Establishing a Robust Platform for Image-Enabled Trials

As the inclusion of patient images in clinical trials starts to become mainstream due to a stiff push from the FDA's Critical Path Initiative, pharmaceutical companies and CROs would be well served to understand the impact that this shift will have on their existing workflow, viewing, and data management systems and applications.

Unlike traditional document-based clinical trials, new clinical trial protocols that feature medical images add a layer of complexity associated with the uniform acquisition, management, reading, and subsequent audit and storage of patient images primarily for regulatory, research or adverse event purposes. These new requirements should be taken into consideration to allow time for pharmas and CROs to plan for the necessary upgrades to their front-end applications and back-end infrastructures.

This article will examine the impact that the inclusion of images in clinical trials will have on trial study acquisition; study workflow, reading, and submission; and data management, archiving, and protection. It will also present some recommendations that organizations can undertake today to minimize the impact that the addition of images will have on their existing applications and infrastructures.

A parallel for pharma in health care

To understand the kind of impact that the shift from documents to digital images can have on an organization's IT systems, it is not necessary to look far: One can easily be found in the shift that took place earlier this decade in the health care industry. With increasing pressure from the government and health insurance companies to improve patient outcomes, and pressure from shareholders to improve business results, many hospitals started to modernize their clinical IT systems in the late 1990s.

The first shift involved moving to integrated hospital information systems that enabled physicians and administrators to manage the flow of patients (admission, treatment, discharge, and billing) more effectively and at a lower cost. A second, more recent, and significant shift has been the aggressive move from film-based medical imaging (remember the film based x-rays that you had to carry around from doctor to doctor?) to newer digital information systems and picture archiving and communication systems (PACS) that enable those images to be digitized, shared, and over time included into an electronic medical record (a single, secure, and shareable view of a patient's medical profile).

The implementation of these PACS systems has been very rapid, with adoption rates that were in the teens just five years ago rocketing to 80% in the United States and Canada today.

This rapid shift to digital from document-based systems, however, was not met with an accompanying upgrade of the existing workflow and data management infrastructure. First, the addition of images that can be 10 to 30 times larger than documents put a massive strain on the storage systems installed at most hospitals. Second, since these images were no longer available in physical form, hospitals were forced to share or replicate these images across locations, thereby straining the hospital's network and middleware infrastructure. Third, significantly easier access to patient images caused clinicians and doctors to ask for more information (e.g., earlier patient studies or newer views of data) across multiple locations, thereby compounding the problem.

As more systems were installed and patient study volume grew, IT was not able to keep up. As a result, health care IT departments have faced significant challenges in meeting the application uptime, reliability, performance, and data protection service levels that they were able to easily offer their users just a few years before. Clearly, many of the challenges that hospitals are facing today are challenges that pharmaceutical companies and CROs will be wrestling with soon.

Figure 1. Three independent but integrated architectural layers

Image-enabled clinical trials infrastructure

When designing a robust and scalable image-enabled trial infrastructure, it is helpful to understand the different architectural layers and their respective functions. Figure 1 shows such an architecture and describes an image-enabled clinical architecture consisting of three independent but integrated layers:

• the application layer

• the image management layer

• the storage management layer.

Application layer. The application layer is probably the most straightforward of the three layers in that it represents the applications and modalities that capture the information at the trial site and passes these images to the image management layer. These applications and modalities could range from radiology, to cardiology, to endoscopy, etc., and will be typically already installed at the clinical trial site. Most trials that include images today will capture images from many different locations and trial sites spread throughout a geographic region or nation.

Potential Image Storage Solutions

What is critical about this stage in a trial is that data acquisition across all sites must be performed in a consistent and auditable manner to minimize any variability that could lead to inconsistencies in the trial itself. Conditions at each site may vary, so it is critical that as soon as an image is processed and de-identified that it be quickly deposited via a standards-based interface (e.g., DICOM, PDF, JPEG, DOC, RTF, WAVE) into an auditable, flexible, and traceable workflow and viewing platform that represents the next layer of the architecture, the image management layer.

Image management layer. Once an image is acquired and digitized, the image management layer enables the transmission, viewing, reading, and sharing of clinical images from the trial site to the CRO and eventually back to the trial sponsor. These image management layers have traditionally been a combination of commercial and custom applications that have been built and expanded over the years to address new clinical trial needs. The inclusion of images and the resulting need for the inclusion of more sophisticated viewers and diagnostic techniques presents a unique opportunity for pharmas and CROs to undertake an evaluation of their existing systems to determine whether or not they may need to be upgraded or replaced.

To address the requirements associated with the inclusion of images in trials, pharmas and CROs should look for image management applications that feature integrated viewers and productivity tools for cardiology, radiology, orthopedics, and other diagnostic imaging. Additionally, applications should deliver (or integrate to) Web-based imaging viewers and clinician work lists with robust workflow capabilities that can facilitate distributed or centralized reading for more efficient analysis and interpretation.

Data and user access should be customizable according to patient, protocol, geography, user profile or any combination of the previous variables so that the application can be easily customized to the ever changing designs and protocols of today's trials without the need for significant manual intervention.

Another important requirement for image management applications is their ability to capture and share images from multiple systems via the use of open standards such as DICOM, HL7, and IHE. Applications that support these formats are able to more easily interoperate with the heterogeneous clinical and imaging systems found in clinical trials today. Adherence to these formats also enables pharmas and CROs to establish a common system infrastructure that will allow easier adaption to new trial designs and future applications.

Finally and most importantly, any image management application must represent a controlled and validated environment that conforms to the guidelines of 21 CFR Part 11.

Storage management layer. Throughout the acquisition, analysis, and interpretation cycle of clinical trials, image management applications must store and make accessible clinical data and images to any number of users across multiple locations and in varying formats. Although the storage and protection of trial data might sound straightforward and simple, the addition of images into this mix adds a surprising level of complexity.

First, digital images are exponentially larger than traditional trial documents and require a much more flexible and scalable storage infrastructure that can support much larger data sets. Second, the size of these images makes their electronic transfer from the trial site to the CRO more challenging, as greater bandwidth is required to generate acceptable performance for acquisition and remote reading. Third, clinical trial images must be stored for regulatory compliance purposes potentially for decades, requiring that organizations ensure that data remains current and corruption free for many years.

In short, the amount of storage required to manage clinical trials today may easily increase ten-fold as the inclusion of images continues to accelerate. This will require that pharmas and CROs design new storage infrastructures, keeping in mind the following key requirements:

• Robust data protection: Trial images and documents, no matter how old, should be protected to ensure that data is not corrupted or lost.

• Automation of storage administration functions: Storage administration functions such as backups, hardware refreshes, and data conversion should be automated as much as possible to increase IT/storage administrator productivity and efficiency.

• Single site or multisite capabilities: Trials will require that images and documents be stored locally but be made accessible at many different locations, potentially over a WAN as well as a LAN.

• Real-time failover and disaster recovery capabilities: Disk failures are inevitable as data ages. Additional protection such as automated failover and rebuild capabilities are required to keep applications running despite unplanned downtime.

Conclusion

The accelerating adoption of images in clinical trials is creating a new set of requirements that pharmas and CROs should evaluate today to remain competitive and improve trial outcomes.

Hugh Rivers is business line executive, WW Healthcare Life Sciences Solutions, IBM Systems and Technology Group.

Eric Ditkowsky* is business development manager at ScImage in Los Altos, CA, email: eric_ditkowsky@scimage.com

*To whom all correspondence should be addressed.