Pemafibrate decreases triglycerides and small, dense LDL, but increases LDL-C depending on baseline triglycerides and LDL-C in type 2 diabetes patients with hypertriglyceridemia: an observational study

Background Pemafibrate, a selective PPARα modulator, has the beneficial effects on serum triglycerides (TGs) and very low density lipoprotein (VLDL), especially in patients with diabetes mellitus or metabolic syndrome. However, its effect on the low density lipoprotein cholesterol (LDL-C) levels is still undefined. LDL-C increased in some cases together with a decrease in TGs, and the profile of lipids, especially LDL-C, during pemafibrate administration was evaluated. Methods Pemafibrate was administered to type 2 diabetes patients with hypertriglyceridemia. Fifty-one type 2 diabetes patients (mean age 62 ± 13 years) with a high rate of hypertension and no renal insufficiency were analyzed. Pemafibrate 0.2 mg (0.1 mg twice daily) was administered, and serum lipids were monitored every 4–8 weeks from 8 weeks before administration to 24 weeks after administration. LDL-C was measured by the direct method. Lipoprotein fractions were measured by electrophoresis (polyacrylamide gel, PAG), and LDL-migration index (LDL-MI) was calculated to estimate small, dense LDL. Results Pemafibrate reduced serum TGs, midband and VLDL fractions by PAG. Pemafibrate increased LDL-C levels from baseline by 5.3% (− 3.8–19.1, IQR). Patients were divided into 2 groups: LDL-C increase of > 5.3% (group I, n = 25) and < 5.3% (group NI, n = 26) after pemafibrate. Compared to group NI, group I had lower LDL-C (2.53 [1.96–3.26] vs. 3.36 [3.05–3.72] mmol/L, P = 0.0009), higher TGs (3.71 [2.62–6.69] vs. 3.25 [2.64–3.80] mmol/L), lower LDL by PAG (34.2 [14.5, SD] vs. 46.4% [6.5], P = 0.0011), higher VLDL by PAG (28.2 [10.8] vs. 22.0% [5.2], P = 0.0234), and higher LDL-MI (0.421 [0.391–0.450] vs. 0.354 [0.341–0.396], P < 0.0001) at baseline. Pemafibrate decreased LDL-MI in group I, and the differences between the groups disappeared. These results showed contradictory effects of pemafibrate on LDL-C levels, and these effects were dependent on the baseline levels of LDL-C and TGs. Conclusions Pemafibrate significantly reduced TGs, VLDL, midband, and small, dense LDL, but increased LDL-C in diabetes patients with higher baseline TGs and lower baseline LDL-C. Even if pre-dose LDL-C remains in the normal range, pemafibrate improves LDL composition and may reduce cardiovascular disease risk.


Background
Major efforts have been made to lower cholesterol (particularly LDL-C) in daily clinical practice [1]. In the case of increased triglycerides (TGs), however, there are insufficient active interventions for reducing TGs, as the TG-lowering effect of existing fibrates is insufficient [2] and the risk of adverse effects of drug combination is increased. In addition, consistent evidence that lowering TG levels reduces the risk of cardiovascular events is not currently available in randomized trials [2].
It has been reported that serum LDL-C levels and arteriosclerosis risk are generally positively correlated and that the decrease in risk due to LDL-C-lowering therapy is proportional to LDL-C reduction rate [1]. So-called statins (HMG-CoA reductase inhibitors) and proprotein convertase subtilisin/kexin type 9 (PCSK-9) inhibitors to reduce LDL-C by approximately half at maximum are used for the purpose of preventing arteriosclerotic diseases including cardiovascular disease (CVD) [3,4]. At present, however, the residual risk of 50% or more has not been resolved, and indicators other than LDL-C level such as lower HDL-C and TG-rich lipoprotein (chylomicron remnants, VLDL remnants, or large VLDL etc.) have been regarded as therapeutic targets [5,6]. A largescale cohort study showed that the reduction of LDL-C was not so important for secondary prevention of CVD [7,8]. Results from large-scale clinical trials have shown that hypertriglyceridemia carries a residual risk of cardiovascular events even with statin use [9,10].
Pemafibrate (Palmodia®, a selective PPARα modulator), which was newly released in 2018, has a stronger TGlowering effect than existing fibrate preparations and can be used in combination with statins [11][12][13]. PPARα, a nuclear receptor expressed mainly in the liver, is involved in regulating genes associated with lipid metabolism. Pemafibrate selectively regulates target genes involved in lipid metabolism among these PPARαregulated genes [14]. The characteristic structure of pemafibrate enables alterations in gene transcription following recruitment of different cofactors [15].
There are many research reports that existing fibrates reduce LDL-C by about 10% [16,17]. The present study clarified the relationship between lipoprotein fractions, LDL-C, and TGs before and after pemafibrate administration to Japanese type 2 diabetes patients with hypertriglyceridemia.

Patients and study procedures
The target subjects were outpatients with type 2 diabetes and lifestyle-related diseases visiting Medical Plaza Daido Central. Patients who were heavy drinkers were excluded from this retrospective study. Patients with eGFR less than 45 ml/min were also excluded because TG-rich lipoprotein and LDL-C accumulate in CKD, especially early stage CKD [18]. Seventy-two patients with type 2 diabetes were enrolled in this study, and 21 patients receiving statins, ezetimibe, or conventional fibrates were excluded, so 51 type 2 diabetes patients were recruited. All patients received pemafibrate 0.2 mg (0.1 mg twice daily) [11]. All patients did not change their exercise or dietary regimens including alcohol consumption, during the entire study period. Sodiumglucose cotransporter 2 inhibitors having lipidimproving activity were not administered during the study period.
Lipoprotein electrophoresis (polyacrylamide gel, PAG) was examined in 42 patients from − 8 weeks before administration to just before pemafibrate administration, and in 40 patients between 4 and 12 weeks after administration. PAG electrophoresis is covered by public health insurance and is often used as a routine diagnostic procedure for dyslipidemia in Japan. PAG electrophoresis revealed 4 lipoprotein fractions (HDL, VLDL, midband and VLDL). The LDL-migration index (LDL-MI) was calculated from the pattern of PAG electrophoresis according to a previous report [19]; that is, the PAG electrophoretic distance between the LDL and VLDL fractions was divided by that between the HDL and VLDL fractions. When this value was > 0.400, it was determined to be an increase in small, dense LDL (sd-LDL) [19,20]. LDL-C was measured by the direct method using Metabolead® LDL-C (Hitachi Kasei Diagnostic Systems, Tokyo, Japan) [21].
Since TG levels are apt to fluctuate under the influence of diet [10], it was confirmed that fasting TG was > 1.69 mmol/L (150 mg/dL) by repeated measurements. Blood samples were collected after 9-12 h fasting. Serum lipids were monitored 4-8 weeks before administration. The average of the results at 8 weeks before (week − 8), at 4 weeks before (week − 4), and at start of pemafibrate (week 0) was used as the baseline value. Blood samples were analyzed every 4-8 weeks after pemafibrate, up to 24 weeks, and the average of the values was taken as the mean post-dose value, and the LDL-C increase rate was calculated using baseline and post-dose values. Assuming that the LDL-C increase rate varied among individual patients, 51 patients were divided into 2 groups, the increased LDL-C group (group I) and the no LDL-C increase group (NI group), with the median LDL-C increase rate as the boundary. Results of basal clinical parameters were shown as the mean (SD). Results of lipid and liver function tests were shown as the median (25-75% quartile, IQR) due to non-parametric distribution.

Statistical analysis
Statistical analysis was performed by t-test, paired t-test, Wilcoxon signed-rank test, Mann-Whitney U test, Kruskal-Wallis test, χ 2 test. Linear regression analysis was applied using least-square method. Multivariate regression analysis was performed using 9 baseline variables (LDL-C, TGs, HDL-C, Non − HDL-C, HDL, LDL, midband, VLDL and LDL-MI) to examine determinants for the LDL-C increase rate. JMP for Windows version 12 software (SAS Institute Japan; Tokyo, Japan) was used for statistical analyses. P values of < 0.05 were considered statistically significant.

Changes in lipid profile after pemafibrate administration
The clinical background of 51 target patients was shown in Table 1. The mean age was 62.2 (SD 12.9) years, the BMI was 26.6 (3.7) kg/m 2 , and the proportion of males was 55%. The average HbA1c was 7.4% (1.4), and the average eGFR was 71.9 (22.3) ml/min/1.73 m 2 . Hypertension and CVD/stroke complications were 59 and 18%, respectively.
The LDL-C increase rate after pemafibrate administration varied from patient to patient. The LDL-C increase rate fluctuated significantly from − 27.8 to 125.6% depending on the case before and after administration with pemafibrate. Thirty-five of 51 patients (69%) showed LDL-C increase. Since the median of LDL-C increase rate in 51 target patients was 5.3% (IQR − 3.8-19.1), patients with an LDL-C increase rate of > 5.3% were defined as the increased LDL-C group (group I; 25 cases) and the median LDL-C increase rate in group I after administration was 19.1% (IQR 13.3-56.3). Those with an LDL-C increase rate < 5.3% were defined as the no LDL-C increase group (NI group; 26 cases) and the median LDL-C increase rate in group NI after administration was − 3.7% (− 13.0-1.2) ( Table 1). As shown in Table 1, there were no significant difference in basal clinical background between the 2 groups.
As mentioned above, LDL-C increased notably after administration in group I; therefore, lipid profiles were compared between groups I and NI in next step. As shown in Table 2 and Fig. 1, the increasing tendency in LDL-C continued throughout the administration period (P = 0.0483, Kruskal-Wallis test). Comparing LDL-C levels in group I between baseline and post-dose, the median LDL-C increased from 2.53 (IQR 1.96-3.26) to 3.31 (2.77-4.11) mmol/L after administration (P < 0.0001, Wilcoxon signed-rank test). Group I was also characterized by high TGs before administration compared with group NI (3. 45-5.20] mmol/L) and decreased in both groups after pemafibrate administration. Table 3 showed the results of lipoprotein fraction analyses and LDL-MI in PAG electrophoresis. PAG electrophoresis revealed 4 lipoprotein fractions (HDL, VLDL, midband and VLDL) but there were cases where the midband fraction does not exist. Before pemafibrate administration, the midband fraction was high in both groups. The increase of midband suggests the increases of intermediate-density lipoprotein, VLDL remnant or remnant-like particle cholesterol, and relates with CVD [22]. The VLDL fraction was higher in group I than in group NI (28.2 [SD 10.8] vs. 22.0 [5.2], P = 0.0234, ttest). Furthermore, the LDL fraction was lower in group I than group NI (34.2 [14.5] vs. 46.4% [6.5], P = 0.0011). Lipoprotein PAG electrophoresis revealed a decrease in the VLDL fraction after pemafibrate administration but did not indicate the complete disappearance of the midband fraction in either group. The LDL fraction in group I increased from 34.2 [14.5] to 47.8% [10.9] after pemafibrate (P < 0.002). However, there was no difference in lipoprotein fractions between the groups after pemafibrate administration ( Table 3).
The LDL-MI before pemafibrate in group I was 0.421 (0.391-0.450), exceeding 0.400, which was significantly different from the NI group (0.354 [0.341-0.396], P < 0.0001, Mann-Whitney U test). These results suggest that sd-LDL increased in group I. The LDL-MI in group I dropped to 0.367 (0.344-0.389) after pemafibrate (P = 0.0002), but it was > 0.400 in some cases. However, the LDL-MI was almost the same in both groups after pemafibrate. Figure 2a showed the correlation between LDL-MI and LDL fraction before pemafibrate in all measured cases. A significant inverse correlation was observed (R 2 = 0.4283, P < 0.0001, least squares). Figure 2b showed the correlation between LDL-MI before pemafibrate and LDL-C increase rate before and after pemafibrate. A significant positive correlation was observed (R 2 = 0.6017, P < 0.0001).

Adverse effects
Of the 51 cases, one patient exhibited muscle pain symptoms, but no significant increase in CPK was observed.

Discussion
This study demonstrated the long-term effect of pemafibrate on LDL-C levels for 24 weeks in patients with type 2 diabetes with hypertriglyceridemia. Pemafibrate 0.2 mg (0.1 mg twice daily) significantly reduced serum TGs and VLDL by PAG electrophoresis and slightly increased HDL-C following administration to type 2 diabetes patients with hypertriglyceridemia. Although LDL-C levels were not considerably altered, pemafibrate for 24 weeks increased LDL-C levels by 5.9% as median. The increase in LDL-C was stably maintained over 24 weeks. LDL-C levels decreased in only 31% (16 of 51) patients after pemafibrate. Furthermore, It was demonstrated for the first time that the increase in LDL-C after pemafibrate had lower LDL-C, higher TGs, higher Non − HDL-C, higher sd-LDL and higher midband at baseline. It should be noted that there was a considerable number of cases of increased LDL-C in this study during treatment with pemafibrate. Previous study showed that the baseline TGs and LDL-C were key determinants of the changes in LDL-C [12]. However, the existence of an  .80 mmol/L) in diabetes patients, and LDL-C increase rate varied according to the baseline LDL-C level, with a significant increase in < 120 mg/dL (3.10 mmol/L) of LDL-C [26]. It was reported that a similar increase effect of LDL-C was brought by weight loss in hypertriglyceridemic patients [27]. This relationship between baseline TGs and LDL-C can be explained by the precursor-product relationship between VLDL and LDL, LDL is produced as an ultimate product of the lipolytic conversion of VLDL [28]. Fibrates, including pemafibrate, activate lipoprotein lipase (LPL) and hepatic triglyceride lipase to promote the catabolism TG-rich lipoprotein and also attenuate VLDL synthesis to reduce serum TGs [29]. It is assumed that LDL and HDL are produced during the catabolism of the TG-rich lipoprotein [27,28]. Since pemafibrate has a stronger TG-rich lipoprotein catabolism than existing fibrates, it is presumed that pemafibrate increases LDL-C and changes LDL-C composition [11][12][13]30]. Pemafibrate has been reported to increase the size and decrease the number of LDL particles [24]. The increase in LDL-C after pemafibrate treatment suggests the actual improvement in lipoprotein metabolism. On the other hand, the increase in LDL-C levels with TG reduction by pemafibrate might lead clinicians to question the clinical efficacy of the treatment. Further studies are needed to define the mechanisms underlying the variability of the effects of pemafibrate on LDL-C.
The baseline LDL-MI in group I exceeded 0.400, suggesting that sd-LDL increased [30]. This study thus revealed that baseline sd-LDL increases and significantly improves after pemafibrate administration. Higher baseline TGs and lower baseline LDL-C appear to be involved in the baseline sd-LDL increase and post-dose  Table 4 Multivariate regression analysis between LDL-C increase rate and baseline lipid parameters LDL-C increase. The ARIC study reported that the risk of CVD was associated with an increase in sd-LDL-C rather than large buoyant LDL-C [31]. In a study of ischemic heart disease in elderly Japanese men, patients with high levels of sd-LDL had increased risk of CVD events over the next 5 years [32]. In the present study, CVD/ stroke complications were not higher in group I, but the number of cases was too small to draw conclusions. Higher values of sd-LDL occur when both non − HDL-C and TGs are high [33]. As the present study showed an inverse correlation between baseline LDL-MI and baseline LDL fraction by PAG electrophoresis, it was suggested that an increase in sd-LDL and a decrease in the LDL fraction are synchronized phenomenon associated with lipid metabolism in patients with type 2 diabetes. There was a positive correlation between baseline LDL-MI and the LDL-C increase rate. This strongly suggests that pemafibrate decreases sd-LDL and increases large buoyant LDL in type 2 diabetes patients.
PCSK9 inhibitors, monoclonal antibodies that bind to free PCSK9, have potent effect on LDL-C reduction, and are recommended to use in patients with high CVD risk or familial hypercholesterolemia [34]. Whether or not PCSK9 inhibitors reduce CVD events is not yet to be clarified [35]. Moreover, PCSK9 inhibitors seem to be less efficient lowering sd-LDL [36], different from the pemafibrate effect. Week −4 Week 0 Week 4 Week 8 Week 12 Week 16 Week 20 Week 24 Hypertriglyceridemia is defined as a fasting TG value of ≥1.69 mmol/L (150 mg/dL). As in hypercholesterolemia (mainly increased LDL-C), hypertriglyceridemia has been epidemiologically associated with many atherosclerotic diseases including CVD [37]. However, unlike cholesterol, TGs do not accumulate in atherosclerotic plaques on the walls of blood vessels, and TGs per se do not promote atherosclerosis [38]. In general, when serum TG concentrations rise, the cholesterol contained in TG-rich lipoprotein increases and total cholesterol concentrations also rise [39]. In patients with hypertension and/or insulin resistance, the metabolism of lipoproteins is delayed and they remain in the blood circulation for a variety of reasons [40]. The retained remnants are taken up by macrophages, and cholesterol accumulates in atherosclerotic lesions [41].
It is well known that VLDL and LDL are apo Bcontaining lipoproteins associated with arteriosclerosis. Lowering LDL-C by 40 mg/dL (1.03 mmol/L) has been reported to reduce the risk of cardiovascular events by 20% [6]. Assuming that all apo B-containing particles have almost the same atherogenic effect [6], the TG value needs to be reduced by 5 times that of LDL-C, approximately 200 mg/dL (2.26 mmol/L), from a simple interpretation of Friedewald's formula. However, the TGlowering effect of existing fibrates is not significant, making it impossible to significantly reduce the risk of major cardiovascular events [23]. On the other hand, in present study, pemafibrate significantly decreased TGs from 3.30 to 2.15 mmol/L in all cases and from 3.71 to 2.11 mmol/L in patients with higher LDL-C increase rate. It was concluded that this significant reduction in TGs caused a decrease in large VLDL and a change in LDL composition [42]. This study suggested that pemafibrate increased LDL-C not by increasing the particle number of LDL, but rather by increasing the cholesterol content of LDL.
In hypertriglyceridemia, VLDL cholesterol and/or non − HDL-C are increased, which has atherosclerosisinducing properties similar to or higher than LDL cholesterolemia [43]. According to current guidelines for arteriosclerosis, non − HDL-C should be evaluated instead of LDL-C in cases such as severe hypertriglyceridemia [5]. The LDL-C value is said to underestimate the cardiovascular risk. When the non − HDL-C level is high, cholesterol lowering therapy is prioritized, as in LDL cholestrolemia [5,44]. Non − HDL-C was certainly high before pemafibrate and decreased after administration in this study.

Study strength and limitations
There are several study strengths in this study. First, the reliable LDL-C direct assay, a Metabolead LDL-C® (Hitachi Kasei Diagnostic Systems), could be used for LDL-C estimation, and the results were consistent with the lipoprotein PAG electrophoresis results. Moreover, this direct method has already been shown to be consistent with ultracentrifugation, unless the TGs exceed 11.29 mmol/L (1,000 mg/dL) [21]. There were no patients with a fasting TG level of 11.29 mmol/L or higher in this study, therefore, any effect of hypertriglyceridemia on the LDL-C assay (false low value) could be ruled out. Second, serum lipids were followed every 4-8 weeks for 24 weeks to evaluate lipids with variable values. Measuring the lipid profile several times is more accurate than simply assessing it before and after pemafibrate administration. Finally, PAG electrophoresis, which is a simple and inexpensive method, was used for the estimation of sd-LDL and lipoprotein fractions. Thus, in some diabetes cases, LDL-C was relatively low before pemafibrate and LDL-C increased after pemafibrate, demonstrating that the composition of LDL was significantly changed due to the TG-rich lipoprotein lowering effect of pemafibrate.
This study has several limitations. First, instead of directly measuring sd-LDL, determination of LDL-MI by PAG electrophoresis was used as a substitute. However, many reports indicated that the results of both are strongly correlated [19,20]. Second, there were not many target patients in this study, and because of the retrospective nature, results of lipoprotein PAG electrophoresis were not obtained in several patients. However, statistically significant results were obtained despite the small number of patients. It would be needed to increase the number of target patients in the future. Finally, this study did not show results for serum apolipoproteins such as apo B and apo E. It cannot be ruled out that present study may include patients with type III hyperlipidemia [45]. As the patients had a high proportion of combined hyperlipidemia, it might be close to the "lipoprotein abnormality similar to type III hyperlipidemia" reported by Matsuzawa et al. [46]. Cardiovascular events are frequently observed in "lipoprotein abnormalities similar to type III hyperlipidemia", so the future risk of CVD/stroke complications in the examined cases should be followed closely [13].

Conclusion
The superiority of pemafibrate allows control of serum TG levels and sd-LDL, which were previously inadequate with conventional treatment [2,11,22]. At the same time, there were cases in which LDL-C levels fluctuate markedly before and after pemafibrate administration. It was also shown that measurement of LDL-C by the direct assay instead of the recommended measurement of non − HDL-C is useful in the pathologic evaluation of type 2 diabetes patients with high TGs. It is necessary to review recommendations in the arteriosclerosis guideline [5]. Moreover, the direct LDL-C assay combined with lipoprotein PAG electrophoresis enables easy evaluation of TG-rich lipoprotein. Even in statin-treated type 2 diabetes patients whose LDL-C remains within the therapeutic range, clear increases in sd-LDL with hypertriglyceridemia are inherent [31], pemafibrate or ezetimibe, which have a TG-rich lipoprotein-lowering effect, might be considered as drugs for add-on therapy [47]. The above data suggest a need to review the diagnostic indices and control standards for lipids in patients with type 2 diabetes.