Long-term efficacy of lipoprotein apheresis and lomitapide in the treatment of homozygous familial hypercholesterolemia (HoFH): a cross-national retrospective survey

Background Homozygous familial hypercholesterolemia (HoFH) is a rare life-threatening condition that represents a therapeutic challenge. The vast majority of HoFH patients fail to achieve LDL-C targets when treated with the standard protocol, which associates maximally tolerated dose of lipid-lowering medications with lipoprotein apheresis (LA). Lomitapide is an emerging therapy in HoFH, but its place in the treatment algorithm is disputed because a comparison of its long-term efficacy versus LA in reducing LDL-C burden is not available. We assessed changes in long-term LDL-C burden and goals achievement in two independent HoFH patients’ cohorts, one treated with lomitapide in Italy (n = 30) and the other with LA in France (n = 29). Results The two cohorts differed significantly for genotype (p = 0.004), baseline lipid profile (p < 0.001), age of treatment initiation (p < 0.001), occurrence of cardiovascular disease (p = 0.003) as well as follow-up duration (p < 0.001). The adjunct of lomitapide to conventional lipid-lowering therapies determined an additional 58.0% reduction of last visit LDL-C levels, compared to 37.1% when LA was added (padj = 0.004). Yearly on-treatment LDL-C < 70 mg/dl and < 55 mg/dl goals were only achieved in 45.5% and 13.5% of HoFH patients treated with lomitapide. The long-term exposure to LDL-C burden was found to be higher in LA than in Lomitapide cohort (13,236.1 ± 5492.1 vs. 11,656.6 ± 4730.9 mg/dL-year respectively, padj = 0.002). A trend towards fewer total cardiovascular events was observed in the Lomitapide than in the LA cohort. Conclusions In comparison with LA, lomitapide appears to provide a better control of LDL-C in HoFH. Further studies are needed to confirm this data and establish whether this translates into a reduction of cardiovascular risk. Supplementary Information The online version contains supplementary material available at 10.1186/s13023-021-01999-8.


Background
Homozygous familial hypercholesterolemia (HoFH) is a rare inherited disorder of lipid metabolism characterized by marked elevation of low-density lipoprotein cholesterol (LDL-C) caused by an almost abolished function of LDL receptor (LDLR) [1]. Typically, HoFH patients develop clinical complications of atherosclerotic cardiovascular disease (ASCVD) early in life [2][3][4][5] due to the very high life-long LDL-C burden associated with this condition [6,7]. Therefore, an early, intensive LDL-C lowering therapy is mandatory to control the high risk of ASCVD in HoFH.
Lipoprotein apheresis (LA) in addition to statins and ezetimibe is considered the standard of therapy in HoFH [1,8]. LA transiently removes LDL particles from plasma during extracorporeal circulation by selectively binding apolipoprotein B (ApoB)-containing lipoproteins and has been demonstrated to be very effective in lowering LDL-C levels [8,9]. However, an optimal inter-procedures LDL-C control is difficult to reach due to the post-treatment LDL-C rebound and, therefore, the residual cardiovascular risk of these patients remains high [10].
In recent years, promising new drugs in the treatment of HoFH have become available [11]. Monoclonal antibodies inhibiting proprotein convertase subtilisin/kexin type 9 (PCSK9i) have been reported to decrease LDL-C by an additional 24% when added to background lipidlowering therapies in HoFH [11]. However, PCSK9i are mostly effective in patients with residual LDLR function, showing very limited effect in those carrying null mutation or homozygous mutation in the LDLRAP1 gene [12].
Lomitapide, an inhibitor of microsomal triglycerides transferase protein (MTP), has been proven to decrease by almost 50% LDL-C levels in HoFH in adjunct to other lipid-lowering medications (including LA) [13]. Recent real-world studies have confirmed the efficacy and safety of lomitapide in the treatment of adults and children with HoFH [11][12][13][14]. As lomitapide acts by decreasing liver secretion of VLDL and, consequently, VLDL-derived LDL production [15], its efficacy appears to be independent from the severity of functional impairment of LDLR pathway [16]. Some real-word studies have reported that the addition of lomitapide to background therapies results in the discontinuation of LA in many patients with HoFH [14].
The advent of novel therapeutic options is rapidly changing the therapeutic armamentarium for the management HoFH. Therefore, some efforts must be done to compare the effectiveness of emerging versus standard therapies in controlling LDL-C burden in these patients. Considering the rarity of the disease and the fact that large and formal randomized clinical trials aimed at comparing lomitapide versus LA are not feasible, we believe that a real-world survey might help in clarifying the benefit of these two treatments. However, a comparison of benefit of lomitapide versus LA in the long-term control of LDL-C burden in HoFH is not available to date.
Our aim was to indirectly compare the long-term LDLlowering effectiveness of these two treatments throughout the evaluation of two independent HoFH cohorts treated with lomitapide or LA. We therefore assessed LDL-C reduction, target achievement and long-term LDL-C burden in these two populations. As an exploratory analysis, we also aimed at evaluating major atherosclerotic cardiovascular events (MACE) incidence during follow-up in the two cohorts.

Results
The two cohorts differed significant for demographic, clinical, genetic, and biochemical characteristics as shown in Table 1. The list of causative HoFH mutations is reported in the Additional file 1: Table 1.
Patients from the LA cohort showed a more severe genotype as compared to the Lomitapide cohort (p = 0.004). The age of initiation of treatment with LA was earlier than that of lomitapide (p < 0.001). First ASCVD occurred earlier in the LA cohort (p = 0.003), while the overall prevalence of ASCVD did not differ between the two cohorts.
Patients in the Lomitapide cohort exhibited lower values of untreated (p < 0.001), lowest on conventional lipidlowering treatment (LLT) (p = 0.01) or baseline (p < 0.001) LDL-C as compared with those in the LA cohort.

Lipid changes during follow-up
The median follow-up duration was 16.7 years in the LA cohort and 2.3 years in Lomitapide cohort (p < 0.001) ( Table 2). Figure 1 shows changes in TC and LDL-C concentrations in the two cohorts before and after addition of lomitapide or LA. The addition of lomitapide to standard LLT resulted in a further 58.0% reduction in the last visit LDL-C, compared to 37.1% LDL-C reduction after the addition of LA (p = 0.013). These results were independent from genotype, ethnicity, untreated LDL-C levels, gender and age at baseline (Lomitapide vs. LA cohort β -0.673 p adj = 0.004, data not shown).
The yearly variation of LDL-C values in the two cohorts is reported in Fig. 2. The mean yearly on-treatment LDL-C percent reduction (Panel A) as well as the absolute yearly on-treatment LDL-C levels (Panel B) at 9 years were significantly lower in the Lomitapide cohort as compared with the LA cohort and this difference persisted after adjusting for genotype, gender, untreated LDL-C,   Finally, when we estimated the LDL-C burden, HoFH patients exposed to lomitapide exhibited a significantly lower total LDL-C burden as compared with patients in the LA cohort and this persisted after adjustment for age at follow-up, untreated LDL-C values, genotype, ethnicity, gender (11,656.6 ± 4730.9 mg/dL-year vs. 13,236.1 ± 5492.1 mg/dL-year, p adj = 0.002, Additional file 2: Fig. 1, Panel A). Indeed, as compared to values before adding lomitapide, LDL-C burden at last visit decreased by 97.9% (Additional file 2: Fig. 1, Panel B), while after addition of LA this parameter was reduced by 52.0% (Additional file 2: Fig. 1, Panel C). These results were confirmed after adjustment for untreated LDL-C values, genotype, ethnicity, gender (p adj = 0.045) and were not related to the ongoing background LLT ( Table 2).

Cardiovascular outcomes
As an exploratory analysis, we compared the occurrence of MACE during follow-up in HoFH patients receiving lomitapide or LA. In this analysis we included only patients that had been on treatment for at least 1 year. The absolute number of incident MACE in the Lomitapide cohort was lower than that in the LA cohort (13.3% vs. 55.2%; p < 0.001). When we estimated the MACE IR standardized for time of exposure, the IR in the Lomitapide was lower than that in the LA cohort (77.6 vs. 86.9 per 1000 person/years) ( Table 2). Results of the multivariate Cox regression analysis did not reveal significant difference in the risk of incident MACE (or its recurrence) in Lomitapide as compared with LA cohort.

Discussion
Our findings suggest that lomitapide in addition to statin/ezetimibe treatment might be more efficacious than LA in the long-term reduction of LDL-C. Indeed, HoFH patients receiving lomitapide presented larger LDL-C percent reduction in the last visit as compared to those on LA and the proportion of patients reaching a yearly on-treatment LDL-C < 70 mg/dl was higher in those treated with lomitapide than in those with LA. Same results were obtained when we analysed changes in the mean percent reduction of LDL-C during time (up to 9 years) showing that patients in the Lomitapide cohort achieved a greater LDL-C percent reduction from baseline independently from possible confounders. Finally, the reduction of cumulative LDL-C burden, which is considered a strong predictor of risk of vascular damage in FH [6,17], was larger in patient exposed to lomitapide than in those exposed to LA.
The LDL-C goals achieved in our LA cohort are consistent with those reported in previous studies where LA was used as the sole therapeutic intervention [6,7,18]. Recently published data from the UK Lipoprotein Apheresis Registry including both HeFH (N = 58) and HoFH (N = 30) have shown that the overall reduction in LDL-C interval means was 43.14% [19]. Furthermore, in a previous analysis by Bruckert et al. [6], authors found a reduction in TC by 20% in LA treated patients. On the other hand, the LDL-C lowering efficacy of lomitapide was almost superimposable to that already described in Phase 3 trial [13] and real-world study [14].
As an exploratory analysis, we noted that HoFH patients who were exposed to lomitapide experienced less cardiovascular events than those who had been treated with LA and this effect was particularly evident for CHD, PAD and aortic valve replacement events. Albeit this observation must be taken with great caution Fig. 2 Yearly changes of LDL-C in Lomitapide (green) and LA (light blue) cohort during follow-up. As the maximum duration of follow-up in the Lomitapide cohort was about 9 years, we compared the two cohorts in the same interval of time. Data are reported as mean values ± 2 standard errors per each time-point. A. HoFH subjects treated with lomitapide achieved significantly greater mean yearly LDL-C percent reduction from baseline as compared with those on LA. Data are reported as mean percent reduction ± 2 standard errors per each time-point. B. HoFH subjects treated with lomitapide achieved significantly lower mean yearly LDL-C values as compared with those on LA. P adj, value is adjusted for genotype, ethnicity, gender, untreated LDL-C and age at baseline. BL, baseline; LDL-C, low density lipoprotein cholesterol; LA, lipoprotein apheresis due to the nature of the comparison and the size of cohorts, it may suggest a benefit of lomitapide on cardiovascular risk. One might speculate that this might be related, at least to some extent, to the increased efficacy of lomitapide in reducing the burden of LDL-C. In fact, the role of cholesterol lifelong exposure on cardiovascular risk in HoFH has been widely explored. Previous data from the French HoFH cohort showed that the cumulative total cholesterol exposure was highly associated with the incidence of MACE [6]. Accordingly, Thompson et al. [7] insisted on the role of on-treatment LDL-C levels as the main determinant of overall HoFH survival. Also, Raal et al. [18] showed that the cardiovascular protection in HoFH was related to the on-treatment LDL-C levels, estimating that a LDL-C reduction of about 30% translated into almost 50% reduction of risk. However, further studies are needed to confirm this observation definitively establishing the role of lomitapide in reducing cardiovascular risk in HoFH. Safety data in the two cohorts have been described in detail elsewhere [14,20]. The most common adverse events (AE) associated with LA included hypotension, nausea, abdominal discomfort, tingling, chest pain, vasovagal reaction and prolonged bleeding from anticoagulant. Anemia was another common adverse event easily managed by iron supplementation. In addition, patients on ACE-inhibitors before the beginning of LA were systematically withdrawn from the treatment in order to prevent anaphylactic-like reactions [21,22]. Conversely, as expected the most common AE in the Lomitapide cohort was diarrhea that occurred in 41.2% of cases (data not shown). The proportion of patients referring gastrointestinal AE, however after 1 year of treatment, decreased to 16.6% after 1 year of treatment. One patient experienced liver function test elevation (> 5 ULN) after 3 months of treatment that was managed by holding lomitapide and statins and then making rechallenge (data not shown).
Some final considerations deserve the cost analysis of both LA and lomitapide treatments. Although this issue is difficult to be handled since the process of reimbursement is different in France and Italy, the direct costs of these therapies appear to be quite comparable between the two countries. Considering a weekly LA regimen treatment, the expected annual cost of LA would be about 90,000 euros while that of lomitapide is 180,000 euros. This economic evaluation does not take into account the social and personal costs of LA, as a result of the number of lost working days, not to mention the psychological burden. Further studies need to be done to better explore the pharmaco-economic aspect of these two therapies thus better allowing to compare the costbenefit of these two treatments.

Limitations
We acknowledge several weaknesses of the present study: the retrospective, cross-sectional observational design, and the heterogeneity of the two cohorts.
Patients from the LA cohort exhibited higher untreated and baseline LDL-C values: this observation might be explained by the higher variability in geographic origin in these patients, which can result in different socio-cultural habits having a different impact on the overall lipid profile. Nevertheless, it can be the reflection of an overall lower efficacy of baseline treatment in relation with a longer follow-up: highdose statins were not used in HoFH children until last decade (and age at first diagnosis was lower in the LA cohort), and ezetimibe was only introduced in 2005. Other treatments, as fibrates and cholestyramine, which were used in the LA cohort primarily because of patients' age, have a more limited effect on LDL-C than high-dose statins plus ezetimibe.
In addition, LDL-C response to treatments might have been also influenced both by the higher percentage of patients classified as carrying null/null mutations and by a more heterogeneous geographic origin within the LA cohort. However, it must be noted that in order to limit the confounding effect of these factors on the results, the efficacy of lomitapide versus LA has been evaluated adjusting for both the severity of genotype and the geographic origin.
It is important to note that we did not systematically collect safety data as this was not the aim of the present study that was primarily focused on describing the independent efficacy of these two treatments. More studies are needed to address this point in order to better clarify the cost-benefits of these two treatments.
We also have to acknowledge that due to the retrospective nature of the study, we did not retrieve information on adherence to concomitant lipid lowering therapies thus not allowing us to make any consideration on the possible effect of the adherence to background lipid lowering therapies on LDL-C control.
Another possible limitation of our results is the duration of follow-up that was markedly different in the two cohorts. This could be easily explained by the fact that LA treatment is available from 1990 and is authorized for use also in children whereas the availability of lomitapide has only been available for less than 10 years with no authorization to use in minors.
Finally, it is possible that the differences in follow-up duration as well as in baseline characteristics between the two cohorts might also have influenced the comparison of the effect of these two treatments on atherosclerotic burden making any observation on MACE only a conjecture. More studies are needed to clarify the benefit of lomitapide on MACE occurrence.
Considering all these limitations, the clinical implications of our analyses should be considered with much caution, especially because these findings are derived from an indirect comparison and might be biased by several confounding factors.

Conclusion
In summary, we observed in an indirect comparison that lomitapide appears to be more effective than LA in controlling atherogenic lipoproteins in HoFH. In major guidelines, LA in adjunct to statins and ezetimibe remains the first-choice therapy in the treatment of HoFH [1]. However, our results may open the perspective in which lomitapide, in addition to statins and ezetimibe could be considered as first-line treatment in HoFH. Further studies are necessary in order to formulate an updated protocol for the treatment of HoFH that integrates old and new treatments in a rationale and costeffective algorithm.

Study aim, setting and design
To assess LDL-C reduction, target achievement and long-term LDL-C burden in two HoFH cohorts, one treated with LA, one with Lomitapide, both on top of current available pharmacological treatment. Taking advantage of the early availability of lomitapide in Italy versus later availability in France, we designed a survey in which we retrospectively collected clinical and biochemical information of two independent cohorts of patients with HoFH treated with lomitapide in Italy or LA in France.

Patients' selection
HoFH patients known to be receiving lomitapide in Italy were considered for this survey. The inclusion criteria were: (1) molecularly or clinically defined HoFH, (2) age > 18 years, (3) treatment with lomitapide for at least 1 month. All identified patients (n = 34) agreed to participate. Of these, 8 patients were receiving LA plus lomitapide at baseline visit. Since in these patients LA therapy was stopped within 1 year from starting lomitapide and the lipid profile did not differ from that of those receiving lomitapide only, they were kept in the present analysis. Conversely, 4 patients with incomplete lipid profiles during follow-up were excluded. Thus, the Lomitapide cohort included 30 HoFH patients.
The French cohort consisted of all molecularly confirmed HoFH patients referred to the Pitié-Salpêtrière University Hospital who had undergone at least one session of LA. These patients were part of the French Registry of Familial Hypercholesterolemia (REFERCHOL). Details about this registry have been previously reported [28]. A total of 29 French HoFH patients were considered for the present analysis (LA cohort).

Data collection
For the Lomitapide cohort, physicians were asked to retrospectively revise medical records and extract demographic and clinical information. Details of concomitant lipid-lowering therapy, dosages of lomitapide and baseline plasma lipid values as well as those ones at last follow-up visit were also retrieved. No information on side effects, adherence to medications or diet was obtained.
For the LA cohort, the same data were extracted by two Authors (LD and AG) from REFERCHOL and harmonized for comparison with the Italian database.
Genotypes underlying HoFH in both Italian and French cohort were retrieved by medical records and ascertained as previously reported [6,14,23,24].
Baseline lipid values were defined as those at the date of initiation of lomitapide or LA treatment. Conversely, last follow-up data were defined as those at the time of the last clinic visit as of December 2019. The duration of follow-up was calculated as the difference between last and baseline visit.
Finally, data on MACE at baseline and during followup were collected. MACE was defined as a composite of angina, acute myocardial infarction, coronary, carotid or peripheral revascularization (as well as hemodynamic stenosis without revascularization) and ischemic stroke, aortic valve replacement and death for cardiovascular disease [25]. They were identified by either self-reported medical history and/or hospital admission documented in the medical record.

Laboratory measurements
In both cohorts, blood samples were collected early in the morning after overnight fasting. In HoFH patients undergoing LA, pre-and post-treatment samples were collected. Aliquots of plasma were used to determine total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C) and triglycerides (TG) following standard procedures. LDL-C values were calculated by using the Friedewald's formula. No complete information was available on ApoB, Lp(a), γGT, CPK and CRP, so that these data were not included in the present analysis.
The estimation of changes in plasma lipid during therapy was carried out using the following reference values: (1) untreated values, corresponding to the lipid profile at the worst LDL-C measurement available in medical charts while the patient was not receiving any treatment; (2) lowest lipid profile on conventional therapies, estimated as the lipid profile corresponding to the lower LDL-C value with "conventional treatment" (as statins ± ezetimibe ± fibrates ± resins ± porto-caval shunt ± PCSK9i) before the beginning of lomitapide/ LA; (3) baseline values, corresponding to the time of beginning lomitapide or LA; (4) last visit values, corresponding to the last visit when patients were receiving lomitapide or LA up to December 2019. It is worth mentioning that only two patients had weekly LA whereas the remaining were treated on a bi-weekly basis. The overall adherence to LA was on average > 80%, as previously reported [20].
For patients on LA pre-and post-apheresis measurements have been collected to estimate the TC and LDL-C interval means according to the following formula: The on-treatment cholesterol ( C follow−up ) was represented by the average of all LDL-C measurements obtained during follow-up in the Lomitapide cohort or by LDL-C interval means in the LA cohort.

Cholesterol burden estimation
Cholesterol burden was calculated according to the following formula: where CholesterolBurdenpre − treatment = (C baseline * * Age firstLAorLomitapideprescrption and CholesterolBurdenon −treatment = (C follow−up * Years follow−up ).
To estimate the average of TC and LDL-C, lipid profiles after baseline were collected every 3 months for the first 3 years, every 6 months for the period between 4 and 10 years and once per year from year 11 until last followup (these measurements were available only for patients on LA).

Statistical analysis
For descriptive statistics, continuous traits were presented as mean and standard deviation or as median and interquartile range as appropriate. Categorical traits were shown as number and proportion. Comparisons were carried out by Mann-Whitney for not-normally distributed and Student's t-test for normally distributed variables. For differences between categorical traits, P-value was calculated by chi-square. Paired T test was used to evaluate the difference between untreated, lowest and last visit total and LDL-C as well as LDL-C burden preversus on-treatment.
Linear regression with stepwise method was used to evaluate the association between the variables and adjustments were performed for the following variables: genotype (Null/null mutation versus other), ethnicity (European-non-Finnish vs. other), gender, untreated LDL-C and age at baseline. Not-normally distributed values were log-transformed before entering the model.
For the analysis on cardiovascular outcomes, we included only patients that had been exposed to lomitapide or LA for more than 1 year [27]. The number of MACE was counted in each cohort and their incidence rates (IRs) were expressed as number of events per 1,000 patient-year [25]. Cox proportional hazards model was TotalCholesterolBurden = Cholesterolburdenpre − treatment Cholesterolburdenon − treatment applied to investigate the predictors of incident MACE [25]. Statistical analyses were performed using the IBM Statistical Package for Social Sciences (IBM SPSS, version 25.0, Inc. Chicago, IL). A P-value < 0.05 was considered statistically significant. participated in the patient selection and recruitment, in the data collection and interpretation, revised critically and finally approved the manuscript to be submitted. All authors read and approved the final manuscript.

Funding
This research has been partially supported by a grant in-aid from Amryt Pharmaceutical to LD. Amryt was not involved in data analysis, in the preparation of the manuscript or in the decision to submit it for publication.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations
Ethics approval and consent to participate For the Italian cohort, informed consent was obtained from all patients before their inclusion into the study which was carried out in accordance with the ethical standards of local institutional committees for human experimentation in agreement with the Helsinki Declaration of 1964, as revised in 2018 (approval code #4928).The French cohort was declared to the ANSM (the French National Agency for Medicines safety) and received a declarant number [unique number identifying a particular research protocol, issued by the ANSM in France]: 2014-A01549-38. Two separate committees assessed the protocol of this study: French advisory committee on the processing of information for medical research (CCTIRS) and the National Commission for computer technology and freedom (CNIL) respectively in May and November 2015.Due to the retrospective nature of the present study, none of enrolled patients had received any procedures outside the standard clinical care.

Consent for publication
Not applicable.