The problems and pitfalls in systemic lupus erythematosus drug discovery

KEYWORDS:

• Systemic lupus erythematosus
• clinical trial
• epidemiologic methods
• pragmatic trial
• drug approval

1. Introduction

Treatment of systemic lupus erythematosus (SLE) was historically based on clinical experience and the evidence from observational studies, suggesting the clinical efficacy of immunosuppressors.[1] However, new drug development in the era of evidence-based medicine requires successful evaluation through different research phases. International regulations in order to have a drug approved require two phase III controlled trials, preferably superiority trials, and with open-label extension.

In 2004, the US Food and Drug Administration (FDA) published a draft guidance document on the development of drugs for SLE. This document was revised and updated in 2010 by incorporating constructive input from different experts.

Over the past decade, development of new drugs for SLE prompted the design of several trials. However, even drugs largely used as off-label therapy for refractory SLE manifestations faced multiple problems in clinical trials. Those trials testing the efficacy of abatacept and rituximab were probably the most representative examples of the lack of success in SLE trials. Fortunately, a new drug – belimumab – was approved for SLE in March 2011, restoring the confidence of research community.[2]

Many lessons can be learned from the several unsuccessful trials in order to avoid past mistakes in the future. Some of the challenges in SLE trial design include determination of adequate size, patient population selection, end-point definition or trial duration, among others.

2. Trial size

Several examples illustrate the difficulty in defining a correct sample size and the importance of finding the correct balance between the power of the trial and the economical efforts required to carry it out. For instance, more than 800 patients were included in the belimumab BLISS-76 successful trial, but it would probably had not found statistically significant differences in efficacy between belimumab and placebo arms with a smaller number of subjects.[3] Conversely, a numerical difference was found between placebo and the treatment group in the lupus nephritis assessment with rituximab (LUNAR) trial including only 144 patients, although it did not reach statistical significance.[4] Thus, before determining the phase III trials size, a careful attention needs to be taken to determine the expected clinical efficacy of the experimental drug based on phase II trials or in other previous observational studies.

3. Patient population selection

One of the most important points to consider when designing a trial is to define the best suitable population for the study. Several consequences arise from the trial eligibility definition, taking into consideration that too restrictive inclusion criteria will sometimes constrain the external validity of the trial. In this sense, post hoc analysis of a subset of patients with nephrotic range proteinuria from the formally failed trial of abatacept in lupus nephritis showed that treatment with abatacept resulted in an approximately 20–30% greater reduction in the mean urinary protein/creatinine ratio in this specific subset.[5] Conversely, the success in both BLISS trials that led to belimumab approval was based on the detailed analysis of the data from a nominally unsuccessful phase II trial, which showed that only a subset of patients benefited from the treatment, thus restricting pivotal phase III trials to them.[3,6,7]

Although some authors have yearned for a universal biomarker to enrich clinical trials with subjects prone to response, a pragmatic approach is encouraged in order to bring trial population closer to the patient population of the real clinical practice. Yuen et al. [8] assessed the eligibility of a cohort of SLE patients for any of 64 clinical trials and found only 40% of patients to be suitable for any trial. It is not surprising to see how promising drugs in clinical trials result not to be that good in clinical practice because patients out of the trials might be particularly prone to develop adverse effects or to experience different outcomes when following therapeutic strategies learned from the clinical trials.

4. End-point definition

Another great challenge when designing trials in SLE is to capture its waxing and waning nature with multiorgan system involvement in a single variable. According to FDA guidelines, a global disease activity index should be used and the British Isles Lupus Assessment Group (BILAG)-2004 is the preferred index, although others (i.e. SLE Disease Activity Index (SLEDAI), European Community Lupus Activity Measurements, Systemic Lupus Activity Measure or their updated versions) are accepted. Simultaneously, patients-reported outcome measures are encouraged. Surprisingly, the activity index used in the pivotal phase III BLISS trials that led to belimumab approval by the FDA was not in their guidelines document. Instead, a new activity index called SLE Responder Index was created as a composite index from SLEDAI, Physician Global Assessment, and the BILAG-2004.

With the aim of obtaining a clear and simple estimation of drug effect, composite disease activity indices are often used in SLE trials. They preclude from competitive effects avoiding the need of multivariable adjustment while, at the same time, decrease sample size simply by increasing the probability to reach the outcome. However, the use of a composite index implies the homogeneity of its variables. But, has a decrease from BILAG-A to BILAG-B in any domain the same meaning of 4 points decrease in the SLEDAI? Does it mean the same for the patient and for the physician? Actually, the dissimilar incidences of clinical manifestations included in disease activity scores influence unequally the outcome.[9,10] Furthermore, since SLE is a multiorgan disease, probably different pathophysiologic pathways lead to each clinical manifestation and, as expected, the experimental drug might have different effect on each domain. Thus, sometimes it becomes hard to understand the real effect of a given drug based on score results. Trying to avoid any misunderstanding, clinical trials often show a table where each component of the score is compared against placebo. However, a reduction on the alpha risk is rarely recognized.

Although some efforts have been performed to define a good response score in lupus nephritis,[11] the traditional outcome in these trials has been based on changes in the proteinuria. However, the degree of improvement required to meet a response definition is still a subject of debate. For instance, a particularly restrictive definition of renal response in the mycophenolate mofetil Aspreva Lupus Management Study trial resulted in trial failure while post hoc evaluation using a more permissive outcome definition showed clinical and statistical significant difference.[12] In this sense, Dall’Era et al. [13] concluded from the analysis of the Euro-lupus regimen trial data that proteinuria less than 0.8 g/day at 12 months after randomization was the single best predictor of good long-term renal function. However, proteinuria might be a clinical expression of kidney inflammation or chronic kidney damage, and it is not clear if proteinuria is an adequate end-point for drugs that can reduce proteinuria by directly affecting the podocyte.[14]

5. Trial duration

FDA recommendations state that randomized trials should last at least 1 year. However, trial duration should not be based on a strict rule but rather on the time when the effect is expected to take place. For instance, a reduction in proteinuria was found to take place after 78 weeks in patients included in the rituximab LUNAR trial while the trial was not able to find any difference when it was completed by the 52nd week.[15]

6. Other challenges

It is also a concern that concomitant medications (for instance, corticosteroids) limit the effect of the drug, or that in the large phase III trials, a number of sites were used that were not experienced in SLE or in SLE disease activity assessments. Furthermore, the heterogeneity of the disease and its waxing and waning course can also be confounder factors.

7. Conclusions

Several aspects on SLE trials design still remain a challenge. Trial size should be carefully evaluated balancing the economic effort and its required power. Although evidence-based medicine sacralized clinical trials for its extraordinary internal validity, nowadays clinicians and investigators have understood the importance of their external validity. A careful definition of inclusion criteria is warranted in order to succeed in the trial. Simultaneously, appropriate outcomes should be selected according to drug effects. Even when randomized controlled trials continue to be essential to prove drug efficacy and have them approved by regulatory agencies, pragmatic observational studies need to be conducted to understand drug usefulness in real life.

8. Expert opinion

After many difficult years, a promising future for clinical trials in SLE is expected. Belimumab approval gave a breath of hope to SLE research community and encouraged several companies to embark on new drug development for SLE. However, a diligent review of past mistakes is warranted in order to learn and avoid them in the future.

We know that SLE is a heterogeneous disease with a waxing and waning course; therefore, future phase III trials should try to homogenize the patient sample while phase IV studies should be encouraged to understand the real usefulness of new approved drugs in clinical practice. In this sense, better understanding of the pathological pathways of SLE might lead to further categorize patients based on their dysregulated physiopathological pathway and this might favor the selection of more homogeneous groups in the clinical trials. New knowledge might come up from current ongoing studies where several biomarkers are being explored to build up new classification criteria for autoimmune diseases. Parallel sub-studies might allow researchers to split SLE clinical manifestations into smaller groups according to their dysregulated pathway.

Furthermore, new trials might consider to include patients without corticosteroid treatment. Different response to treatment adds disparity between patients, thus requiring trial size to be bigger in order to find differences between groups. Probably, new trials will not need such a big size as required in previous trials and this will encourage more drug companies to embark on other trials.

Concurrently, careful end-point selection needs to take into consideration their benefits and drawbacks. Probably, more simple end-points might lead to an easier understanding of drug effects. For instance, proteinuria less than 0.8 g/day at 12 months after randomization could be a simple end-point in lupus nephritis trials, and it should be tested. In addition, terminating a clinical trial as soon as a robust result becomes evident should be an ethical and practical imperative, and group sequential clinical trial designs are a modern means to this end, which should also be implemented in the design for SLE studies.[16]

Declaration of interests

R Cervera has received sponsorship from GlaxoSmithKline. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.