Deep technology for the optimization of cryostorage

Abstract

Cryopreservation, for many reasons, has assumed a central role in IVF treatment cycles, which has resulted in rapidly expanding cryopreserved oocyte and embryo inventory of IVF clinics. We aspire to consider how and with what resources and tools “deep” technology can offer solutions to these cryobiology programs. “Deep tech” has been applied as a global term to encompass the most advanced application of big data analysis for the most informed construction of algorithms and most sophisticated instrument design, utilizing, when appropriate and possible, models of automation and robotics to realize all opportunities for highest efficacy, efficiency, and consistency in a process.

Introduction

The birth of Louise Joy Brown in 1978 from in vitro fertilization (IVF) heralded a revolution to overcome infertility; four decades on, aspiring parents continue to benefit from the unrelenting pursuit of excellence for higher rates in clinical outcomes. Advances in the full array of assisted reproductive technologies have yielded more opportunities for fertilization, embryo development, and embryo selection through intra-cytoplasmic sperm injection (ICSI), improvements in culture medium and environment, and pre-implantation genetic testing, respectively.

In contrast to these profound advancements to drive fertilization, embryo development, and implantation, the methods, and resources for storage, maintenance, and inventory management of cryopreserved embryos and oocytes, have remained virtually static. It is ironic that the ability to cryopreserve “surplus” embryos, i.e., those embryos of good or high quality remaining after a “fresh” embryo transfer, was the first adjunctive method for IVF, enabling patients to save spare embryos for future embryo transfers. While the techniques for cryopreservation—the freezing of embryos and oocytes—evolved dramatically from slow-cooling protocols that required programmable freezing machines to vitrification, the resources for storing the inventory entail the same double-walled vacuum tanks that were designed by Sir James Dewar in the late nineteenth century. Inventory logging and tracking may have evolved from paper ledgers, ring binders, or index card catalogs to electronic spreadsheets—but inventory management has remained a highly manual and error-prone data input activity.

Cryopreservation, for many reasons, has assumed a central role in IVF treatment cycles. Fertility physicians often delay embryo transfer to reduce the risk associated with OHSS and to allow the body to return to baseline after treatment with high doses of exogenous gonadotrophins. With current pre-implantation genetic testing methods, embryos must be cryopreserved prior to transfer to allow the results of the biopsied cells to be received. Oocyte cryopreservation for fertility preservation is increasingly utilized by women who collect and store their oocytes during years of high ovarian reserve, staving off the stress of managing decreased oocyte quality and the lower fertility potential that comes with advanced reproductive age. These trends, combined with the adoption of “freeze-all” cycles in which all embryos are cryopreserved to avoid transfer to a sub-optimal uterine environment in the ovarian stimulation cycle, have resulted in rapidly expanding cryopreserved oocyte and embryo inventories in the majority of IVF clinics [1].

The management of higher sample volumes is challenged by the square footage needed to accommodate increasing numbers of storage tanks, the time and labor required to monitor and replenish the tanks’ liquid nitrogen levels and to conduct yearly audits to confirm physical against recorded inventories, increase in importing and exporting cryopreserved gametes and embryos from other laboratories or commercial gamete banks, and requirement for the ability to respond to a significant emergency, such as the structural failure of a cryostorage tank.

We aspire to consider how and with what resources and tools “deep” technology can offer solutions to the cryobiology programs that are expanding rapidly in IVF clinics. These must be effective in optimizing measures to achieve the highest security and safety of the cryopreserved inventory, reflect accurate inventory and ability to track and trace each stored specimen and achieve these goals with the greatest efficiency for the embryologists who are the ultimate users and have the highest expectations of this technology.

What is “deep” technology?

The expanding catalog of utilities that collect, organize, and interpret “big” data for the prediction and identification of trends, as well as innovations across the breadth of mechanical and electrical engineering, represent the scope of technology. We offer the definition that deep technology is one or more of the array of technologies for data management, algorithm and instrumentation development, and the design and plan for deployment and integration of the most advanced methods and concepts from these fields to optimize outcomes from a given process. “Deep tech” has been applied as a global term to encompass the most advanced application of big data analysis for the most informed construction of algorithms and most sophisticated instrument design, utilizing, when appropriate and possible, models of automation and robotics to realize all opportunities for highest efficacy, efficiency, and consistency in a process. Deep tech is expected to catalyze even greater quality control and quality assurance.

Requirements for optimal cryostorage

The exceptional challenges of contemporary cryostorage; its rapid expansion, the singular nature of the stored inventory, and the most stringent requirements for maintenance of cold chain and most rapid response to a system failure endangering the integrity of the specimens, must be highlighted.

Current treatment paradigms in IVF point to consistent increases in inventory through the daily addition of cryopreserved embryos and oocytes. These specimens are, literally, in many cases, priceless and transcend the assignment of value, as they represent immeasurable and possibly final and limited efforts to achieve parenthood. Both capacities for storage and the most effective system to monitor temperature and respond to any deviations from a safe zone (i.e., below the glass transition temperature at which warming is consistent with de-vitrification) are important attributes to safeguard specimens that simply cannot be replaced.

Technological advances in system monitoring will involve the integration of several features, such as the urgent warning and notification of users, confirmation of that notification, and trigger of mechanisms to conserve conditions against failure until an emergent, real-time solution can be implemented. Staving off an emergent response can be achieved by the reporting of continuous analytics to the end users, who will discern trends from well-designed data that may portend irregularities or unexpected increases in, for instance, liquid nitrogen consumption, or declination in liquid nitrogen levels. Data analysis [2] can serve as an early warning and pre-empt mechanical or structural failures. The structural aspects of cryostorage (monitoring the physical integrity and internal environments of the storage units) are just one aspect. Inventory management has also acquired both the volume and complexity to warrant technological enhancement.

Collections of cryopreserved embryos have swelled with the advent of single embryo transfer freeze-all cycles and genetic testing of embryos. These collections must be adequately described, with data points, for example, (1) owner’s name; (2) second unique identifier; (3) day of development; (4) morphologic grade; (5) infectious disease screening results of the owner; (6) genetic testing result; (7) status or disposition, e.g., frozen, thawed/warmed, exported, donated, discarded; and (8) any special descriptors or notes. Each cryodevice or cryo-container could also have a unique identifier and must be described for the number of specimens stored on or within it, e.g., a cryodevice holding two mature oocytes.

An ideal inventory technology would avail itself of a labeling model armed with an electronic reader [3] so that as the sample is added to or withdrawn from storage a record of its movement is tracked in real time. Such a model will preclude inconsistencies between the physical inventory and the spreadsheet or handwritten inventory, which results from lapses in manual data entry and record of sample movement into or out of a storage tank. The fact that regular cryostorage inventory audits are required speaks to the inadequacy of how specimens are currently being managed.

In addition to augmenting record keeping, the best technology for cryostorage will allow real-time sample tracking during storage or shipping and provide data on the specimen’s ambient condition. Users will appreciate the continuous assurance that the cold chain is preserved when the specimens are outside of their physical control. The increased geolocation capability [4] will mitigate the stresses encountered by dry shippers gone “missing” in the shipping process.

Current resources and practices of cryostorage for embryos and gametes

Cryostorage in reproductive laboratories is surprisingly unchanged from the earliest iterations. Cryodevices are typically stored in goblets attached to aluminum canes. The density of storage is dependent on the cryodevice type, which has evolved over the years from straws to slender plastic devices appropriate to vitrification that are more compact. Canes and goblets are stored in the individual canisters of dewars that can vary from the six to twelve-canister models that when placed on wheeled bases afford mobility, to high-capacity bulk tanks.

Specimen deposit into or removal from the tanks is entirely manual, requiring the tank to be opened by lifting and removal of the lid, identification of the appropriate canister, and translocation of the canister to a position where its contents can be examined, followed by moving it up and into the center of the tank, examining the contents by reading the labels affixed to the tops of canes and then extracting the specific cane to remove the desired specimen(s). With the multiple opportunities to incur innocent exposure of the specimens to transient warming, stringent training in specimen management using the tanks or moving specimens into or out of them must emphasize the safest handling.

Storage security, specifically ensuring the vapor or liquid nitrogen levels in the tanks are maintained, is achieved by after-market installation of temperature probes (occasionally) pierced through the dewar’s lid in the smaller format tanks. Most frequently wireless now, these detection systems alert the users when the temperatures rise to some threshold. Whether this allows adequate time for an effective response to this emergency will depend on the nature of the cause for the temperature increase, e.g., loss of vacuum or too low a level of liquid nitrogen from failure to replenish [5]. Accredited laboratories are required to have an emergency response plan, and many maintain a spare, fully charged tank to receive specimens that may have to be removed from a failing tank. Some of these systems have a requirement to “arm” or “secure” them at the end of the workday, injecting the possibility of a lack of monitoring from the omission of this step from the laboratory’s quality protocol and procedure. Storage security paradigms are often user-defined and there is no standardization [6] in how guidelines and regulations are interpreted.

While automated LN2-filling systems [7] can be installed in bulk storage tanks, smaller dewars require manual monitoring and replenishment of LN2 levels. With expanding numbers of tanks, sometimes stored in multiple sites if space is limited, a single repository cannot be achieved in some clinics. This critical task can be exceptionally onerous, requiring the handling of potentially dozens of tanks. Each dewar must be opened and, using a meter stick, sampled to view the frost line. In cases where the LN2 manifold and distribution nozzle is distal to the storage site, each tank must be moved to that location for filling, or LN2 must be dispensed into a container suitable for transport, and then brought to each dewar for filling, increasing the potential for accidents and spills.

The applications of AI in the maintenance and security of a cryostorage facility can be apprehended as providing the earliest alert to an emerging problem and effective interim solutions. Machine learning could be used to build models that could recognize, for instance, the incremental increases in LN2 consumption that may portend a structural defect in the tank’s seal at the neck, one of the most vulnerable points. To stave off the concern that a single individual will respond to the tank alarm system, AI algorithms [8] could concurrently notify additional resources who could, for example, deliver extra supplies such as LN2 or more tanks to receive specimens from a failing unit. The advantages of having AI fortify emergent responses and actions will be invaluable in relieving the stress of the laboratory staff and more effectively meeting the quality program’s aspiration of biorepository security and continuous integrity [9].

The evolution of an aspect of cryostorage as fundamental to specimen integrity as labeling can be traced [10, 11] in IVF laboratories, from handwriting on the earliest cryo-containers used for cleavage-stage embryos: hand-pulled glass vials, plastic vials, and straws—to printed labels that are applied to the singularly slender dimensions of the cryodevices used for vitrification.

The benefits of AI will be realized more readily when the most contemporary models for labeling and tracking the disposition of individual samples are applied, and when the identification scheme has evolved from human-derived to the automated assignment of unique identifiers. Barcoded or RFID-enabled labels will deliver that attribute and contribute to the development of databases for cryopreserved specimens useful for identifying trends and outcomes, particularly coupled with specimen descriptors such as patient demographics, clinic, and laboratory demographics, and practices, morphology, stage of development, and genetic status. Epidemiology studies as it relates to ART will be enhanced by the adoption of standardized and objective data capture systems [12]. Analysis of data may hone algorithms for clinically relevant information such as the selection of an embryo for transfer or the number of oocytes to warm for an IVF treatment cycle. For a resource as central to ART as cryobiology, AI and its derivative algorithms must supplant all analog methods of specimen information and data, e.g., registration, listing, description, and disposition that are still practiced in most centers using hardcopy ledgers or spreadsheets. AI may also overcome the inefficiencies posed by the non-standardized practices in labeling cryopreserved oocytes, sperm, and embryos [10].

On labels for cryopreserved specimens, laboratories typically include two unique patient or owner identifiers but may choose among an array of additional information to include on the label: date of cryopreservation, developmental stage, morphology, number of specimens, cryodevice number, partner identifiers, embryo serial number or unique identifier. Barcoded labels will simplify the label and its “reading” by containment of all requisite information in the cryo-database, precluding labels with more difficult-to-read content. Moving to systems that automate the unique identification of these specimens and devices will remove the human glue currently required to stitch the specimen labeling and audit trail together.

Innovations to cryopreservation

Consistent with the aspirations to identify, customize, and deploy elements of AI and automation to cryostorage, a robotic system complemented by data collection and analytics has been proposed and described [13, 14]. Continuous monitoring with reporting to not only the laboratory team—the embryologists—but to a central site provides remote intervention when needed. This may be the option to remotely fill a storage unit with LN2 during the interval between alarm and transit to respondents to the location and notification to dispatch a replacement tank to the laboratory.

Synergy to security is achieved through improved containerization of cryo-specimens. In place of the long-used cane and goblet model—a tube clipped to an aluminum holder—a stronger, thicker-walled unit comprising a barcoded tube with RFID and a cap for complete enclosure of the specimens provides advantages. In addition to the advantage of more secure containment, i.e., elimination of the possibility of goblets becoming dislodged from canes with the spilling of unsecured specimens into the dewar, the RFID further enables reporting and tracking of geographic location.

Enhancements to monitoring and storage can also be realized by the integration of electronic medical records and cryo-inventory software, given that these may be two independent databases in many clinics. Through communication of these two sources of records, coordination of patient care can be more efficient, e.g., the addition of specimens to the database during a treatment cycle, or withdrawal of an embryo from the inventory for a frozen embryo transfer (FET) is recorded in a single session in one software program. This model also staves off inaccuracies in the cryo-inventory that can result from the recording of specimen disposition in a spreadsheet or written ledger model and ensures consistency with the patient’s medical record.

Having a single source of data also allows the generation of specimen container labels, offering a significant safety feature of assured, indelible and legible patient information in conjunction with specimen descriptors, e.g., cryopreservation date and serial numbers. When labels are barcoded, entry into and out of the cryo-inventory can be logged. Dual advantages are obtained: documentation of all events affecting the inventory with full details available to the user of specimen identity and disposition and application of this attribute for inventory control. A valuable enhancement is the ability to receive and upload patient-specific information to the cryo-database through an interface, specifically, the analytic results from genetic testing of a patient’s embryos. This is another instance of increased efficiency and transcription-error mitigation, by eliminating the receipt of the genetic laboratory’s report followed by the manual event of entering the report details into the cryo-embryo inventory.

With a model of electronic labeling and continuous monitoring and tracking of labeled specimens, a full accounting of the cryo-inventory is accessible by any authorized user at any time. This is useful to embryologists, clinicians, and regulatory agency inspectors who assess the quality of laboratory cryobiology practices in inventory management for accreditation and certification, and to patients who can now realize a novel and maximally informed connection to their cryo-specimens. In current models, patients may have to contact their clinics to learn or be reminded of how many cryopreserved specimens they have and what the status of each is. Enabling a patient to access his or her cryo-inventory—to read, for instance, a list of specimens on a mobile telephone screen at the clinic in the same way a bank account can be read—represents an enhancement to the patient care experience through the provision of vital information. This has a particular urgency given the central place that cryopreservation has attained in current assisted reproductive treatments where oocyte and embryo freezing is central to the patient’s objectives of fertility preservation and family building, respectively. A model in which patients can access their cryo-inventory details may mitigate the phenomenon of “unclaimed” cryo-embryos [15, 16] and achieve sustained engagement in both their potential future treatment plans and the disposition of their embryos.

An environment in which cryo-inventory data are continuously monitored, updated, and analyzed is one in which AI can be developed and then applied. The opportunities for querying the database from a given laboratory or a network of laboratories can result in a wealth of information about trends in patient care, treatment outcomes, and cryo-repository activities, and can inform future utilities and assets for best practices in the safekeeping of invaluable reproductive cells and tissues.

Potential use and advantages of deep technology in cryostorage

In addition to the anticipated achievements using AI to produce a maximally informed, data-driven ranking of embryos for intra-uterine transfer and all future applications of a cryopreserved embryo or gamete cohort, AI can be integral to new process design.

In conjunction with robotics, AI may be instrumental in making the fully automated “laboratory-on-a-chip” a reality, a model that will enable embryologists to devote their increasingly valuable expertise to the exploration and research that will enhance clinical outcomes, develop new technologies, and attend to the ever-expanding educational, training, and administrative scope of laboratories for human reproduction. One could envision an algorithm-driven, robotic platform integrated with the patient’s electronic medical record and all of the pertinent clinical and scientific data that would allow (a) an electronic order for a frozen embryo transfer to be scheduled; (b) the survey of the cryopreserved, genetically tested cohort; (c) ranking of the appropriate cryopreserved embryos by AI [17–19]; (d) confirmation of embryo selection to the clinical partners for the treatment cycle to go forward; (e) confirmation of the preparation of the warming and culture media for the cycle; and (f) execution of the order by which the selected cryopreserved embryo is retrieved from storage, released from the cryodevice into the warming media with initiated microscopic imaging, and deposit into the culture medium in the transfer culture plate awaiting loading into the transfer catheter. All manual handling and visual reading are eliminated for efficiency, transparency, and safety while attaining an intact chain of custody that is recorded and documented at every step.

Conclusion

While AI, likely complemented with applications of automation and robotics, promises the benefits of informed decisional, ranking, and selective processes, it stands as an accessory to, not replacement, the human interactions and relationships vital to clinical practice. Cryostorage, and the particular importance of specimen and inventory management, represent a need for better technology and tools to manage and safeguard these tissues for our patients. One can hope that when assisted by AI, humans will enjoy more time to afford the invaluable attributes of attention, perception, and support that are central to treatment and all human commerce. Development and utilization of technology that removes opportunities for human error, and subjectivity in decision-making, and improves safety and operational efficiencies will allow us to deliver a better standard of care and gives the promise of improved outcome measures for our patients. Errors in the fertility industry are considered grave and even make the list of serious reportable events (aka SRE or “never events”) by the National Quality Forum [20]—we owe it to our patients to relentlessly develop and adopt new and better ways of helping them achieve their family building dreams.

One of the authors (C.H.) is currently involved in a commercial effort by TMRW Life Sciences, Inc. to realize the elements highlighted in this paper. These operational innovations include mechanical automation to minimize handling of the cryo-inventory, maintain cold chain, and preclude opportunities for temperature excursions. Continuous monitoring of cryostorage conditions alongside the ability to respond to any deviation allows for proactive measures to be taken. In addition, software automation is utilized to assign storage locations and unique identifiers, integrating electronically mediated specimen labeling (barcodes, RFID) for logging and tracking.

Declarations

Conflict of interest

Cynthia Hudson is an employee of TMRW Life Sciences, Inc. Kathryn J. Go, PhD., H.C.L.D., none.