0.924; LDL-S: ρ = 0.843, 95%CI 0.710–0.918; LDL-F: ρ = 0.836, 95%CI 0.697–0.914). However, Sampson’s formula had the highest correlation compared to the other equations for triglycerides > 800mg/dL (LDL-S: ρ = 0.586, 95%CI 0.081–0.852; LDL-M: ρ = 0.088, 95%CI -0.464-0.590; LDL-F: ρ = 0.082, 95%CI -0.468-0.587).
Comparison of LDL-C formulas for LDL-C and Apo B targets
When assessing concordance in lipid target goals (LDL-C < 100mg/dL), the highest concordance and AUROC were observed for Sampson’s formula overall and in both isolated hypercholesterolemia and mixed dyslipidemia (Table 3). For more stringent lipid targets (LDL-C < 70mg/dL) Sampson’s formula had lower concordance compared to Martin’s, but higher AUROC, which was comparable to Friedewald’s equation in isolated hypercholesterolemia. In mixed dyslipidemia, Martin’s equation had highest concordance, but Sampson’s formula had highest AUROC. This is consistent with previous findings relating LDL-M in FCHL. Finally, when assessing concordance in lipid target goals (ApoB < 80mg/dL) the highest concordance and AUROC were observed for Martin’s formula overall and in both isolated hypercholesterolemia and mixed dyslipidemia. For more stringent lipid targets (ApoB < 65mg/dL), Martin’s showed consistently improved concordances and AUROC compared to Sampson’s and Friedewald’s formulas overall and in both phenotypes.
Table 3. Comparison of lipid targets for all three formulas compared to LDL-C estimated using VLDL-C measured by ultracentrifugation in the overall population and stratified by FCHL dyslipidemia phenotype.
Metric
|
LDL-F
|
LDL-M
|
LDL-S
LDL-F
|
Isolated Hypercholesterolemia
|
Mixed dyslipidemia
|
LDL-M
|
LDL-S
|
LDL-F
|
LDL-M
|
LDL-S
|
|
LDL-C goal <100mg/dL
|
κ (95%CI)
|
0.730
(0.642-0.818)
|
0.764
(0.674-0.854)
|
0.819
(0.740-0.997)
|
0.723
(0.630-0.843)
|
0.752
(0.636-0.868)
|
0.779
(0.666-0.892)
|
0.709
(0.568-0.850)
|
0.731
(0.566-0.897)
|
0.852
(0.735-0.968)
|
AUC (95%CI)
|
0.923
(0.884-0.961)
|
0.892
(0.835-0.948)
|
0.933
(0.890-975)
|
0.898
(0.839-0.956)
|
0.909
(0.851-0.965)
|
0.911
(0.851-0.970)
|
0.938
(0.883-0.994)
|
0.815
(0.690-0.940)
|
0.943
(0.871-1.00)
|
LDL-C goal <70mg/dL
|
κ (95%CI)
|
0.338
(0.127-0.549)
|
0.506
(0.252-0.759)
|
0.462
(0.216-0.707)
|
0.560
(0.115-1.00)
|
0.453
(0.012-0.839)
|
0.560
(0.115-1.00)
|
0.279
(0.054-0.504)
|
0.557
(0.242-0.872)
|
0.424
(0.136-0.712)
|
AUC (95%CI)
|
0.870
(0.750-0.990)
|
0.869
(0.742-0.995)
|
0.878
(0.759-0.997)
|
0.730
(0.412-1.00)
|
0.727
(0.405-1.00)
|
0.731
(0.409-1.00)
|
0.942
(0.877-1.00)
|
0.929
(0.816-1.00)
|
0.951
(0.877-1.00)
|
Abbreviations= AUROC: Area Under the ROC Curve; 95%CI: 95% Confidence Interval; LDL-F: LDL-C estimated by the Friedewald equation; LDL-M: LDL-C estimated by Martin’s formula; LDL-S: LDL-C estimated by Sampson’s formula.