RESULTS
In line with our previous observations in 12 patients (11), a subset of the 27 patients who were included in the current study, the VLCD profoundly reduced bodyweight from 113.1 ± 3.7 to 87.7 ± 2.9 kg (P < 0.05) and decreased BMI from 37.2 ± 0.9 to 28.9 ± 0.8 kg/m2 (P < 0.05). In addition, in the 12 patients from whom 1H-MRS scans could be obtained, hepatic TG content reduced considerably from 21.2 ± 4.2 to 3.0 ± 0.9% (n = 12; P < 0.001) as reported previously (11).

Plasma CETP and (apo)lipoproteins
When compared with baseline, VLCD decreased plasma CETP concentration (−18.2%; P < 0.01). In addition, VLCD reduced plasma levels of total cholesterol (−13.1%; P < 0.001), TG (−45.1%; P < 0.001), phospholipid (−15.2%; P < 0.0001), LDL-cholesterol (−15.8%; P < 0.01), and apoB100 (−13.9%; P < 0.01). VLCD did not alter plasma HDL-cholesterol and HDL-phospholipids, but increased apoAI (+16.2%; P < 0.05) (Table 1). The change in body weight after 16 weeks of VLCD did not correlate with the change in either plasma TG (R2 = 0.0000; P = 0.9952), total cholesterol (R2 = 0.0471; P = 0.2769), phospholipid (R2 = 0.0305; P = 0.3837), LDL-cholesterol (R2 = 0.0320; P = 0.3717), HDL-cholesterol (R2 = 0.0086; P = 0.6460), or HDL-phospholipid (R2 = 0.0013; P = 0.8570).
Table 1 Plasma CETP and (apo)lipoprotein levels in obese patients with type 2 diabetes mellitus and hepatic steatosis in response to 16 weeks of VLCD
Plasma parameters Baseline After VLCD Δ (%) P value
CETP (μg/mL) 2.48 ± 0.15 2.03 ± 0.14 −18.2 0.0021
Total cholesterol (mmol/L) 5.76 ± 0.30 5.00 ± 0.22 −13.1 0.0007
Triglycerides (mmol/L) 2.41 ± 0.28 1.32 ± 0.10 −45.1 0.0003
Phospholipids (mmol/L) 2.69 ± 0.11 2.28 ± 0.07 −15.2 0.0000
LDL-cholesterol (mmol/L) 3.99 ± 0.27 3.36 ± 0.20 −15.8 0.0028
ApoB100 (mg/dL) 130 ± 6 111 ± 5 −13.9 0.0016
HDL-cholesterol (mmol/L) 0.84 ± 0.04 0.91 ± 0.06 — NS
HDL-phospholipids (mmol/L) 0.98 ± 0.03 0.98 ± 0.05 — NS
ApoAI (mg/dL) 135 ± 10 156 ± 11 +16.2 0.0174
Data are presented as means ± SEM (n = 27). Δ values are calculated by comparing values obtained after VLCD with those obtained at baseline from obese patients with type 2 diabetes mellitus. P values are calculated using paired Student t test. NS, not significant.

Cholesterol efflux
VLCD decreased cholesterol efflux from THP-1 cells to total plasma obtained from patients compared with plasma from baseline (−14.5%; P < 0.001; Fig. 1A). Similarly, the capacity of apoB-depleted plasma obtained after VLCD to promote cholesterol efflux was lower than that of apoB-depleted plasma obtained at baseline (−14.9%; P < 0.001; Fig. 1B).
Figure 1 Cholesterol efflux to total plasma and apoB-depleted plasma. THP-1 cells were loaded with [3H]cholesterol and incubated for 4 h at 37°C with total plasma (1% vol/vol; A) or apoB-depleted plasma (1% vol/vol; B), obtained before and after VLCD from 27 obese patients with type 2 diabetes mellitus. Cholesterol efflux rate is calculated by dividing 3H-activity in the medium by the sum of the 3H-radioactivity in the medium and cell extract. Data are means ± SEM. P values are calculated using paired Student t test. ***P < 0.001 as compared with baseline.Correlation analysis showed that cholesterol efflux to total plasma positively correlated with plasma total cholesterol (R2 = 0.2416; P < 0.001) and plasma total phospholipid (R2 = 0.3499; P < 0.001). Cholesterol efflux to total plasma positively correlated with non-HDL-cholesterol (R2 = 0.2339; P < 0.001) and non-HDL-phospholipid (R2 = 0.2855; P < 0.001) rather than HDL-cholesterol or HDL-phospholipid (both P > 0.05; Supplementary Fig. 1). Moreover, no significant correlation was observed between cholesterol efflux to plasma and apoAI (Supplementary Fig. 2).