It’s not news to anyone in the pharma or biotech industries that clinical trials have become increasingly complex over the last decade. With the shift towards precision medicine, greater emphasis on exploratory endpoints, and trials with multiple arms or dosing regimens, this trend is sure to continue. Studies now incorporate more procedures and more specimens are collected than ever before. Hard measures on the increase of the number of samples collected from patients are difficult to come by, but the general sentiment of staff involved with the management of the samples is that the amount has grown significantly.
The work of sample management is often shared among several different people at the sponsor, central lab provider, and the CRO involved with delivering the study. In some cases, companies have even created specialized positions whose sole duty is to track and manage clinical samples. To make the situation even more challenging, a push towards outsourcing often requires managing sample logistics across a network of lab suppliers. Add on the expectations from senior management and investors to get near real-time updates on the status of the trial and the pressure on staff can be immense.
At the same time, the evolution of technology for managing clinical samples has not kept up. It’s still common for staff to maintain large spreadsheets to manually track the status and location of the samples. Input needs to be gathered from multiple sources with varied formats and frequencies of update. The risk for errors to occur in such a process only grows with the complexity of the study.
Fortunately, we are now seeing automated sample management solutions enter the marketplace. To meet the needs for efficient tracking, these systems should have three basic features: an up to date inventory of all samples in the study, an ability to model what samples should be present and reconcile it against the inventory, and a compliant way of managing consent information.
Inventory Tracking
The most basic need of a sample tracking system is to be able to show where samples are located in the clinical trial supplier network in as close to real-time as possible. While this sounds easy in theory, it’s a fairly difficult task in today’s environment.
Lab providers have for a long time utilized Laboratory Information Management Systems (LIMS) to manage their own internal inventories. But, when inventory information is passed on to a customer it’s usually done in the form of a manual export and sent as a snapshot in time. This information is then combined by the staff on the receiving end with all of the other available inventories to create a complete view of the samples in the study. Unfortunately, these data quickly become obsolete and the whole exercise needs to be repeated. No wonder specialized positions have been created to manage this information.
The power of using an automated solution is that information can be exported from each lab provider’s LIMS and imported into a single tool. If these data transfers are set up to be as frequent as samples are physically transferred to the testing facility, the overall inventory would be up to date in near “real time”. Ideally the tracking system is able to accept information in a variety of formats since there are many different LIMS systems on the market and information captured about samples is not yet standardized.
Reconciliation
The second need the tracking system should address is the ability to be able to tell not just what samples are present and where they are (the inventory), but also the ability to model what samples should be present at any time.
With the knowledge of when subjects enroll in a study and the collection schedule that defines what samples are collected at each study visit, an automated system can model what samples should have been collected. This requires integration with clinical databases to gather subject enrollment and visit dates and also the ability to create flexible algorithms to deal with multiple collection schedules in one study. By comparing the results of these calculations with a real time inventory, the tracking system can flag any discrepancies as they occur rather than waiting for a failure downstream when it may be impossible to correct. There are a large number of ways that things can go wrong when dealing with clinical samples that an automated system can detect as discrepancies. Examples include transcription errors, mislabeled tubes, shipments to the wrong facilities, and missing collections. By capturing these errors, they can be flagged and the appropriate resolution process started immediately.
Consent Tracking
The third and arguably most important feature of an automated sample tracking solution is to ensure proper consent is documented for each sample collected.
With the increased focus on exploratory analyses and other research use of samples, scientists need an efficient way of associating consent with each sample. As science and knowledge progresses we may want to go back and perform additional analyses on a previously collected specimen. We would then need the ability to search for samples that had clearly documented consent for those types of analyses to be performed.
Of course, for this type of reporting to be possible we need the consent information to be present in the system. Ideally this is performed electronically via an entry into an electronic data capture system at the clinical site or through direct capture in an electronic consent tool.
An automated sample system also needs the capability to deal with changes in consent status. If a patient decides to withdraw from the study or change the status of their consented use, the samples that have been collected may have to be destroyed. Ideally, the system will automatically trigger a process to ensure the proper destruction of the samples, and monitor its completion. Because samples can potentially be located at one of many locations, existence of a complete inventory is a prerequisite for ensuring all samples can be located and properly destroyed.
Samples will also need to be tagged for destruction at the end of their storage life. The study master consent may stipulate a certain number of years of allowed storage, but often there are regional or even institutional changes to the allowed time period. An automated system needs to be able to assign the expiry for a particular sample based on the specific consent signed by the subject. Determining exactly when a specific sample should be destroyed can then efficiently be automated.
Additional Benefits of an Automated Tracking System
An automated sample tracking system removes much of the manual work of managing samples, but it also enables functionality other than just automating tasks.
With all of the data about clinical samples collected in one place, trends can be identified in ways that were previously impossible. For example, it may become clear that particular sites or countries have not been properly trained on sample processes and therefore have higher error rates than others. Without such a rich database, these patterns would likely only be captured on a study by study basis and problem areas may not be as readily apparent. An automated system allows analysis of patterns to become routine and regularly performed, alerting staff to errors before they become limiting to the success of a study.
Another need often expressed by project teams is the ability to better predict when samples will be available for testing. In fast moving studies, the data from even a single subject can drive critical decisions. By being able to model when samples will be collected based on enrollment and visit schedules, the system can be used to predict when a specific amount of samples will be ready for testing. This information enables the project team to better plan critical decision points and also allows the testing labs to ensure that the proper capacity is available to match demand.
Challenges to Implementation
While the need for more efficient sample tracking systems is clear, adoption and penetration into the industry is still low and several challenges exist for easy implementation.
Access to data from all of the providers involved in a study is critical in order to have a complete inventory. For clinical data there are pre-defined standards (such as CDISC) for how these data should be organized. As of now, we don’t have any such standards for data about samples. Without standards, it becomes necessary to negotiate data transfers individually with each provider in the network for a particular study. Because of the diversity of inventory management systems in use at lab providers they do not always capture all of the necessary data fields to create a complete inventory. Especially challenging is getting access to accurate receiving and shipping dates as well as indicators of sample status changes such as destruction at end of testing. Many providers simply do not collect these data fields in detail. Additionally, because tracking systems are not yet common, lab providers are not used to request for these types of data. Testing labs commonly transfer analytical result data but request for data transfers of sample metadata are often met with confusion. Generally, they will require custom programming, incurring both cost and additional time to implement. Once it becomes more routine, the larger laboratories will likely start providing these transfers as an expected service.
Finding a software provider that is capable of meeting all of the requirements of a modern sample tracking system is also challenging. There are many solutions in the marketplace that can meet basic inventory management and reconciliation needs, but there are very few that have the flexibility to accept varied data formats across several lab providers and also integrate consent information. Maturation in this space is likely to continue, but for implementation today, there are not many options to choose from.
Future Directions
As the industry continues to mature in this space, there are several areas that will need improvement.
All companies involved in the delivery of clinical studies would benefit from a data standard for sample metadata. It would define the information to be captured on sample location, shipping, and status so that it would be readily available for all parties. This standard would ideally be discussed in a forum that includes pharmaceutical companies, central and testing laboratories, and software vendors labs so that the needs of all parties can be addressed. With the processes and systems available today it’s difficult to capture sample collection status at clinical sites. This means that the first reasonable point of sample reconciliation is at the first laboratory (often the central lab) where the sample arrives after collection. If samples are shipped infrequently it can be days and sometimes weeks before errors in sample collection are found. Designing and deploying sample collection systems directly to clinical sites and integrating those with an overall sample tracking system would allow even earlier detection of errors.
Some of the sample tracking systems in place today can integrate with couriers to perform simple monitoring of samples during transit. Couriers currently can provide updates on when samples leave one facility and arrive at another, but not much more detail. Integrating data from embedded temperature monitors and even GPS coordinates for a package would further complete the chain of custody and integrity record of the samples as they travel between locations.
The sample systems available today can generate predictions of when samples will be available for testing, but only for subjects that have already enrolled. Being able to add inputs from enrollment prediction algorithms would provide a picture of not just the testing demand for the subjects already enrolled, but also for subjects enrolling in the future. This functionality can be a powerful tool to model when data will be available for decision making and to give laboratory managers even better information to plan their testing capacity.
Once automated sample systems are implemented across the industry the benefits will be immense. Adoption is in the early phases, but even with the solutions available today the three basic needs of a complete inventory, automated reconciliation and consent tracking can be achieved. However, for the full benefit to be easily realized the industry needs to organize around a set of standards and expectations for what data needs to be collected about clinical samples.
Daniel Joelsson is Director of R&D Strategy at MedImmune. Daniel has over 16 years experience in pharmaceutical R&D. He spent 12 years in analytical and bioprocess development with a focus on bioassays before transitioning into various roles in R&D operations and strategy. He currently leads strategic planning for Clinical Biologics at MedImmune.