SEARCH

SEARCH BY CITATION

Key points

  1. Top of page
  2. Key points
  3. Introduction
  4. Background: association between Lp(a) and cardiovascular risk
  5. Rationale and design of the ELAILa trial
  6. Planning of the trial, current status and randomization ethics
  7. Conclusions
  8. Disclaimer
  9. Addresss
  10. References
  •  Lipoprotein(a) has been associated with cardiovascular disease risk, but no randomized study has shown yet that lowering Lp(a) decreases cardiovascular risk.
  •  One treatment option for lowering Lp(a) in high-risk patients is lipoprotein apheresis, which has been approved for reimbursement now in Germany. This decision has been coupled with the mandate to perform a controlled endpoint trial.
  •  The ELAILa trial protocol has been filed (NCT01064934), but the study is on hold because of a negative ethics committee vote after strong opposition of part of the medical community due to ethical concerns.
  •  The authors argue that a randomized trial is necessary to investigate the effectiveness of the invasive, life-long and costly therapeutic procedure. Meanwhile, reimbursement continues.
  •  The study would be the first endpoint trial investigating whether lowering isolated Lp(a) elevation decreases cardiovascular endpoints.

Introduction

  1. Top of page
  2. Key points
  3. Introduction
  4. Background: association between Lp(a) and cardiovascular risk
  5. Rationale and design of the ELAILa trial
  6. Planning of the trial, current status and randomization ethics
  7. Conclusions
  8. Disclaimer
  9. Addresss
  10. References

The association between lipoprotein(a) [Lp(a)] and cardiovascular disease (CVD) risk has generated much interest in the last few years [1]. However, randomized controlled trials examining the effects of Lp(a) lowering on cardiovascular outcomes are lacking, and thus, an assumption of a beneficial effect remains speculative. In Germany, there was recently an opportunity to perform such a trial using lipoprotein apheresis to investigate whether decreasing Lp(a) improves cardiovascular outcomes in patients with isolated elevated Lp(a) concentrations and progressive cardiovascular disease. Lipoprotein apheresis for this indication is reimbursable in Germany.

The study is presently on hold due to the strong opposition of part of the medical community questioning the ethical justification of randomization. The aim of this study is to raise the discussion to an international level, to foster the debate across healthcare systems and to discuss the ethical considerations of not performing such a study.

Background: association between Lp(a) and cardiovascular risk

  1. Top of page
  2. Key points
  3. Introduction
  4. Background: association between Lp(a) and cardiovascular risk
  5. Rationale and design of the ELAILa trial
  6. Planning of the trial, current status and randomization ethics
  7. Conclusions
  8. Disclaimer
  9. Addresss
  10. References

Lp(a) is a low-density lipoprotein (LDL)-like particle consisting of an apolipoprotein-B100 molecule covalently linked to the large glycoprotein apolipoprotein(a) (apo(a)) [2]. The precise physiological role of these particles remains unclear. The distribution of plasma Lp(a) levels in the population is skewed, ranging from <0·1 to >300 mg/dL, and there is an interindividual variation by a factor of ∼1000 [3]. In Caucasians, the 80th percentile of Lp(a) concentrations is 50 mg/dL [1]. Numerous prospective epidemiological studies and meta-analyses [4–6] have reported associations between Lp(a) concentrations and CVD [7]. Two recent studies provided strong evidence for a causal association between increased levels of Lp(a) and coronary heart disease [8,9].

However, it remains unclear whether lowering Lp(a) concentrations reduces atherosclerosis and CVD outcomes. This situation recalls the situation at the end of the last century when many physicians refused to accept that lowering LDL was beneficial until the results of the 4S study [10] unequivocally established that it was. With Lp(a), a major problem in proving such an effect is the lack of an agent specifically and substantially decreasing Lp(a) levels [2]. The current drug treatment of choice for elevated Lp(a), niacin [1], decreases its concentrations by only 20–30% [11] and may reduce cardiovascular events [12], but at the same time affects other lipoprotein fractions so that its overall effects cannot be solely attributed, if at all, to the modest decreases in Lp(a) concentrations.

Lipoprotein apheresis, besides decreasing LDL cholesterol, is able to decrease Lp(a) concentrations by 50% to 70% [13]. In Germany, the Federal Joint Committee (Gemeinsamer Bundesausschuss; G-BA), the highest decision-making body of the so-called self-governing health system, decided in 2008 to approve the reimbursement of lipoprotein apheresis for individual patients with isolated hyperlipoproteinaemia(a) (Lp(a) >60 mg/dL) and progressive CVD when all other measures to stop disease progression have failed. This reimbursement decision was coupled with a mandate to perform a controlled trial which would definitely prove (or disprove) the benefit of Lp(a)-lowering using apheresis in these patients. An independent steering committee was created, and an investigator-initiated study protocol for such a study (ELAILa trial; http://www.clinicaltrials.gov identifier NCT01064934) was duly submitted to the competent ethics committee (EC). The study was rejected, however, after lengthy discussions with and appeals from various groups of medical professionals, mainly associated with the procedure. Their criticism was based on the argument that randomization for the study is not ethically justified. The EC eventually decided that it was indeed unethical to perform a randomized trial because an observational trial addressing this issue existed [14], and the data it provided deemed adequate proof of benefit of apheresis. However, this observational study did not include a control group.

The question remains whether there is enough evidence to support that Lp(a) apheresis improves the outcomes of patients with isolated elevation of Lp(a) and progressive CVD, thus obviating the need for a randomized trial. To us, the answer is a clear no, as uncontrolled observations do not fulfil the criteria of evidence-based medicine in 2012.

Rationale and design of the ELAILa trial

  1. Top of page
  2. Key points
  3. Introduction
  4. Background: association between Lp(a) and cardiovascular risk
  5. Rationale and design of the ELAILa trial
  6. Planning of the trial, current status and randomization ethics
  7. Conclusions
  8. Disclaimer
  9. Addresss
  10. References

The ELAILa trial was designed as a hybrid trial, consisting of both, a randomized controlled trial (RCT) and an observational trial (OT) for the patients not willing to be randomized. In addition, patients having reached a nonfatal endpoint in the RCT could be switched over to the OT for long-term follow-up (Fig. 1). The OT would thus also serve the purpose of establishing a national registry of Lp(a) apheresis patients, another mandate of G-BA (the respective documents can be found on http://www.g-ba.de).

image

Figure 1.  Flow of participants.

Download figure to PowerPoint

Primary outcome measure of the trial was a composite endpoint, defined as first occurrence of one of the following: myocardial infarction, interventional coronary therapeutic procedure, coronary artery bypass grafting (CABG), cerebrovascular accident, hospitalization due to acute coronary syndrome (ACS), peripheral arterial revascularization and death from cardiovascular cause. Secondary outcome measures were the components of the primary endpoint considered individually and death from any cause. Safety of the procedure and quality of life were also to be investigated.

The trial would enrol male and female patients ≥18 years of age having Lp(a) concentrations ≥60 mg/dL, LDL cholesterol <100 mg/dL and progressive CVD. As demanded by G-BA, CVD progression despite maximally tolerated conservative therapy would have to be determined not only clinically but also with imaging methods. Patients with LDL cholesterol of <100 mg/dL were selected for the trial since the effect of lowering Lp(a) and not of lowering LDL cholesterol on cardiovascular outcomes was to be investigated.

After randomization (1 : 1 ratio stratified by centre), patients would be treated by lipoprotein apheresis (all approved apheresis methods will be allowed in the trial). In the control arm, patients will be treated with maximally tolerated conservative treatment to reduce cardiovascular risk. Lp(a) concentrations would be measured in a central accredited laboratory with an isoform-insensitive assay, and changes in Lp(a) levels would be modelled as a covariate.

In preparation of a reliable estimate of the predicted number of events and for calculating the required sample size, statistical considerations were hampered by a lack of data on the parameters of interest. A 50% risk reduction was considered to be clinically relevant, a decision based on the effect sizes usually observed in statin trials (about one-third event reduction) and on the high-risk state of the study population. The assumptions for the final sample size calculation for the primary outcome included equal allocation of patients to the apheresis or control group, an accrual time of 48 months and an additional follow-up of 12 months, a median event-free time for the control group of 24 months and an event-free time twice as large for the apheresis group. It was furthermore assumed that patients will be accrued uniformly over time. Calculating the sample size, a relative risk reduction of 50% can be detected at a two-sided significance level of 0·05 and with a power of 80% by evaluating a total number of 135 patients.

The observational trial (OT) would also consist of two arms (see Fig. 1), an apheresis arm and a control arm (for patients unwilling to either be randomized or to receive apheresis treatment). If patients in the RCT would reach a nonfatal endpoint, they would be able to switch over to the OT arm.

Planning of the trial, current status and randomization ethics

  1. Top of page
  2. Key points
  3. Introduction
  4. Background: association between Lp(a) and cardiovascular risk
  5. Rationale and design of the ELAILa trial
  6. Planning of the trial, current status and randomization ethics
  7. Conclusions
  8. Disclaimer
  9. Addresss
  10. References

In December 2008, G-BA had agreed to the general outline of the ELAILa trial. Moreover, the German Cardiac Society and the German Society for Nephrology supported this trial. Financing would be supplied by four of the largest manufacturers of lipoprotein apheresis systems in Germany, which have established a consortium that would sponsor the organizational part of the trial. Importantly, in an additional decision of July 2009, G-BA ascertained that reimbursement of the treatment costs was granted for study participants by mandatory health insurance coverage. This decision has been endorsed by the Federal Ministry of Health. In January 2010, the finalized study protocol was submitted to the ethics committee (EC) at Charité University Medicine in Berlin. After several hearings and after having been contacted by groups of medical professionals who strongly opposed the study, the EC decided that it was crucial for their final decision to have an independent biostatistical expertise on a retrospective cohort study by Jaeger et al. [14], presumably showing a reduction of major coronary events with lipoprotein apheresis. Such a report was obtained, and a final decision was made in July 2010, ruling that the proposed RCT is unethical, while the OT received a positive vote. The EC came to the conclusion that it is unethical to randomize patients ‘as long as there is no active, probably effective control group and as long as there are no data from new cohort studies that question the results of the study of Jaeger et al.’. An appeal of the study trialists against the decision of the EC was rejected. The study has been ‘on hold’ since then.

We believe that this decision is very unfortunate as (i) it deprives both patients and the scientific community of the true answer to a significant question and (ii) the study on which it was based [14] has significant limitations, most importantly because it is retrospective and uncontrolled. This nonrandomized retrospective cohort study in 120 subjects showed that decreasing Lp(a) levels by 73% using apheresis associates with decreases in the annual rate of major adverse cardiovascular events (MACE) from 1·056 to 0·144 before and after starting apheresis [14]. At first sight, this finding seems very persuasive; however, potential limitations in design and statistical methodology should be considered. For example, in time-to-failure data (or person–time data in general), the basic assumption is that the risk of failure (event) in one group is the same constant multiple of the other group at any point in the follow-up time. The study violated the principle of independence of observations and used a questionable statistical test. Other problematic issues with this study are the fact that the patients were not maximally treated (e.g. only 8·3% were receiving niacin and only 3·3% cholestyramine), that there was no prospective documentation of events (events were retrospectively assigned as such) and that events were not adjudicated independently. Calculating intraindividual ‘event rates’ is a statistically dubious concept [15]. In summary, the conclusions of this study are limited by its design, and it does not meet the criteria of evidence-based medicine in 2012. We thus believe that, although laudable, this study cannot be considered as an adequate alternative to a randomized controlled trial as proof of the effectiveness of Lp(a) apheresis.

Moreover, a recent small randomized trial from Italy showed that in 21 patients with angiographically documented CHD apheresis decreased Lp(a) by 57% but showed no difference in the rate of cardiovascular events in a follow-up of 1 year [16].

While in their demand for a controlled trial, the G-BA did not specifically request a randomized trial, they characterize such as the ideal design. A nonrandomized trial design bears substantial selection bias (confounding by indication). In such a study, it cannot be excluded (not to say it would be rather likely) that patients with higher cardiovascular risk will be allocated to apheresis treatment while the ones with lower cardiovascular risk will receive standard care. In conclusion, another uncontrolled trial (or registry) will not be able to prove the effectiveness of apheresis over conservative treatment.

We acknowledge that the decision in Germany to reimburse this procedure, suggesting to both, medical professionals and patients, that the required evidence for superiority of apheresis exists, would make randomization and therefore recruiting difficult. On the other hand, it could be argued that it is difficult to defend offering patients an invasive, expensive, life-long procedure without having high-quality, evidence-based, proof of benefit. Even the Committee on Ethical and Scientific Issues in Studying the Safety of Approved Drugs of the Institute of Medicine [17] concludes that ‘… the FDA may be justified in requiring studies that could expose patients to heightened risk–but only if a public health question of pressing importance is at stake, if no other study design could supply the needed evidence,…’. We believe that this approach exactly reflects the situation with Lp(a) apheresis.

Of note, any cost-effectiveness data regarding Lp(a) apheresis for this indication are also lacking.

Conclusions

  1. Top of page
  2. Key points
  3. Introduction
  4. Background: association between Lp(a) and cardiovascular risk
  5. Rationale and design of the ELAILa trial
  6. Planning of the trial, current status and randomization ethics
  7. Conclusions
  8. Disclaimer
  9. Addresss
  10. References

Although observational data suggest that lipoprotein apheresis to decrease elevated Lp(a) concentrations may improve CVD risk, observational data cannot adjust for unmeasured confounders. However, they have been accepted in this particular case as adequate proof of beneficial effects of a life-long invasive procedure. How can this be explained? Certainly, the aspect of the ‘rare disease’ state of the patients was an argument involved in the decision. There is a clear need for guidance or guidelines evaluating the treatment of rare diseases to help health professionals navigate through the maze of various purportedly effective therapeutic options. On the other hand, one might ask why the scientific standards (and statistical methods) for investigating rare and common diseases should be different. Moreover, the decision may be a reflection of what has been described as ‘the illusion of validity’, the phenomenon of unwarranted confidence on a specific outcome which is produced by a good fit between the input information (significant decrease in Lp(a) concentrations) and the predicted outcome (decrease in cardiovascular events). Interestingly, this ‘illusion’, which has been observed even among the most experienced of researchers, persists even when the scientist is aware of the factors that limit the accuracy of his/her predictions [18].

The ethical basis for entering patients in randomized trials has been in general under debate. Some doctors espouse the uncertainty principle whereby randomization to treatment is acceptable when an individual doctor is genuinely unsure which treatment is best for a patient. Others believe that clinical equipoise, reflecting collective professional uncertainty over treatment, is the soundest ethical criterion [19].

Numerical modifications of surrogate markers of CVD risk do not automatically translate to an actual reduction of CVD events, as has been recently shown, for example, with CETP inhibitors, drugs that increase HDL-C and decrease LDL-C [20]. Observational and randomization data have often reached surprisingly disparate conclusions, and effects of surrogate markers were misleading in the field of cardiology, as we have seen, for example, with hormone replacement therapy and intensive glycaemic control in patients with diabetes. A randomized controlled trial to investigate the effects of lipoprotein apheresis to decrease Lp(a) needs to be performed sooner than later. Randomization is a powerful tool that cannot be reliably reproduced by statistical modelling, and therefore, observational data cannot be used as substitutes for randomized trial results [21]. The ELAILa trial may be an illustrative example where selectively missing information on a research subject does not only pertain to studies that have already been performed but for the much greater challenge to understand how many potential studies do not exist when they could readily have been conducted [22].

Disclaimer

  1. Top of page
  2. Key points
  3. Introduction
  4. Background: association between Lp(a) and cardiovascular risk
  5. Rationale and design of the ELAILa trial
  6. Planning of the trial, current status and randomization ethics
  7. Conclusions
  8. Disclaimer
  9. Addresss
  10. References

Drs. Heiner K. Berthold and Ioanna Gouni-Berthold are members of the ELAILa trial steering committee. The views expressed here are those of the authors and do not necessarily reflect the views of the other members. The authors have no conflict of interest associated with the subject matter.

Addresss

  1. Top of page
  2. Key points
  3. Introduction
  4. Background: association between Lp(a) and cardiovascular risk
  5. Rationale and design of the ELAILa trial
  6. Planning of the trial, current status and randomization ethics
  7. Conclusions
  8. Disclaimer
  9. Addresss
  10. References

Charité University Medicine Berlin, Virchow Clinic Campus, Lipid Clinic at the Interdisciplinary Metabolism Center, Berlin, Germany (H. K. Berthold); Evangelical Geriatrics Center Berlin (EGZB), Berlin, Germany (H. K. Berthold); Département de Médecine Interne et Centre de Recherche Médicale de Jolimont, Hôpital de Jolimont, Haine Saint-Paul, Belgium (O. S. Descamps); Center for Endocrinology, Diabetes, and Preventive Medicine, University of Cologne, Cologne, Germany (I. Gouni-Berthold).

References

  1. Top of page
  2. Key points
  3. Introduction
  4. Background: association between Lp(a) and cardiovascular risk
  5. Rationale and design of the ELAILa trial
  6. Planning of the trial, current status and randomization ethics
  7. Conclusions
  8. Disclaimer
  9. Addresss
  10. References