No medicine is without hazard. In 2004, multiple sclerosis (MS) patients were offered a breakthrough compound called natalizumab. Branded as the first in a new class of therapeutics known as selective adhesion molecule inhibitors to prevent relapse, vision loss and cognitive decline associated with the disease, it proved effective for many, but sinister side effects began to appear in a small number of people.

The monoclonal antibody, known by its trade name Tysabri, was found to be associated with progressive multifocal leukoencephalopathy (PML), a rare and usually fatal neurological condition. Characterised by progressive inflammation of the brain's white matter, it's caused by the John Cunningham virus, normally present and kept under control by our immune systems. However, in immunocompromised individuals, the usually inert pathogen can become deadly. PML has a mortality rate of 30-50% in the first few months and, for those who survive, neurological disabilities are common.

After three cases of PML had been identified in patients taking the drug in 2005, natalizumab was voluntarily withdrawn from the market. However, after an intensive global risk management programme, it was reintroduced in 2006.

After its removal, the 3,000 patients who were treated with the medicine were studied carefully for 18 months. At the end of the study, the risk of developing PML while being treated with natalizumab was estimated to be one in 1,000. The FDA re-released the compound as a disease modifying therapy for relapsing remitting MS but only if administered by itself and not in combination with other medications for the disease. Natalizumab is now prescribed under a restricted distribution programme, which requires specialised enrolment, counselling and follow up for each patient given the drug.

"For biologicals, the production process may be changed post-marketing, which could give rise to new safety concerns."

Cases like natalizumab serve as reminders that clinical trials aren't necessarily representative of real-world experience. Before a new product is granted market access, it will have been well studied in a limited number of patients from a relatively selected population who are followed up for a limited period of time. It can be hard to spot adverse drug reactions (ADRs) that might only appear in certain types of people.

That's why the field of pharmacovigilance – the practice of monitoring drug safety in the marketplace – is so important. Since the thalidomide tragedy in the 1960s, most countries have systems in place for reporting ADR information to the authorities. Once placed on the market, drugs continue to be monitored to ensure any aspect that could impact the safety profile of a medicine is assessed and that the necessary measures are taken.

"Every pharmaceutical company should have a pharmacovigilance system in place," explains Niels Vermeer from the division of parmacoepidemiology and clinical pharmacology at Ultrecht University. "If you receive any ADR reports, you assess them and store them in a database, and regularly screen the database for unrecognised associations between drug exposure and clinical events. It's part of a range of routine pharmacovigilance activities that every sponsor has to do."

Best-laid plans
In 2005, EMA introduced risk management plans (RMPs) as the instrument for the planning of pharmacovigilance activities and risk minimisation of new drugs.
It was hoped this marked a shift towards a more proactive approach to gaining knowledge of safety concerns for newly approved therapeutics. In the European Union (EU), companies must submit an RMP at the time of application for marketing authorisation. RMPs include information on:

  • a medicine's safety profile
  • how its risks will be prevented or minimised in patients
  • plans for studies and other activities to gain more knowledge about the safety and efficacy of the medicine
  • risk factors for developing side effects
  • measuring the effectiveness of risk-minimisation measures.

Before this, regulators had relied primarily on spontaneous reports and industry-initiated studies but, as high-profile drug withdrawals occurred and increasingly complex drugs entered the market, EMA saw a need for a new risk assessment and pharmacovigilance method.

"Prior to the introduction of RMPs, pharmacovigilance was mostly reactive," explains Vermeer. "A product was put on the market, and then we waited for the ADR reports to be submitted, or additional studies to be conducted. The regulatory system was waiting for the data to come in, but usually didn't proactively specify what data was actually needed to address existing uncertainties."

Vermeer holds a PhD candidate position at Utrecht. As part of the regulatory science collaboration between the Medicines Evaluation Board (MEB) and the university, he conducts research in the field of pharmacovigilance of biologicals. When he's not doing that, he works as a pharmacovigilance assessor at the MEB, in which he assesses safety data for medicinal products that are either nationally authorised in the Netherlands, or centrally authorised in Europe.

"RMPs enabled us, for the first time, to systematically document which uncertainties there were at the moment of market approval, and that allowed us to study the evolution of uncertainties over time."

Biologicals, a relatively new class of drugs to which natalizumab belongs, pose novel challenges in pharmacovigilance. That's important because biologicals already dominate the list of best-selling medicines in Europe; eight out of the top ten fit into this category. In contrast to conventional small molecules, which are synthesised chemically, biologicals are derived from living sources.

Because the production and purification process is more complex, minor changes in manufacturing can have major implications. Take pure red cell aplasia, for example. This rare disorder of profound anaemia appeared in patients taking a particular formulation of recombinant human epoetin in 1998. Causation was attributed to formulations without human serum albumin, subcutaneous administration and uncoated rubber stoppers.

"For biologicals, the production process may be changed post-marketing, which could give rise to new safety concerns. The characteristics of the product might change and it may influence the safety profile of the product. This isn't something we see with small-molecule drugs," Vermeer explains.

But could RMPs, now a cornerstone in the pharmacovigilance of new drugs in Europe, result in sufficient knowledge gain over time, particularly for more complex drugs? Vermeer and colleagues embarked on a study to examine the development of safety concerns in the RMP after approval. They wanted to know to what extent uncertainties are resolved under this new method. Their findings are published in Nature's Clinical Pharmacology & Therapeutics.

The team reviewed subsequent RMP updates for a cohort of recently approved drugs. They chose two groups of products to investigate: biologicals and small molecule drugs intended for chronic use. Biologicals were chosen because of their association with specific risks and more uncertainties at approvals, and chronic-use drugs were selected as their exposure time is relatively limited at time of approval, so side effects might be missed in clinical trials.

Unlucky for some
Over 600 safety concerns were described in the initial RMPs for the 48 products in the cohort, which corresponded to a median of 13 per product. The majority of these (74%) reflected uncertainties (potential risks or missing information). And in general, more safety concerns per product were described for biologicals than for small-molecule drugs.

"We found that, after five years, 20% of the uncertainties that existed at approval had been resolved," Vermeer reveals.

The rate at which uncertainties were resolved did not differ between small-molecule drugs and biologicals.
"We are not able to compare this finding with the previous practice," adds Vermeer. "RMPs enabled us, for the first time, to systematically document which uncertainties there were at the moment of market approval, and that allowed us to study the evolution of uncertainties over time. So this is the first time we can actually investigate this."

It sounds a relatively modest accrual of knowledge but, while Vermeer expected the number of resolved uncertainties to be higher with the RMP approach, he acknowledges there could be multiple explanations for the resulting value.
"The study duration was five years and it may be too short to adequately study for example malignancies," he opines. "It could also be that we measured this knowledge gain for these products along this resolving of uncertainties, but it could also be the case that for certain uncertainties, new knowledge has been made and put in the product information, but that the safety concern has remained as an uncertainty in the RMP."

No one can deny that pharmacovigilance is a fundamental practice, and a field that has developed in response to lessons learned from landmark safety issues over the years. RMPs are an important refinement of European drug safety, but there are still limitations that must be addressed. Further research toward better understanding of pharmacovigilance systems is needed.

"A RMP is a tool that's very good to have and it's very useful to proactively think about what data we need in the post-approval phase and how we can really ensure that the uncertainties are being addressed faster," concludes Vermeer. "But I think we have to learn as regulators and as industry how to deploy them to our best advantage. And when we're more experienced, we will use them in a better way."