Addressing the health technology assessment of biosimilar pharmaceuticals

Abstract

The growing number of biosimilars presents challenges to regulatory and health technology assessment (HTA) systems. This paper illustrates these challenges by focusing on biosimilars used in the oncological setting. In particular, discordances between data required by regulatory and HTA authorities potentially deprive patients of effective treatments and hinder optimal resource allocation. Regulatory and HTA authorities need to harmonize requirements to foster the development and widespread use of biosimilars, which potentially release considerable resources. The authors believe that often-inappropriate methodology creates a very real chance that HTA authorities will reject some biosimilars. This would effectively extend patent protection and, in the absence of competitor pressure from biosimilars, result in prices remaining unnecessarily high. The authors propose that HTA organizations should accept pharmacokinetic and pharmacodynamic equivalence between the brand and the biosimilar as a proxy of biological comparability. HTA organizations should then adopt, in the absence of compelling reasons otherwise, cost-minimization analysis (CMA) as the basis of the cost-effectiveness deliberations. In the absence of adequate studies demonstrating equivalent efficacy, a prerequisite of CMA, HTA organizations should require threshold analysis. Once approved, biosimilar manufacturers and regulators should maintain rigorous pharmacovigilance to exclude immunoreactivity or other rare adverse events. Furthermore, cancer centres and trusts should regularly audit and publish the impact of biosimilars on clinical outcomes and resource use. When appropriate, regulatory and HTA authorities should demand revised cost-effectiveness analyses from biosimilar manufacturers. This approach would hone the accuracy of the cost-effectiveness analyses, protect patients and allow health services rapid access to low cost treatments.

Key words::

• Biosimilars
• Cancer treatment
• Cost and cost analysis
• Drugs generic
• Technology assessment biomedical
• Therapeutic equivalency

Introduction

Regulatory and reimbursement authorities worldwide increasingly employ Health Technology Assessment (HTA) to assess the social consequences of using a particular medical intervention. However, definitions of HTA vary between authorities1. In the UK, the National Institute for Health and Clinical Excellence (NICE) includes three key elements in HTAs: the broad balance of clinical benefits and costs offered by the technology; the degree of clinical need among patients with the condition or disease; and the potential for long-term benefits to the NHS that arise from the innovation2.

HTA helped develop more transparent and explicit methods of allocating limited healthcare resources. However, this paper argues that current HTA systems may discriminate against the growing number of biosimilars – pharmacokinetically equivalent versions of branded biopharmaceuticals. This potentially deprives patients of the opportunity to benefit from effective treatment and compromises optimal resource allocation. This paper considers some specific HTA requirements that may present challenges in biosimilar submissions. The authors highlight the need for a harmonized approach from HTA and regulatory authorities. However, healthcare systems cannot afford to wait for the result of these deliberations. Interim measures are proposed that could inform the development of a more equitable system as well as making biosimilars available more rapidly than at present.

The fourth hurdle

Governments worldwide need to control growing healthcare expenditure and ensure that each pound, dollar or euro invested optimizes health gain. As a result, HTA increasingly constitutes a ‘fourth hurdle’3 that medicines must overcome before reaching widespread clinical use. (Quality, safety and efficacy represent the three traditional ‘regulatory’ hurdles.) Indeed, HTA, which depends on demonstrating cost-effectiveness compared with current treatments, is arguably now the main factor determining whether clinicians can prescribe a new treatment. In the UK, for example, NICE rejected some treatments that prolong life in advanced malignancies with poor prognoses based on ‘unacceptable’ incremental cost-effectiveness ratios (ICER) despite compelling clinical trial evidence of efficacy4,5.

As this suggests, data required for marketing authorization often differs from that needed for HTA approval. For example, the European Medicines Agency (EMEA) and the US Food and Drug Administration (FDA) require at least one randomized controlled trial demonstrating efficacy against an appropriate comparator6,7. To avoid potential confounders and to limit risk to patients, the comparator in regulatory studies is often a placebo. In contrast, HTA bodies generally require pharmacoeconomic models comparing the treatment to current best practice, preferably based on a direct comparative study8–10. While this discordance initially caused difficulties, most phase III research programmes now generate data that satisfies regulatory and HTA requirements.

Once patent protection expires, other companies can produce pharmacokinetically and, pharmacodynamically equivalent versions of a drug or biopharmaceutical. These are cheaper than the original brand and, therefore, generate considerable savings, which health services invest in meeting other priorities. A classic ‘generic’ refers to a bioequivalent version of a small chemical, such as methotrexate or 6-mercaptopurine. A ‘biosimilar’ refers to pharmacologically and biologically equivalent version of a biopharmaceutical, such as erythropoietin or filgrastim. As discussed below, biosimilars are not synonymous with generics. In clinical terms, biosimilars are more similar to the pharmacoevolutional, incremental improvements made by second and subsequent members of a therapeutic class than to a conventional generic.

When considering a generic, the EMEA does not expect independent efficacy data in animals or humans11. Rather EMEA licences generics based on studies that demonstrate the formulation's bioequivalence to the branded ‘reference’ product. In general, the 90% confidence intervals for the area under the concentration–time curve (AUC) and maximum plasma concentration (C_max) for the generic should be within 0.80–1.25 of the reference. In some cases, the EMEA varies the definition of ‘bioequivalence’, such as tightening the range for drugs with narrow therapeutic indices12.

In most clinical situations, the pharmacokinetic variation inherent in the bioequivalence range does not influence patient outcomes. Unless prescribers specify a brand, pharmacists can substitute any approved generic. Therefore, while the prescriber is unlikely to know whether pharmacists dispensed the original brand or a generic, the switch is usually unlikely to influence outcomes and the health service can reinvest the resources released. Nevertheless, based on the EMEA requirements for bioequivalence, switching between a generic at the low end of the range (0.80 of the reference) to one at the high end of the range (1.25 of the reference) could result in AUC and C_max changing by up to 50%. In several clinical settings, this variation can undermine efficacy or exacerbate toxicity. As a result, healthcare professionals should not prescribe certain medicines – such as some antiepileptic drugs, certain modified-release formulations, and a number of controlled analgesics – generically13.

Biosimilars: a more complex situation

Unfortunately, the relatively straightforward model used to assess bioequivalence for generics does not apply to biosimilars. Since 2006, the EMEA has approved several biosimilars, including filgrastim, recombinant human erythropoietin (rHuEPO) and human growth hormone. In common with generics, the manufacturer intends the biosimilar to show the same pharmacokinetic and pharmacodynamic profile as the brand. However, unlike generics, the biosimilar manufacturer is unlikely to use the same production process, which complicates regulatory and HTA procedures. For example, the EMEA approved different precautions and warnings for a biosimilar growth hormone to the reference, a situation that probably arose because different cell lines were used to produce the brand and the biosimilar14.

To create a biopharmaceutical, living cells produce ‘raw’ protein that the manufacturer purifies and formulates. Despite the end of patent protection, many aspects of this process remain proprietary knowledge. Therefore, biosimilar manufacturers establish a unique manufacturing and purification process11,15 that, almost inevitably, shows subtle differences from the method used to produce the reference.

In most cases, these differences are clinically irrelevant. However, in theory at least, process variations could potentially influence the biopharmaceutical's characteristics, such as three-dimensional structure, acid-base variants and glycosylation. Furthermore, cells used to produce biopharmaceuticals usually release several isoforms12 rather than a single, distinct chemical. Theoretically, such differences could exacerbate immunogenicity or undermine tolerability and efficacy11,16. For example, removing an amide group from asparagine and glutamine, oxidizing methionine, and dimerization can reduce filgrastim's activity to 15% of that of endogenous granulocyte colony stimulating factor (G-CSF)17.

Because of these factors, the EMEA demands a higher level of evidence before granting marketing authorization to a biosimilar than that required for generics. For example, the biosimilar manufacturer must perform additional animal and human studies to assess pharmacokinetics and pharmacodynamics. The EMEA also recommends comparative clinical trials to demonstrate equivalent efficacy15.

Nevertheless, current analytical techniques and preclinical studies cannot detect or predict all the potential differences between biosimilars and the reference. For example, patients with chronic renal disease taking a formulation of rHuEPO containing polysorbate 80 and glycine rather than human serum albumin appeared to be at increased risk of developing pure red cell aplasia (PRCA)16. Cancer patients do not seem to develop rHuEPO-associated PRCA, possibly because of differences in immune competence, concurrent therapies and duration of exposure14. Because some rare conditions emerge after a large number of patients are exposed, biosimilars require intensive post-marketing scrutiny12,14, which is one reason why clinicians should prescribe biosimilars by brand.

Because of these differences, prescribers should regard a switch between the original product and the biosimilar (or vice versa) or between biosimilars ‘as a change in clinical management’12. However, although the EMEA recommends comparative clinical trials to demonstrate equivalent efficacy, these are not an absolute requirement15. Such discordance complicates clinicians’ ability to evaluate this effective change in management.

In the case of filgrastim, the EMEA has licensed two biosimilars: Ratiograstim (ratiopharm direct)18 and Filgrastim (Hexal)19. The first was authorized on the basis of both pharmacological and comparative efficacy studies versus the branded comparator (Neupogen) while the second was granted on the basis of pharmacological equivalence, with only supportive evidence from a non-comparative clinical efficacy study. Both agents are licensed for the full range of filgrastim indications, on the basis that they had demonstrated equivalence to the comparator and shared the same mode of action. For a clinician, this presents problems – a decision must be made as to whether the pharmacological evidence is sufficient to justify a change of product or whether, as seems more prudent to prescribe the biosimilar supported by the most rigorous and comprehensive evidence base. The situation, however is considerably complicated when considering HTA requirements.

HTA assessment and biosimilars

While several aspects of regulatory procedure need harmonization and clarification, HTA assessment of biosimilars is even more unclear and policies are constantly evolving20. Moreover, there is a fundamental discordance in the attitudes of regulatory bodies and HTA authorities towards biosimilars. Regulatory bodies regard biosimilars as sufficiently similar to the reference product to require an abbreviated submission. In contrast, HTA authorities regard biosimilars as sufficiently different from the reference to require a full submission, incorporating economic analysis. This discordance could lead to delays in approval and even the rejection of a biosimilar by an HTA body, which, in turn, potentially deprives patients of an effective treatment and hinders optimal resource allocation.

Reviewing requirements for various HTA bodies demonstrates a requirement for a common dataset (Table 1)8–10 that potentially creates an insurmountable hurdle for some biosimilars. The authors believe that an emphasis on an often-inappropriate health economic method creates a very real chance that HTA authorities will reject some biosimilars. This would effectively extend patent protection and, in the absence of competitor pressure from biosimilars, result in prices remaining unnecessarily high.

Table 1.  The common dataset for HTA submissions.

Item	Example
Basic efficacy and safety data	Comparable to that required for licensing
Clinical effectiveness data	Comparison between new treatment and existing options
Cost-effectiveness	Typically, cost-utility analysis versus reference treatment
Budget impact	Integrating cost information with burden of disease

Typically, HTA authorities require a cost-utility analysis (CUA) to estimate ICER. A CUA compares interventions that produce different consequences, such as the quantity and quality of life (Table 1)21. A utility measure, usually a quality adjusted life year (QALY), combines duration of benefit and quality of life into a single variable. Most HTA authorities establish a ‘threshold’, usually defined as a cost per QALY, above which a product is unlikely to be approved. Moving resources from interventions associated with a high to low cost per QALY maximizes the health gain from each pound, dollar or euro invested. Importantly, CUA uses a common framework and outcome allows policy makers to compare dissimilar treatments for the same medical condition and, more controversially, between diseases22–25. While controversial and subject to basis, QALYs allow healthcare purchasers to ‘objectify’ budgetary decisions. This, combined with their long history of use, means that cost per QALY is likely to remain the pre-eminent ICER method in HTA.

Currently, NICE's upper limit of the ‘range of acceptable cost effectiveness’ is around £30 000 per QALY gained. Above this, factors other than ICER must suggest that the technology is an effective use of resources. For example, NICE supports technologies with ICER above £30 000 per QALY gained for small populations of patients suffering from incurable illnesses, provided their life expectancy is, normally, less than 24 months and the intervention extends life, normally, by at least 3 months, compared to current NHS treatment26.

As discussed above, the EMEA licences biosimilars on the assumption that they produce equivalent utilities, but the newer drug will have a lower acquisition cost (roughly equivalent to price). Therefore, CUA are inappropriate for a cost-effectiveness analysis of biosimilars. By definition, CUAs compare interventions that produce different utilities21. As biosimilars have no effect on QALYs, relative to the comparator, the ICER will be infinite, regardless of price differences. Where a CUA could be appropriate is if clinical experience shows a difference in rare adverse events or resource utilization between the originator and the biosimilar, or between biosimilars. However, such cases are likely to exceptional and are likely to emerge, if at all, only when a large number of patients have received the biosimilar.

In contrast, cost-minimization analyses (CMA) assume that the consequences of the interventions are the same and that only costs differ. In other words, CMAs determine the least costly way of achieving a certain outcome21,27. Justifying CMA requires a high level of proof based on three lines of evidence. Firstly, an a priori expectation that the treatments should perform equally and, secondly, pharmacodynamic and pharmacokinetic evidence that treatments are clinically equivalent. As discussed above, the biosimilar manufacturer needs to demonstrate these criteria to obtain marketing authorization.

Thirdly, CMA requires evidence from adequately designed and powered equivalence or non-inferiority studies. Licensing bodies do not necessarily require a comparison of clinical outcomes in equivalence or non-inferiority studies. While such data is a prerequisite for HTA8,10,28, such studies are available for only some biosimilars. HTA authorities in Scotland29 and Wales30 approved Ratiograstim based on a CMA derived from a direct comparative study with the reference product in one of the licensed indications31. However, given that other versions of filgrastim lack such comparative studies, it is open to question whether they could comply with the third criterion for a CMA, despite being EMEA approved.

Assessing relative effectiveness

From the discussion above, it is clear that CUAs assume dissimilar relative effectiveness, while CMAs assume equivalent effectiveness. However, these assumptions raise the vexed issue of defining relative effectiveness. Few, if any, identical placebo-controlled studies compare a biosimilar and reference. Inevitably, therefore, HTAs will need to accept, for example, secondary rather than primary outcomes, slightly different treatment regimens or different inclusion criteria, despite the resulting uncertainty. Therefore, any differences should seem clinically reasonable and comprehensive sensitivity analyses must support any conclusions drawn. It is worth considering a detailed example illustrate this principle.

Eldar-Lissai et al. published a CUA comparing a single dose of prophylactic pegfilgrastim to either 8 days of filgrastim or no treatment in patients undergoing chemotherapy32. The analysis used a single cycle of chemotherapy in patients with solid tumours and a baseline risk of febrile neutropenia of 20%. Pegfilgrastim was associated with lower cost and improved outcomes compared with filgrastim.

However, different assumptions of relative efficacy could markedly influence the conclusions. Eldar-Lissai et al. assumed that prophylactic filgrastim and pegfilgrastim reduce risk of febrile neutropenia by 39% and 90%, respectively. Yet no study directly compared efficacy. Therefore, the authors used two placebo-controlled studies that employed different chemotherapy regimens and, thus, showed varying baseline risks of febrile neutropenia.

Firstly, in a randomized placebo controlled trial performed by Vogel et al. that enrolled breast cancer patients33, 17% of the placebo group developed febrile neutropenia, compared with 1% of pegfilgrastim-treated subjects, equivalent to a relative risk reduction (RRR) of 90% (95% CI: 72–96%). Secondly, a meta-analysis of nine placebo-controlled filgrastim studies in patients with various solid tumours or lymphoma34 reported that the risk of febrile neutropenia in control patients ranged from 21% to 77%, compared with 5–40% in filgrastim treated patients. The pooled RRR was 39% (95% CI: 28–48%).

Eldar-Lissai et al. argue that sensitivity analyses show that pegfilgrastim usually remains dominant despite varying effectiveness across ‘plausible clinical values’ or assuming equivalent RRR32. Nevertheless, as there are no other published placebo-controlled trials of pegfilgrastim, it is unclear whether the Vogel study represents the expected efficacy in different cohorts or settings. Furthermore, the meta-analysis included nine studies with a wide range of baseline risk. Subanalysis reveals patients at higher risk of febrile neutropenia showed a lesser RRR than those at lower risk. Consequently, the pooled result for all risk levels will be unrepresentative of a low-risk group, such as that examined by Vogel.

Only one filgrastim study enrolled patients with a comparable level of risk35. In this trial, 21% of patients treated with placebo had febrile neutropenia, compared with 5% of filgrastim-treated patients, yielding a relative risk reduction of 75% (95% CI: 28–91%). This falls within the 95% confidence intervals for pegfilgrastim, suggesting no significant difference between the two agents. In other words, several outstanding issues concerning the validity of the outcomes used and the poor comparability of the evidence base means that there is no rigorous evidence that pegfilgrastim is more cost effective than filgrastim. While pegfilgrastim is not strictly a biosimilar, this example illustrates the complexity of assessing relative efficacy.

Cost: further complication

Current HTA mechanisms usually assess therapies available at fixed reimbursement prices. Indeed, the three HTA organizations in the UK state explicitly that cost-effectiveness analyses must compare only list prices8,10,28. For most drugs prescribed in primary care, this approach seems reasonable. Firstly, for example, the NHS reimburses patent-protected drugs dispensed in primary care at a published price. Secondly, economists can employ a variety of well-established sources to determine consequential costs (e.g., hospital admissions, GP consultations, social care costs) and address any uncertainties in the relevance to the clinical setting using sensitivity analyses. Thirdly, dispensing pharmacists and consortia of primary care trusts can negotiate discounts with manufacturer, although these tend to be exceptions rather than rules. NHS budgets include these as an average correction figure. Finally, biosimilars have yet to reach primary care, although this may change with the advent of biosimilar insulins.

This approach is less well suited to biosimilars used in secondary care. Hospital pharmacy services actively seek and achieve discounts in most clinical sectors. The departmental budget retains any ‘profit’ over the reimbursement price. Furthermore, price competition in the UK generics market is intense – and the biosimilars market is likely to follow suit. The competition is particularly relevant for biopharmaceuticals, which generally carry a much higher cost than conventional pharmaceuticals. Indeed, discounts on list price of up to 70% are common [personal communication; ratiopharm direct]. Given the market's competitive nature, it is likely that calculated cost differences between treatments will vary substantially from those in practice. Thus, cost-effectiveness analyses of biosimilars performed according to current HTA criteria are likely to be out of date or irrelevant to many secondary care trusts.

Proposal

Clearly, regulatory and HTA authorities need to address particular discordances and harmonize requirements to foster the development of biosimilars, which potentially release considerable resources. However, the NHS cannot afford to wait for these deliberations. Given the constant pressure on healthcare budgets, there is a need to develop interim mechanisms that allow HTA to assess biosimilars demonstrating comparable pharmacological and biological activity to reference products. Insights and experiences gained using this interim model would inform the definitive mechanism.

The authors propose that HTA organizations should follow the lead of the EMEA and accept pharmacokinetic and pharmacodynamic equivalence as a proxy of biosimilarity. Therefore, in the absence of compelling reasons otherwise in particular cases, HTA bodies should accept CMA as the basis of the cost-effectiveness deliberations. Where direct comparative evidence exists, the approach would be straightforward, along the lines adopted for Ratiograstim29,30. Any cost differences arising from discounts or competition would, of course, further improve the ICER.

In the absence of comparative studies, HTA bodies should accept efficacy, utility and costs derived from studies of the reference treatment, ideally based on a meta-analysis, and use threshold analysis to explore relative efficacy. Threshold analysis constructs a CUA model using whatever data are available to support equivalence. The baseline case assumes the two treatments produce equal QALYs. The analysis then determines difference in efficacy needed to exceed a willingness to pay threshold – such as the lower bound of NICE's range of acceptable cost-effectiveness (£20 000 per QALY).

The HTA authority should define an acceptable threshold for treatment effect. For example, if a 30% difference in treatment effect crosses the threshold, the HTA body could reasonably approve the biosimilar. If a 5% difference crossed the threshold, the HTA body might reasonably reject the biosimilar. Sensitivity analyses should assess the impact of cost variation arising from competition or discounts.

These models need to remain dynamic. Biosimilar manufacturers and regulators should maintain rigorous pharmacovigilance for immunoreactivity or other rare adverse events. In the unlikely event these emerge, a CUA could replace a CMA. Furthermore, cancer centres and trusts should regularly audit and publish the impact of biosimilars on clinical outcomes and resource use. As evidence accumulates, the CMA and threshold analysis become increasingly accurate.

Despite being available for a relatively short time, audits of biosimilar filgrastim are beginning to emerge. For example, in November 2008, Southampton University Hospitals NHS Trust (SUHT) became the first UK transplant centre to switch to a biosimilar filgrastim exclusively for peripheral blood stem cell (PBSC) mobilizations before autologous stem cell transplantation. SUHT initially audited 64 PBSC harvests, all of which were successful. Mean CD34⁺ cell counts were 94.3 and 76.6 with filgrastim and Ratiograstim, respectively. Median CD34⁺ cell counts were 33.0 and 39.3, respectively. The mean CD34⁺ collections were 6.4  × 10⁹/kg in both groups36.

SUHT established a 10% efficiency target for PBSC mobilization. Mean efficiency was 9.1% and 8.6% with filgrastim and Ratiograstim, respectively; median efficiency was 8.2% and 8.1%, respectively. No difference emerged in time to maximal PBSC and no unexpected toxicity emerged. This audit suggests that switching from filgrastim to Ratiograstim did not compromise patient outcomes or toxicity profiles in a clinical setting.

Conclusions

Biosimilars offer the prospect of using biopharmaceuticals that are as effective as the brand, but produce marked savings. Biosimilar filgrastim is the first to reach the oncological armamentarium, but many others will follow. The consensus of opinion, an audit36 and clinical studies37,38 suggest that the reduction in the incidence of neutropenic sepsis when used as primary prophylaxis and the rapid neutrophil response seen following neutropenic sepsis produced by biosimilar filgrastim is equivalent to that seen with the original product. However, clinicians should regard any switch from brand to biosimilar ‘as a change in clinical management’12 and use the biosimilar supported by the most comprehensive, rigorous evidence base.

The potential differences between brand and biosimilar suggest that clinicians should prescribe biopharmaceuticals by brand, which will also facilitate pharmacovigilance. Theoretical concerns that the immunogenicity profile may differ and loss of efficacy might emerge after multiple courses has not, to date, been realized with biosimilar filgrastim. However, rigorous pharmacovigilance must continue.

The potential for differences between biosimilars behoves regulators to require manufacturers to produce a consistent body of evidence demonstrating clinical effectiveness. Moreover, there is a fundamental discordance in the attitudes of regulatory bodies and HTA authorities towards biosimilars. Regulatory bodies regard biosimilars as sufficiently similar to the reference to require an abbreviated submission. In contrast, HTA authorities regard biosimilars as sufficiently different to require a full submission.

Conflicts between licensing and HTA requirements raise the prospect that HTA authorities will not endorse some biosimiliars, leaving only the higher cost patented product available. Whilst there is, of course, no guarantee that a biosimilar with a low list price will remain the cheapest after discounting, their absence will reduce the incentive for discounting and prices will remain high. Given the constant pressure on healthcare budgets, there is a need to develop mechanisms allowing HTAs to assess biosimilars in a fair manner, to ensure that competition for patent-expired products drives prices down.

Regulatory and HTA authorities need to harmonize requirements internally and between the bodies. In the meantime, the authors propose that HTA organizations should follow the lead of the EMEA and accept pharmacokinetic and pharmacodynamic equivalence between the brand and the biosimilar as a proxy of bioequivalence. HTA organizations should then adopt, in the absence of compelling reasons otherwise, CMA as the basis of the cost-effectiveness deliberations. In the absence of adequate demonstration of relative efficacy, a prerequisite of CMA, HTA organizations should require threshold analysis. Biosimilar manufacturers and regulators should maintain rigorous pharmacovigilance to exclude the risk that rare adverse events may emerge. Furthermore, cancer centres and trusts should regularly audit the impact of biosimilars on clinical outcomes and resource use. When appropriate, HTA authorities should mandate biosimilar manufacturers to update their cost-effectiveness analysis based on these data. This approach offers a pragmatic means to allow health services access to low cost treatments as rapidly as possible.

Transparency

Declaration of funding

This review was funded by an unconditional grant from ratiopharm direct. The funder was invited to comment on drafts of the paper but all editorial decisions rested with the authors, who take full responsibility for the content.

Declaration of financial/other relationships

J.B. has disclosed that he has carried out paid consultancy for a wide range of pharmaceutical companies. He received payment from ratiopharm direct relating to this paper and also for contributions to UK health technology assessment submissions for their products. A.S. has disclosed that she has received payment from ratiopharm direct for contributing additional materials to this paper and commenting on drafts and also for participation in an advisory board meeting. P.A. has disclosed that he has received payment from ratiopharm direct for contributing additional materials to this paper and commenting on drafts.

Some peer reviewers receive honoraria from CMRO for their review work. The peer reviewers of this paper have disclosed that they have no relevant financial relationships.

Acknowledgement

Medical writer, Mark Greener, helped with revisions to the initial draft of the manuscript, for which he received payment from ratiopharm direct.

The paper was conceived and written entirely by the authors. J.B. conceived the project, carried out literature searches and prepared the initial draft of the paper. A.S. and P.A. commented on the initial draft and contributed additional materials.