Roche Combines Analytics + Clinical Operations For Better Feasibility Plans

By Ed Miseta, Chief Editor, Clinical Leader

Corsee Sanders is a statistician by training, having earned both a bachelor’s degree and Ph.D. in statistics. Unlike many in similar positions, she did not grow up in a clinical trials environment, having only joined the clinical operations team at Roche four years ago, moving over from a position as the global head of biometrics. However, as the SVP and global head of clinical operations and external collaborations, her new role in overseeing trials with Roche is key to the success of her company’s studies.

When working in the biometrics space, Sanders was used to having clean, manageable, and analyzable data that could be used for modeling and predictions to assess and raise a program’s probability of success. She has found that in clinical operations she also has to deal with a lot of data, not just about patients, but about the clinical trials themselves.

“The data tells us how many trials are running, where they are, who is running them, how long it takes to run a clinical trial in certain countries, and information on how long it takes to recruit a patient in a specific area. All of this information is extremely quantifiable and valuable to the conduct of a study.”

But her experience working in clinical operations also has allowed her to look beyond the data. She has found that a lot of the information used in the decisionmaking process is not data captured in a system or in an evaluable format. It is information and knowledge that exists in the minds of the team members. While there is certainly value in that information, especially when coming from a clinical trials expert with extensive experience, the information is still a snapshot. It is one data point and, as Sanders likes to point out, scientists and statisticians never rely on one data point.

“I think this is true throughout the industry,” she says. “We have all of this learned information, but it is dispersed throughout the company, and it is not in a format that is analyzable. Additionally, the decisions we make based on this information do not have confidence intervals. One of my most experienced team members might tell me we should be able to recruit a certain study in 24 months. But we have no consistent and quantifiable way of reflecting how sure or confident we are in that prediction, and we are not set up to capture the basis of that prediction for future reference.”

Take Individual Experience Out Of The Equation
Roche is a huge organization, and Sanders has a staff of over 2,500 individuals working in clinical operations and conducting clinical trials across 60 countries. The size of the organization complicates the importance of the decisions she often has to make with incomplete information.

“We are making decisions that have a significant impact on both clinical operations and the company as a whole,” she says. “When a company commits to planning for a certain number of programs, those timelines are based on information from the clinical operations group. But too often those decisions are not based on information that is easy to analyze or quantify. As someone who is the head of global clinical operations, I can tell you those decisions can make me feel very uncomfortable.”

Sanders is quick to point out that the source of the information does not make it wrong or mean the decisions based on it are incorrect. She stresses again that often the information is coming from brilliant people with a lot of experience performing trials. But as someone who has come into her position from a technical and scientific perspective, she felt the company — and the industry — could do better. If a method could be devised that would apply scientific, quantitative, and highly technical scientific methodologies to the decision-making process, those decisions would no longer need to be dependent on the experience of select individuals.

The first thing she wanted to do was look at what information was available to her. She knew she would need to look at solid data to accurately see what was going on, and she also knew that not having access to that data would again make her feel very uncomfortable with the decisions she would be forced to make.

“We looked at our clinical trials systems and other systems that were available to us,” she notes. “I don’t think clinical operations had been the focus of high-end technologies in the past, and as a result you find many disparate systems that do not link or overlap in any way. I think this situation is common across the industry, and yet companies still need to collect data from these systems. When the information you need resides in two different places, it is not straightforward to get that information in a timely manner and know that it is accurate.”

The Search For An Umbrella Solution
To try to simplify the duplicative data that existed, Roche’s late-stage clinical operations group, which is headed by Sanders, launched a program titled Trials and Related Information Systems (TARIS) to try to make sense of the linked and related systems. She acknowledges this effort has taken a lot of resources, and it will take some time to realize the full benefits, but feels that it will all be worthwhile if the end result is a single source of reliable, pertinent data that can be accessed and analyzed at different levels.

“One of the goals is to create an information layer that is an umbrella for all of our clinical trials systems,” she says. “If someone wants information, they will not have to go to different systems to get it. In my role, I want to know how we are performing relative to our goals. To do that, I should not have to go to different systems and call on five different CROs. I should be able to see that information on this information layer, because all of the data is being fed directly into it. And when I find the data I want, I also should have a very high level of confidence in it.”

Several different vendors that have information layer technologies are currently being evaluated, and Sanders is also looking at leveraging technologies already in place at Roche. A vendor decision is expected to be made by the end of the second quarter of 2016. She notes there are numerous challenges, including the fact that many of these technologies are still relatively new. Another is getting all internal stakeholders on board. She is taking a slow and measured approach to ensure buy-in from all job functions that will be impacted by the changes. In the meantime, data is being broken up into “core data packages” so as to better move them into the new architecture being built. “This has made it easier to get alignment and has enabled us to test structures with smaller amounts of data and then apply the lessons learned as we move on to the next package,” she says.

While the new information layer will not eliminate all of the different systems already in place, it will allow the company to share data more easily and seamlessly across systems and analyze that data via reports and dashboards. According to Sanders, the systems are transactional tools; they facilitate the input or extraction of data, but the information layer will hold the information and remain the single source of truth.

The program already seems to be simplifying the jobs of many in her organization. Roche conducts an incredible number of programs collecting large amounts of trial data. The teams should be able to use this data, with matching analytics and benchmark information, to assess how long it will take to complete a study under different assumptions — and to determine if they can do better.

When the fully implemented information layer is in place, teams and senior leaders like Sanders will have the ability to interrogate the data through queries. Examples of these queries include:

• Which countries/sites have the highest likelihood of delivering on their commitments to patient enrollment?
• If Germany drops out of a study, how will it affect our model?
• If we add more sites to a study, what is the likelihood that the timelines will move up by a quarter?
• How is Roche clinical operations performing relative to industry peers?
• Given our potential portfolio two to three years from now, what operational factors should we start preparing to ensure high likelihood of execution success?

“We need this capability to conduct predictive modeling within the system,” she adds. “We need to have the ability to put science, analytics, and sophistication into our feasibility and oversight activities, which will improve the quality of the answers we get. We believe this is a big game changer when it comes to performing clinical trials because to get this capability, we had to deal with many systems. When you have multiple systems in place, things can get very complex if you are not able to ‘discipline’ the systems to match your needs in the simplest possible way.”

Does Clinical Trial Intelligence Belong With You Or Your CRO?
With the value of data increasing every day, many pharma companies are faced with a tough decision: Should we have that intelligence in-house or allow it to reside with CRO partners? On this point, Sanders and her leadership team thought long and hard before coming to a decision. Ultimately, Roche will focus on managing its late-stage studies within the company, with few exceptions, and outsourcing only site monitoring. “The goal is to have Roche staff around the world working with our investigators, being local experts in the therapeutic areas, having close oversight of study quality, and having information from these trials in a standard, comprehensive, and accessible format to make better informed decisions and explorations,” Sanders explains.

She cites a few considerations for her team’s reasoning. “Global CROs are professionals with expertise and proven ability to execute clinical trials and the flexibility to absorb the large shifts in resource demands inherent in Phase 3 programs. There are many instances when leveraging global CROs is the best, or sometimes the only, way to execute trials. The direction we are taking was driven primarily by our desire to have direct site relationships and in-depth understanding of the environment at the country level to inform the design and planning of trial execution, especially in areas new to us. An equally important driver is the need to have a consistent model to enable rapid and easy data capture and access to clinical trial information across the hundreds of studies and sites in a way that is meaningful.”

When asked about the concept of Big Pharma buying a large CRO, Sanders notes that although this is an interesting evolution of the pharma-CRO relationship, Roche late-stage clinical operations has not considered this approach.

Sharing Success Stories With Others
Sanders believes there are many things the industry can do better regarding clinical trials, and she thinks getting companies to work together on solutions will speed their development and implementation. For that reason, she heads up external collaboration initiatives for Roche. She also sits on the board of directors at TransCelerate Biopharma Inc. as the vice chair.

“I always have felt that cooperation is key to success in pharma,” she states. “There are areas in which we all compete, but there are areas where we don’t and where working together is smarter. We have an internal theme within Roche product development that’s called Smarter Together. We want our people to feel that is the best thing for us to do.”

Externally, Sanders believes there has been a major shift in thinking across the industry. Roche was one of the first 10 participants in TransCelerate, and when she was first called about participating in it, her response was “Forget it.”

“I was certain that nothing was going to come out of it,” she now says with a smile. “I was very cynical of the whole effort. I felt there would be a lot of talking back and forth, and it would take five years before anything would be approved. But after attending just one meeting, I realized it would be a serious collaboration. The structure that Dalvir [Dalvir Gill, CEO of TransCelerate] put in place has made it successful. At the industry level, I think we will continue to see this type of cross-company collaboration.”

One success story that Sanders proudly states has already come out of TransCelerate is the Site Qualification Training Program. At Roche there are several clinical operations groups, and by design they all operate independently. That means an investigator working with Roche may have to be trained several times. Taking that to an even higher level, staff at a site working with multiple companies may have to go through the same training multiple times. If there were just 15 individuals working at that site needing 30 training instances, it would result in 450 separate qualifications. That can become a huge problem for both the sponsors and the sites.

“What pharma companies are doing via TransCelerate is just a much smarter way for all of us to do business,” says Sanders. “Members have agreed to a set of minimum criteria to enable a voluntary mutual recognition process of GCP Training across member companies. This allows clinical trial investigators and other site personnel to complete GCP training, which may be recognized by other TransCelerate member companies, making it unnecessary to train separately for each participating company. That is a huge relief for any investigator site. It is an investigator- centric approach that we feel will greatly ease their lives and help pharma get medicines to patients faster.”