Tiny Bodies, Big Needs: Prospective Biobanking of Neonatal Clinical Remnant Samples

Abstract

Repurposing biological samples collected for required diagnostic purposes into suitable biobanking projects is a particularly useful method for enabling research in vulnerable populations. This approach is especially appropriate for the neonate in the neonatal intensive care unit (NICU), where blood volume reductions can quickly increase beyond minimal risk for adverse events, such as iatrogenic anemia, and proxy consent provided by parents or guardians is required. The method described in this study provides a framework to prospectively collect and store blood-derived clinical samples after all clinical and regulatory requirements are fulfilled. The consent approach incorporated a 30-day window to allow parents and guardians ample consideration time with follow-up involvement with NICU embedded study team members. The study enrolled 875 participants over a 3-year period. This established a critically needed biobank to support investigator-initiated research with explicit study aims requiring samples at defined day of life frequencies within the NICU and created a normative control reference bank for case comparisons for premature and full-term neonates with brain injury.

Introduction

Critically ill neonates are challenging in every sense of the word. Despite considerable improvement in neonatal outcomes overall, clinical and translational research and the pursuit of predictive biomarkers for this population are often stymied by lack of access to high-quality, prospectively collected samples. This is particularly problematic for very low-birth-weight (VLBW) infants, where intraventricular hemorrhage, the most common brain injury in this population, occurs in the first 36 hours of life. Therefore, sample collection at birth to understand disease mechanisms and developing methods of earlier diagnosis and intervention become critical for improved long-term outcomes and new treatment modalites.¹ Sample procurement in this setting is a multifactorial and complex problem.

Notwithstanding the dim economic realities of pediatric research (underfunded or low-resourced infrastructure), the ethical dilemma surrounding the appropriate times to approach overburdened and anxious parents to obtain consent is complicated by the need to obtain research samples as soon as possible to establish baseline measures before, during, and after important clinical events. Prior research has shown that parents in the neonatal intensive care unit (NICU) prefer the choice to participate in research studies even if it adds some measure of stress, as long as the health care and research team provide information and support.² This type of nonpaternalistic preference has support across diverse ethnic and socioeconomic strata for research with minimal to low added risk.³ An important determining factor to the success of the informed consent process requires diligent communication and the establishment of trust between the medical team and parents, wherein “effective interactions,” if possible, become essential.⁴ This concept establishes the focus of the research team to engage in a long-term relationship with parents beyond the initial consent approach that provides a long consideration phase and the opportunity to ask questions or otherwise interact with embedded clinical research nurses (CRNs) in the NICU.

Establishing a larger window for informed consent and guardian deliberation, however, compounds the challenges associated with collecting samples representative of disease onset that do not suffer from quality degradation. The narrow and urgent collection window for samples from the 1st hours and days of life must be considered with greater prioritization given to clinical needs over research use. Maximum 24-hour blood draw guidelines from the Institutional Review Board (IRB) suggest a 3 mL/kg limit (3.8% total blood volume) postneonatally will impose minimal risks to a pediatric subject.⁵ This type of limitation for neonates, after accounting for blood transfer losses and separation of plasma/sera from whole blood with an assumed hematocrit value of 50%, means it is likely only one milliliter per kg may be available for all intended clinical and research purposes. Utilization of clinical remnant samples in minimal-risk research is thus the only avenue available for investigators. Infrastructure required to implement an effective collection strategy must account for preanalytical variations in clinical processing, sample misidentification if using an honest broker de-identification system (a particular challenge in the NICU where patient names may change one or more times), and post-testing storage practices before biorepository procurement.

The need for better sample strategies, optimized for tiny participants, is apparent. A robust and efficient protocol to prospectively target and procure residual clinical samples from NICU patients will be highly site specific; however, the method presented in this study provides a configurable framework for other institutions to modify depending on availability of clinical and research resources. This interdisciplinary approach begins with the screening and enrollment of potential participants within 30 days of birth by research nurses embedded in the NICU and describes in detail the necessary steps to identify and procure remnant samples after completion of all clinical testing, yet before sample quality starts to deteriorate. This methodology was implemented over a 3-year enrollment window, resulting in the biobanking of nearly 8000 samples from 875 participants, which otherwise would have gone unused.

Methods

IRB review/approval and informed consent

The described activities received IRB approval from the Johns Hopkins Medicine IRB before any study activity. While both studies were considered minimal risk by only utilizing remnant blood samples, and thus could have qualified for a waiver of informed consent, the IRB committee felt it important to give parents the opportunity to determine their willingness to allow their child to participate. However, this meant remnant diagnostic samples would be discarded by clinical laboratory before the study team had an opportunity to reach the parent to obtain parental permission. Therefore, the investigators, biorepository scientific leadership, and clinical laboratory leadership collaboratively developed a plan that allowed remnant samples to be procured from the clinical laboratory following the completion of all diagnostic testing, temporarily moved to the biorepository, and held for temporary storage in appropriate conditions to protect sample quality and integrity until obtaining parental permission. The study team then had 30 days to approach the parent for informed consent before samples reach their ultimate destination (banked or destroyed).

Research coordinator role: screening log

The investigators assembled a study team comprising a clinical research coordinator (CRC), CRN, and research assistant (RA). The CRC was responsible for overall IRB and regulatory documentation while the CRN and RA were integral to day-to-day study execution. Infant admissions to the 97-bed NICU were prescreened each workday by the CRN/RA and the patient's name was added to a screening log that was accessible by the biorepository.

Infants meeting inclusion criteria included those born 23 to 42 weeks with severe neurological pathology such as hypoxic-ischemic encephalopathy requiring whole body hypothermia therapy, brain stroke, and seizure disorder, as well as all VLBW infants (<1500 g). Gestational age-matched controls included infants not meeting the case criteria. Infants with chromosomal abnormalities, genetic syndromes and major congenital malformations were excluded. A total of 3090 participants were prescreened for eligibility. Parents of eligible infants were approached for written parental permission. Due to the nature of visitations within the NICU, this usually occurred over the course of a few visits. The patient's disposition was updated by the CRN/RA on the shared screening log, which alerted the biorepository staff as to which patient samples should be formally accessioned and banked (parental permission granted) and which should be destroyed (parents declined or permission not otherwise obtained).

Biorepository staff role

Clinical chemistry samples, once testing is completed, are archived in the Johns Hopkins All Children's Pathology and Clinical Laboratory Department by sectional accession numbers assigned sequentially by the laboratory's laboratory information system (LIS), based only on the timing of the order drop that day. Sample types (serum, plasma, urine, and cerebrospinal fluid) are not considered in the ordering of accession numbers, but may be discerned by order type. This accession number resets each day. Samples are archived at 4°C with a patient label containing at least three identifiers in the order of this accession number, and there are typically between 300 to 400 unique accession numbers generated each day in the chemistry section for all outpatient and inpatient laboratory services. Furthermore, policies supporting the College of American Pathologists accreditation standards require clinical samples be retained for 48 hours after all testing is completed to allow for re-testing, additional testing, or other required validations necessitated by clinical guidelines.

Biorepository staff generated a sample list daily (Monday to Friday) using an LIS query that selected for the following: chemistry section accession identifiers (C, CZ, and CF), completed status, collection time 48 hours prior, and a, NICU patient location (including subunit locations A, B, and C). This query ran several hours after physicians had completed rounds to allow for any additional add-on testing as an added safeguard. It is important to note this list provides all samples originating from the NICU, including study-ineligible or non-consented individuals, and contains protected health information, including name, medical record number (MRN), financial account numbers, the corresponding specimen accession number, and processing-related metadata (i.e., collection date and time and collection vacutainer). This list does not provide test results. This LIS list was cross-referenced against the screening log completed by the study coordinators to identify only the appropriate samples from participants who provided consent and identified using names and MRNs, as well as those potentially eligible for the study, who had not yet provided consent. The prevalence of non-singlet neonates with common naming conventions (i.e., Baby Girl Smith) requires multiple identifiers to prevent misidentification from occurring. Consented patients were assigned a unique research identifier and provided to the biorepository using the screening log described previously. Participants ineligible for the study or those who had opted out were removed from the list, and samples from these patients were disregarded. Identified samples were procured from clinical chemistry refrigerators using the unique specimen accession number for that date and moved to the Biorepository processing area for de-identification and storage in research-only areas directly adjacent to clinical chemistry.

Each sample was immediately transferred to a 1.4 mL low-retention polypropylene cryovial (Micronic, Lelystad, Netherlands) identified with a unique, 10-digit 2D barcode. The barcode was scanned into the daily LIS list spreadsheet next to the sample accession information and any sample quality issue was noted. Hemolysis grades were estimated using a comparison to a visual standard, using grade 0 for minimal hemoglobin (hgb) present and grade 1 (∼50 mg/dL), grade 2 (∼100–250 mg/dL), and grade 3 (∼500–1000 mg/dL) for prefreezing characterization to help inform downstream analytical utility. The type of vacutainer used in the primary collection was also annotated to determine if the volume was derived as serum or plasma. Samples were transferred into the aliquot tubes by inversion and poured to decrease volumetric loss sometimes observed using pipettes from gel separator vacutainers. Aliquot tubes were then frozen on dry ice for all further handling.

Samples were then separated into two categories—consent obtained or consent pending. If consent was already obtained, samples were immediately received into the biorepository LIS, labeled with their unique research participant identifier, and stored in long-term −80°C storage facilities for future distribution. Frozen sample volumes were estimated using 50 μL visual reference standards. Samples waiting on consent were not received into the biorepository system, and stored separately in a triage location at −80°C. These samples were checked daily for consent changes and disposed immediately if consent was not obtained before discharge or if the patients' guardians did not want to participate in the study.

Results

Study enrollment occurred over a 3-year period. In this time, the study team enrolled 810 participants into the control arm and 65 participants into the case arm of the study. This represents an enrollment rate of 28.3% (875 consented/3090 total pre-screened). This enrollment calculation includes patients in the NICU who did not meet study criteria and were prescreen failures, were discharged before securing consent, or were otherwise ineligible for the study. This approach provides a more complete metric for extrapolation to other centers. The actual consent rate is 86.8% (875 provided consent/1008 approached). The average time to obtain consent is 10.1 days from the date of admission. In total, 19,126 samples were collected over the course of this study, although only 7964 samples were consented for research use. Biorepository costs totaled $99,516 and the CRC, CRN, and RA costs totaled $66,863, which breaks down to $20.89 total costs per usable sample. At the time of this writing, 3495 of these samples were distributed for analysis using enzyme-linked immunosorbent and Slow Off-rate Modified Aptamer (SomaLogic, Boulder, CO) assays, reflecting a 43.6% utilization rate expected to increase with future use.

Sample characteristics

Samples showed similar characteristics (Table 1), although cases tended to have larger median volumes (300 μL) compared to controls (200 μL). It is unclear if this might be a reflection of laboratory ordering practices for patients with brain injury or an unrecognized tendency for slightly higher collection volumes to prevent extra blood collections if additional orders are anticipated (Fig. 1).

Table 1.

Sample Characteristics Across Groups

	Case	Control
Samples, n	720	7244
Volume, mean (μL)	326	276
Median (μL)	300	200
Hemolysis, %	6.39	1.58
Mild (%)	4.16	1.06
Marked (%)	2.08	0.42
Gross (%)	0.15	0.10

Cases were defined as infants meeting inclusion criteria, those born 23 to 42 weeks with hypoxic-ischemic encephalopathy admitted for whole-body hypothermia therapy, strokes, seizures, severe hypotonia, and all very low-birth-weight infants (<1500 g). Gestational age-matched controls included infants not meeting the case criteria. Infants with chromosomal abnormalities, genetic syndromes, and major congenital malformations were excluded.

FIG. 1.

Violin plots demonstrate sample volume (μL) distributions for control and case participants. This type of plot shows the range of sample volumes for both plasma and serum, as well as the frequency of measurements at particular volumes. Small volume tendencies are consistent in both participant groups for both plasma and serum derivatives.

The distribution of sample collections across the first 7 days of life was also assessed (Fig. 2). Day 0 was determined using the day of birth with a matching sample collection date, with subsequent dates calculated from that starting point.

FIG. 2.

Day of life (0–7) sample collection frequencies overall for control (left) and case (right) participants.

Discussion

This biobank was built to support investigator-initiated research with explicit study aims requiring prospective collection of samples from specific patient cohorts at defined day of life frequencies within the NICU and to create a normative control reference sample bank for case comparisons for premature and full-term neonates with brain injury.^6–8 This need justified the costs required to create and sustain a systematic biobanking approach focused on time-sensitive sample procurement and which provided a large enrollment window to approach potential participant guardians for informed consent.

Metrics from this analysis establish a benchmark of efficiency and cost associated with procurement of clinical chemistry remnant samples derived from blood, and provide important insights into phlebotomy practices and clinical laboratory requirements relating to the frequency and volume of blood sampling in NICU patients. These data further expand on previous literature demonstrating the feasibility and utility of “scavenging” clinical samples from preterm infants,^9,10 specifically by enhancing the parent-research team connection and improving sample stability during transfer of custody from the clinical laboratory to the biorepository. Moreover, our general practice to embed CRCs in the NICU and biorepository staff in the clinical laboratory as shared hospital resources establishes a permanent system more conducive to research in general with wider access not normally available to investigators and study team members operating in academic siloes or those limited to single-study support. Institutions that may not be able to adopt this infrastructure should consider a hybrid approach for certain situations where parents or guardians are difficult to reach or visit sporadically, wherein researchers recruit clinical staff, such as health unit coordinators or nurses, to act as liaisons that contact the research team when parents are at the bedside. On the laboratory side, it is relatively rare for research teams to directly access a clinical LIMS system; however, these types of systems should have mechanisms for generating similar reports without much effort by the clinical support team, which can then be appropriately disseminated to research staff acting as honest brokers (i.e., someone approved at the institution to de-identify samples and data).

In addition, calculation of per-sample costs at $20.89, which includes the costs to enroll, collect data, and the procurement of samples for de-identification (including processing and consumable costs), storage, and distribution, may be useful for future research planning using similar strategies. It is unclear at this time if the study enrollment rate (22.3%) is more or less optimal, as effective comparisons to equivalent studies are lacking, and may be community or regionally distinct. The enrollment rate includes prescreen failures to provide insights into enrollment performance characteristics in a 97-bed NICU and to serve as an indicator of samples that may be procured temporarily without realized research utilization. It is highly likely this rate may also be a reflection of our delayed consenting approach, as parents of potential participants discharged from the hospital before consent may have opted to participate in the study if approached in time; however, the enrollment rate was not deemed detrimental to the success of the study. Indeed, the actual consent rate of 86.8% is much higher, reflecting a high level of enthusiastic participation in this population.

Sample quality issues were not a major issue in this study; however, investigators seeking to utilize this model should conduct validation studies, if possible, for downstream biomarkers of interest that may be unstable or more labile in nature where refrigeration with delayed freezing (>48 hours) is unavoidable. This includes special consideration for clinical trials wherein outcomes are more acute, requiring more frequent or more focused hourly collections as blood volumes allow (i.e., pharmacokinetic studies). Other considerations affecting sample quality not captured in this study include delayed freezing times for samples handled over the weekend, as well as the frequency of samples consumed from intended clinical testing with no measurable volumes remaining for research.

Disclaimer

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Author Disclosure Statement

No conflicting financial interests exist.

Funding Information

Research reported in this publication was supported by the generous support of the Johns Hopkins All Children's Foundation Institutional Research Grant Program and by the Eunice Kennedy Shriver National Institute of Child Health and Human Development of the National Institutes of Health under Award Number R01HD086058 (A.D.E., S.B.).