Author:

Nicolò Morina

Abstract

Clinical trials can be intrinsically slow and high-risk, negatively affecting research for new treatments in challenging areas like oncology where there is a need to make faster progress. These properties are mainly due to methodological aspects. For decades, the conventional procedure of clinical trials has compared a new therapy to the standard of care for a specific disease in a two-arm setting. This methodology ensures the integrity of results, at the cost of minimal flexibility throughout the study. A solution to this limitation has been proposed in the early 2000s in the UK: the multi-arm multi-stage platform trial design. By changing the underlying questions, this approach has overcome the barriers created by a scheme that has always been trusted, but that definitely needs to be improved.

Introduction

Among biomedical science fields, there’s one which stands out for being both completely different from the others and their natural prosecution at the same time: clinical trials, essential to promote the development and evaluation of treatments for known diseases, from the most common to the most complex and rare. Clinical trials are a unique area of medicine due to the fact their results can have an almost immediate impact on society. Indeed, when the chance of improving health conditions of patients becomes a real possibility thanks to a new drug, the need to assess its efficacy and safety in the best and fastest way possible is fundamental. Thus, the science of clinical trials, coupling the selection of proper hypotheses with statistical rigour, has greatly influenced the medical practice during the 20^th century. While the advantages it ensures are unquestionable, it is also true that what have been the standard of care in trials’ design for decades has its limitations. This story witnesses a revolution that took place at the beginning of the 21^stcentury, aimed at overcoming these very limitations: the Multi-Arm Multi-Stage (MAMS) platform design.

The origin of the standard of clinical trials

First, a bit of context: the history of clinical research owes a great deal of its reliability to essential trials carried out in the 1940s by the Medical Research Council (MRC) in the UK, paving the way for a methodology applied to this day. In 1943 and 1944, a trial was performed to try out the efficacy of patulin, a new compound derived from Penicillium patulinum, for the treatment of common cold. It had been proposed as a potential alternative to penicillin, but it proved ineffective in what has been the first double-blind trial performed on over a thousand enrolled people. In fact, both the doctors and the patients were unaware of the administered treatment, whether it was patulin or a placebo. A few years later, between 1947 and 1948, streptomycin was tested on individuals suffering from pulmonary tuberculosis. The novelty this trial introduced was the randomization of patients among treatment groups, although the patulin trial exploited an alternate allocation method, which can’t be considered completely random. Concurrently, the streptomycin test had no double-blind features and lacked a placebo administration (control patients were simply untreated) (D’Arcy Hart, 1999). Thus, both studies showed key traits of the future answer to the need of a robust, trustable design. From these complementary experiences, the double-blind randomized controlled trial (RCT) emerged and imposed itself as the gold standard for clinical research in the many years to come.

The specific features of this design allow to reach solid results while responding to the main purpose behind each clinical trial: the willingness to test a hypothesis by evaluating a new intervention, compared with a control counterpart. This is what drives a phase-III trial, the common final step before the license of a new drug, after phases I and II, aimed at assess its safety and activity, respectively, on smaller cohorts of patients (Sedgwick, 2011). Following this concept, it appears logical that once the conditions are set, it is recommended not to deviate from the original plans, and that for a simple reason: it is the easiest way to retain the integrity. Since clinical trials are the last passage before a new treatment is released in society, it is pivotal they have the utmost integrity, so to be sure of, and trust, the obtained results. Thereby, the comparison of one new treatment versus one control treatment has been the trait of the standard approach to evaluation for more than fifty years. Nevertheless, over time one thing has become clear to the medical community: this approach is slow. Extremely slow, some would say. This slowness couples with the substantial costs of a trial, which can sometimes reach several million euros at final stages. The fact that such an expensive process, lasting years, involving many people and requiring massive efforts, might end in failure of one new treatment, can be very frustrating.

Looking back to the turn of the century, there were many people suffering from these limitations and claiming a substantial change: one of these people was Prof. Mahesh Parmar, now Director of the MRC Clinical Trials Unit and the Institute of Clinical Trials and Methodology, and Professor of Medical Statistics and Epidemiology at the University College of London (UCL). As a statistician for the MRC since 1987, Parmar participated to the ongoing trials of the time: “In the late 1990s, after seeing that many new treatments weren’t better than the old ones, with a lot of phase-III trials inexorably failing, I started thinking that things were too slow”. The main weakness was the comparison of two treatments only, with no chance of modifications: a very high-risk bet. Parmar is clear about this: “There is nothing magical about number two”. Thus, the tenet can be changed. The ingredients of this change are different perspectives, questions, and the willingness to adopt the trial while retaining its scientific integrity. The introduction of MAMS design provided them.

The innovation of MAMS design and the first trials

Looking at something known from another angle helps finding new solutions. What Prof. Parmar and colleagues did in the early 2000s goes in this direction. The idea behind the typical two-arm randomized trial is simple: how does this new intervention, specifically applied, improve outcomes for patients? Nothing wrong with it but, as Parmar states, “if you change the question and ask yourself: ‘how do we improve outcomes for patients with this disease as quickly as possible? And what are the family of interventions I could look at?’, then you start thinking broader and bigger!”. The quest for quick and efficient trials, able to retain the integrity of results even with new underlying questions, found its logical conclusion in the multi-arm multi-stage design platform, a real change of paradigm in the field.

The key principles of MAMS are two: flexibility and adaptation. The multi-arm feature enables to compare distinct treatments simultaneously, with no obligation of maintaining them until the end if they don’t give encouraging results at intermediate stages. Indeed, this approach allows to drop inactive arms and add other ones if new drugs are ready to be tested, giving shape to the multi-stage property. “You adapt as you go along. And you do it all within one protocol”, says Parmar. Thanks to this thinking, as innovative as logical, trials adopting the MAMS platform procedure can reduce costs and time required to improve the outcomes. And the first field to benefit from these advantages, not surprisingly, has been clinical oncology. A phase-III five-arm study, named ICON5, started in 2001 through an international collaboration. It aimed at evaluating alternatives in regimens of chemotherapy for patients with stage III/IV ovarian cancer, by comparing and combining them with the standard therapy based on carboplatin and paclitaxel (Copeland, Bookman, & Trimble, 2003). “We wanted to test a few different treatments (polyethylene-glycol-liposomal doxorubicin, gemcitabine and topotecan [AN]), and we had no reason to choose between one or the other”, remembers Parmar. For this reason, they considered them all in the same study, defining the first multi-arm trial. “It was the first toe in the water about this idea for us”, he says. ICON5 showed that the addition of a third cytotoxic agent didn’t improve the progression-free and overall survival of patients, but its results provided an important knowledge on the effects of different therapies, with a significant saving of time (Bookman, 2006; Sia et al., 2023).

While ICON5 was the first application, it’s with the STAMPEDE trial for prostate cancer, started in 2005, that Parmar and colleagues exploited the full potential of MAMS design, with astonishing results. In those days, there was an extreme need for progress in the treatment of this tumour. As Parmar explains, “after the discovery of hormone therapy for patients with prostate cancer in the 1940s, nothing had changed, and for over 5 decades we had consistently failed to make any progress. It was a disaster”. The emergence of new approaches could restore the hope in the field, and the MRC had an advantage: as an academic institution, the interest lied not in a single drug, but in improving outcomes. Hence, potential solutions were considered all at once. The first comparisons considered the bisphosphonate zoledronic acid, the cytotoxic taxane docetaxel, and the inhibitor of cyclooxygenase (Cox-2) celecoxib, administered alone or in combination, in addition to the standard of care, androgen deprivation therapy (ADT). Among them, only docetaxel ameliorated patient’s survival, but the increase of adverse effects suggested caution in its usage as a new part of the standard therapy (James et al., 2016, 2012).

Over the years, new arms were added and dropped to STAMPEDE: from 2013 to 2018, radiotherapy was tested on men suffering from advanced prostate cancer, increasing overall survival in those with a low metastatic burden (Parker et al., 2018); a further research arm incorporating abiraterone, an inhibitor of steroid synthesis, with or without enzalutamide to contrast androgen receptor signalling, showed that its addition alone to ADT improves both failure-free and overall survival in men with metastatic or locally advanced cancers (Attard et al., 2023; James et al., 2017). The last treatments, involving metformin, currently used in type-2 diabetes, and transdermal oestradiol hormone patches, are now being evaluated, and represent the conclusion of STAMPEDE. The plethora of significant results and information achieved in 20 years is a break with the past: “Testing each of those interventions separately in two-arm trials would probably have taken us 80 or 100 years” concludes Parmar.

The STAMPEDE trial highlights another element of MAMS design, always considered while planning new arms and stages: the different mechanisms of action targeted through each evaluated treatment. Two main reasons justify this choice: “First, there are more opportunities of reaching better outcomes”, says Parmar, “and second, most of the major diseases are currently treated by combination therapies to get real effects. With MAMS protocols, the strategy of evaluating treatments with different modes of action means that it is possible to couple two or more interventions that proved successful”.

The hidden complexity of MAMS design

The efficiency brought to clinical practice by the MAMS platform trial logic is indisputable, to the extent that a question arises spontaneously: why had no one thought of this before? Actually, someone did. Between 1947 and 1951, the MRC carried out additional trials on patients with pulmonary tuberculosis, as a natural prosecution of the randomized trial to assess the efficacy of streptomycin: a second trial to test para-aminosalicylic acid (PAS) alone or in combination with streptomycin, and a third one to study the correlation between PAS dosage and streptomycin-resistant strains (Daniels & Hill, 1952). The MRC applied a strategy that may recall the MAMS one, by concatenating the trials through a scheme very similar to STAMPEDE. Thus, the idea of a multiple comparison of interventions has been present since the dawn of contemporary clinical research. Still, it was an unripe attempt, since something crucial was missing: a solid and reliable methodology.

Beyond the appearance of something logical and simple, the MAMS platform design is built on a complex statistical methodology, the basis of its capability to retain the integrity of trials, making them trustable. “It’s like driving a car, it’s quite simple and you know it”, confirms Parmar, “but if you open the bonnet, you realize that the making and fixing of that car is an entirely different matter”. The complicated thinking underneath this approach led the MRC Clinical Trials Unit to develop a dedicated website with extended information and useful tools to help planning trials. Nevertheless, this complexity can be scary. At the time of MAMS first proposal, some clinicians raised doubts and concerns: a trial adopting many research arms could not give reliable results, turning patients away from recruitment. “It was seen as madness”, continues Parmar, “but actually, it was the opposite”. Indeed, MAMS platform trials ensure their integrity through the direct randomization at every stage, and they have always shown very fast enrolments among patients, generally enthusiast for the chance of a quicker improvement of therapies.

The peculiarity of a multi-arm multi-stage adaptive platform makes even more interesting the fact that, on the other side of the ocean and in the same years, a similar thinking was developed. A multicentre collaboration in the US gave rise to the I-SPY 2 trial, aimed at identifying effective therapies for the treatment of breast cancer (Barker et al., 2009; Rugo et al., 2016). While the feature of adding and dropping arms makes it akin to the UK platform, its connotation as a phase-II trial represents a substantial difference from ICON5 and STAMPEDE studies. As Parmar says, “a phase-II trial is still to persuade yourself there’s value in testing a treatment further, while a phase-III one must persuade the rest of the world of the integrity of the obtained results”. MAMS platform trials allow this, thanks to their sound methodological basis.

Conclusions

During the last two decades, various trials have taken advantage from the MAMS platform protocols. Many of them regard different types of tumours, but this scheme has been applied to other major diseases which are desperate to make faster progress: examples include neurodegenerative, vascular and infectious diseases. As Prof. Parmar states, “there is a real need to change the paradigm in areas where we've made little or no progress for years, otherwise we will carry on failing, and failing slowly.” In other cases, the attention of new MAMS studies is aimed to identify the optimal frequency of approved treatments, as for cancer immunotherapy with the ongoing REFINE study (Merrick et al., 2023).

Thinking about it, the MAMS experience gives hopes for the future: “There is much more acceptance of this design and a growing appetite for it from funders, patients, doctors, regulators, and the general public; this is encouraging”, says Parmar. This aspect couples with the overall understanding of the simple logic within MAMS trials. It all started by recognizing that nothing magical characterizes the number two. And it ended with an approach that, in a certain way, brought some new magic to clinical research.

Acknowledgements

The author wishes to thank Prof. Mahesh Parmar, for his helpfulness and passion in providing an important historical overview of the subject.

References

Attard, G., Murphy, L., Clarke, N. W., et al. (2023). Abiraterone acetate plus prednisolone with or without enzalutamide for patients with metastatic prostate cancer starting androgen deprivation therapy: final results from two randomised phase 3 trials of the STAMPEDE platform protocol. The Lancet Oncology, 24(5), 443–456.

Barker, A. D., Sigman, C. C., Kelloff, G. J., et al. (2009). I-SPY 2: An adaptive breast cancer trial design in the setting of neoadjuvant chemotherapy. Clinical Pharmacology and Therapeutics, 86(1), 97–100.

Bookman, M. A. (2006). GOG0182-ICON5: 5-arm phase III randomized trial of paclitaxel (P) and carboplatin (C) vs combinations with gemcitabine (G), PEG-lipososomal doxorubicin (D), or topotecan (T) in patients (pts) with advanced-stage epithelial ovarian (EOC) or primary periton. 2006 ASCO Annual Meeting.

Copeland, L. J., Bookman, M., & Trimble, E. (2003). Clinical trials of newer regimens for treating ovarian cancer: the rationale for Gynecologic Oncology Group Protocol GOG 182-ICON5. Gynecologic Oncology, 90(2), S1–S7.

D’Arcy Hart, P. (1999). A change in scientific approach: From alternation to randomised allocation in clinical trials in the 1940s. British Medical Journal, 319(7209), 572–573.

Daniels, M., & Hill, A. B. (1952). Chemotherapy of pulmonary tuberculosis in young adults; an analysis of the combined results of three Medical Research Council trials. British Medical Journal, 1(4769), 1162–1168.

James, N. D., de Bono, J. S., Spears, M. R., et al. (2017). Abiraterone for prostate cancer not previously treated with hormone therapy. New England Journal of Medicine, 377(4), 338–351.

James, N. D., Sydes, M. R., Clarke, N. W., et al. (2016). Addition of docetaxel, zoledronic acid, or both to first-line long-term hormone therapy in prostate cancer (STAMPEDE): Survival results from an adaptive, multiarm, multistage, platform randomised controlled trial. The Lancet, 387(10024), 1163–1177.

James, N. D., Sydes, M. R., Mason, M. D., et al. (2012). Celecoxib plus hormone therapy versus hormone therapy alone for hormone-sensitive prostate cancer: First results from the STAMPEDE multiarm, multistage, randomised controlled trial. The Lancet Oncology, 13(5), 549–558.

Merrick, S., Nankivell, M., Quartagno, M., et al. (2023). REFINE (REduced Frequency ImmuNE checkpoint inhibition in cancers): A multi-arm phase II basket trial testing reduced intensity immunotherapy across different cancers. Contemporary Clinical Trials, 124 (October 2022), 107030.

Parker, C. C., James, N. D., Brawley, C. D., et al. (2018). Radiotherapy to the primary tumour for newly diagnosed, metastatic prostate cancer (STAMPEDE): a randomised controlled phase 3 trial. The Lancet, 392(10162), 2353–2366.

Rugo, H. S., Olopade, O. I., DeMichele, A., et al. (2016). Adaptive randomization of Veliparib–Carboplatin treatment in breast cancer. New England Journal of Medicine, 375(1), 23–34.

Sedgwick, P. (2011). Phases of clinical trials. BMJ, 343(7825).

Sia, T. Y., Tew, W. P., Purdy, C., et al. (2023). The effect of older age on treatment outcomes in women with advanced ovarian cancer receiving chemotherapy: An NRG-Oncology/Gynecologic Oncology Group (GOG-0182-ICON5) ancillary study. Gynecologic Oncology, 173, 130–137.

Bibliography

MAMS platform main website:

https://www.mrcctu.ucl.ac.uk/our-research/methodology/design/multi-arm-multi-stage-mams-platform-trials

MAMS platform link for resources:

https://www.mrcctu.ucl.ac.uk/our-research/methodology/design/sharing-our-expertise-in-mams-platform-designs

STAMPEDE trial overview:

https://www.mrcctu.ucl.ac.uk/studies/all-studies/s/stampede

I-SPY2 trial overview:

https://clinicaltrials.ucsf.edu/trial/NCT01042379

REFINE trial overview:

https://www.mrcctu.ucl.ac.uk/studies/all-studies/r/refine/