Search Results

The SEEKER System, including the SEEKER Instrument and the SEEKER LSD Reagent Kit-IDUA|GAA|GBA|GLA for use on the SEEKER Instrument, is intended for quantitative measurement of the activity of a-L-iduronidase, a-D-glucosidase, Bglucocerebrosidase and a-D-galactosidase A from newborn dried blood spot specimens as an aid in screening newborns for Mucopolysaccharidosis Type I, Pompe, Gaucher and Fabry diseases. Reduced activity of these enzymes may be indicative of these lysosomal storage diseases. The enzymes measured using the SEEKER LSD Reagent Kit-IDUA|GAA|GBA|GLA and their associated lysosomal storage diseases are listed below.

Enzyme (abbreviation)	Disease
α-L-iduronidase (IDUA)	Mucopolysaccharidosis Type I (MPS I)
α-D-glucosidase (GAA)	Pompe
β-glucocerebrosidase (GBA)	Gaucher
α-D-galactosidase A (GLA)	Fabry

Device Description

The SEEKER System employs digital microfluidic technology to measure multiple lysosomal enzymatic activities quantitatively from newborn dried blood spot specimens. The following components are provided:

1. SEEKER Instrument (including USB and power cables), Desktop PC with monitor, keyboard, mouse, and the Spot Logic software.
1. SEEKER LSD Reagent Kit IDUA|GAA|GBA|GLA containing enzyme specific substrate reagents, dried blood spot extraction buffer, reaction stop buffer, 4 levels of calibrators and quality control dried blood spots containing 4 levels of quality control (OC) material. Each Reagent Kit contains sufficient consumables for 1440 tests. The contents of the kit are listed below:

Component	Contents	Quantity
Quality controldried blood spots	QC-Base Pool (QCBP)	1x15 spots
	QC-Low (QCL)	1x15 spots
	QC-Medium (QCM)	1x15 spots
	QC-High (QCH)	1x15 spots
Enzyme substrates	IDUA	9x100 µL
	GAA	9x100 µL
	GBA	9x100 µL
	GLA	9x100 µL
Calibrators	Calibrant A (CALA)	9x50 µL
	Calibrant B (CALB)	9x50 µL
	Calibrant C (CALC)	9x50 µL
	Calibrant D (CALD)	9x50 µL

Other components needed to run tests include the following:

Component	Contents	Quantity
Other Reagents	Extraction Buffer (EXT)	9 x 30 mL
	Filler Fluid	9 x 10 mL
Cartridge	SEEKER cartridge	1 x 36

The composition of the enzyme reagents and buffers are summarized below:

Reagent (Description)	Composition
IDUA(α-L-iduronidase substrate)	2 mM 4-MU-α-L-iduronide sodium salt3mM D-saccharolactone0.04 M acetate buffer, pH 3.520 mM methyl β-cyclodextrin0.01% Tween20
GAA(α-D-glucosidase substrate)	5 mM 4-MU-α-glucopyranoside12 µM acarbose0.04 M acetate buffer, pH 3.820 mM methyl β-cyclodextrin0.01% Tween20
GBA(β-glucocerebrosidase substrate)	16 mM 4-MU-β-glucopyranoside0.05 M/0.1M citrate phosphate buffer, pH 5.20.01% Tween201.5% sodium taurocholate
GLA(α-D-galactosidase A substrate)	10 mM 4-MU-α-galactopyranoside145 μM N-acetyl galactosamine0.04 M acetate buffer, pH 4.620 mM methyl β-cyclodextrin0.01% Tween20
Stop Buffer STB(reaction stopping buffer)	0.6 M NaHCO3, pH 11.0 in 0.04% Tween 20
Extraction Buffer EXT(dried blood spot extraction buffer)	0.1% Tween 20 in water
Filler Fluid(medium for droplet movement)	0.1% Triton X-15 in 5cSt silicone oil

The Seeker Calibrators are supplied as part of the SEEKER LSD Reagent Kit -IDUA|GAA|GBA|GLA. The calibrators consist of 4 levels of aqueous preparation of 4methylumbelliferone sodium salt (4-MU) in 0.6M sodium bicarbonate buffer, pH 11.0 with 0.01% Tween 20. The concentration of 4-MU in each of the 4 calibrators is indicated in the table below:

Calibrator Level	Concentration of 4-MU
Calibrant A (CAL A)	0.0375 µM
Calibrant B (CAL B)	0.0750 µM
Calibrant C (CAL C)	0.1500 µM
Calibrant D (CAL D)	0.3000 µM

The quality control dried blood spots include 4 levels of control material: OC Low (QCL), QC Medium (QCM) and QC High (QCH). The composition of the 4 quality control dried blood spot (DBS) is summarized below. QCBP is used to fill empty wells on a cartridge.

Quality Control Level	Composition
QCBP	Heat inactivated human serum, adjusted to ~50% hematocritusing human red blood cells
QCL	5% cord blood and 95% heat inactivated serum, adjusted to~50% hematocrit using human red blood cells
QCM	50% cord blood and 50% heat inactivated serum adjusted to~50% hematocrit using human red blood cells
QCH	Human umbilical cord blood, adjusted to ~50% hematocritusing human red blood cells

The enzymatic activity values for the quality control DBS measured by the manufacturer are given on the lot specific quality control certificate included in each assay kit for all levels except QCBP. Each laboratory should establish its own mean and acceptable ranges for the quality control materials.

3. SEEKER Cartridges

1. Finnipipette Novus 8-channel automatic pipette 1-10 uL
1. Finnipipette Novus 1-channel automatic pipette 10-100 uL

AI/ML Overview

The SEEKER System is a device designed for quantitative measurement of the activity of α-L-iduronidase (IDUA), α-D-glucosidase (GAA), β-glucocerebrosidase (GBA), and α-D-galactosidase A (GLA) from newborn dried blood spot specimens. This serves as an aid in screening newborns for Mucopolysaccharidosis Type I (MPS I), Pompe, Gaucher, and Fabry diseases.

Here's an analysis of the acceptance criteria and the study that proves the device meets those criteria:

1. Acceptance Criteria and Reported Device Performance

The acceptance criteria for the SEEKER System are implicitly defined by the analytical and clinical performance standards required for its De Novo classification as a Class II device with special controls. These controls mandate thorough demonstration of performance characteristics, including clinical validity, reference intervals, carry-over, detection limits, and imprecision. The clinical validity is primarily demonstrated by the false positive and false negative rates observed in the pivotal clinical study.

Table: Acceptance Criteria (Implied by Regulatory Requirements) and Reported Device Performance

Performance Characteristic	Acceptance Criteria (Implied)	Reported Device Performance (Pivotal Phase Study)	Notes
Clinical Validity (False Positive Rate)	Acceptable false positive rates to minimize unnecessary confirmatory testing and emotional burden.	IDUA: 0.035%GAA: 0.092%GBA: 0.047%GLA: 0.097%	These rates are considered acceptable, particularly given the benefits of early detection and the fact that newborns with "presumed normal" results after risk assessment were not followed up, potentially leading to an overestimation of the false positive rate.
Clinical Validity (False Negative Rate)	Acceptable false negative rates to ensure affected newborns are detected and receive timely therapy.	IDUA: 0%GAA: 0%GBA: 0%GLA: 0% (Note: 2 newborns that would have been FN at 7.0µmol/L/hr cutoff were TP at 8.0µmol/L/hr cutoff. Labeling specifies known limitations for female Fabry and late-onset Pompe).	The reported 0% false negative rate for IDUA, GAA, and GBA is based on active surveillance and the assumption that early-onset cases would be reported. However, some limitations are disclosed in the labeling, acknowledging potential false negatives due to clinical variability (e.g., late-onset Pompe, female Fabry patients). The adjustment of cutoffs over time also indicates complexities in achieving a perfect 0% FN rate. The note about GLA false negatives suggests that the 0% reported might be dependent on the specific cutoff used at the time.
Imprecision (Reproducibility)	Reproducibility across different instruments, reagent lots, and days, within acceptable limits. (No explicit %CV target stated, but general medical device standards suggest low CVs for reliable quantitative assays).	IDUA: 14.2% - 28.5% CV (depending on concentration)GAA: 12.0% - 17.0% CVGBA: 15.7% - 38.0% CVGLA: 9.4% - 16.3% CV	These CVs are generally considered acceptable for screening assays, especially at lower concentrations where biological variability or technical limitations might be higher. The CLSI EP5-A3 guideline was followed.
Linearity/Reportable Range	Linear response over the intended measurement range, covering normal and disease-associated values.	IDUA: 2.77 to 50.75 µmol/L/hGAA: 2.18 to 94.66 µmol/L/hGBA: 2.14 to 73.24 µmol/L/hGLA: 4.88 to 153.74 µmol/L/h. Max deviation from linearity <10%.	Acceptable linearity was demonstrated. Outliers were noted but did not impact overall linearity claims, though their presence highlights the importance of robust outlier handling in routine testing.
Detection Limits (LoQ)	Limit of Quantitation (LoQ) should be below or at relevant clinical cutoffs to ensure accurate measurement in the critical low activity range.	IDUA: LoQ = 2.77 µmol/L/h (some clinical cutoffs were below this, e.g., High Risk 1.5 µmol/L/h)GAA: LoQ = 2.18 µmol/L/hGBA: LoQ = 1.85 µmol/L/hGLA: LoQ = 4.88 µmol/L/h (some clinical cutoffs were below this, e.g., 14+ days High Risk 3.0 µmol/L/h)	The LoQ values are provided. A significant limitation noted is that some clinical cutoffs for IDUA and GLA were below the LoQ, meaning that results in this critical range might not be reliably quantifiable, introducing an element of uncertainty in diagnosis for values between LoQ and these lower cutoffs. This is explicitly recognized as a limitation in the study findings.
Analytical Specificity	Minimal interference from common endogenous and exogenous substances.	Most common interferents (bilirubin, galactose, glucose, EDTA, heparin, intralipid) showed <+10% bias. IDUA specific interferences: total protein at >88 mg/mL and hematocrit at 66% caused significant negative bias; heparin increased imprecision.	Generally good, but specific interferences identified for IDUA (high protein, high hematocrit, heparin) are crucial for laboratories to consider. These are included in the package insert.
Carry-over	Demonstrate minimal or no significant impact from carry-over.	Most assays showed minimal bias (<10%). GLA showed up to 27% bias in one specific condition, leading to a recommendation for retesting.	Carry-over was identified for GLA under specific conditions (high GLA sample preceding a borderline sample), leading to a specific retesting recommendation in the package insert. This directly addresses the identified risk.

The studies described predominantly cover the analytical performance requirements (reproducibility, linearity, detection limits, analytical specificity, carry-over) and the clinical performance (false positive/negative rates).

2. Sample Size Used for the Test Set and Data Provenance

The primary clinical effectiveness study (pivotal phase) involved screening 105,599 newborns.

Data Provenance: The study was performed over 24 months at a public state health laboratory that performs newborn screening. This implies the data were collected in the United States. The study was prospective in nature, as it involved actively screening newborns using the device over a defined period (pivotal phase 18 months). A preliminary study, prior to the pilot/pivotal studies, used approximately 13,000 presumed normal de-identified specimens and 29 known affected specimens, suggesting a mix of retrospective (de-identified) and potentially prospective collection for the known affected cases.

3. Number of Experts Used to Establish the Ground Truth for the Test Set and Their Qualifications

The document does not explicitly state the "number of experts" or their specific "qualifications" involved in establishing the ground truth for the clinical test set in a centralized, adjudicatory manner (e.g., as consensus readers for imaging). Instead, the ground truth was established through confirmatory diagnostic methods and clinical follow-up within the standard newborn screening program.

For patients referred for follow-up:

"Babies considered to still be at high risk for the screened condition after the risk assessment were referred for confirmatory testing."
"Confirmed positive specimens must have a diagnosis based on confirmatory diagnostic methods or clinically meaningful information regarding the status of the subject must be obtained."

The "risk assessment" itself was performed by the laboratory, and would involve experienced personnel (e.g., laboratory directors, genetic counselors, metabolic specialists). While not explicitly listed as "experts establishing ground truth," these individuals make the critical decisions leading to confirmatory diagnosis. The state public health laboratory also had an active surveillance program to track false negative results, which likely involves experts in metabolic diseases.

4. Adjudication Method for the Test Set

The adjudication method for the clinical test set was an elaborate, multi-step process ingrained in the newborn screening workflow of the state health laboratory:

Initial Screening: DBS cards were tested in singlicate.
Retesting for Borderline Results: DBS with at least one enzyme below the borderline cutoff were retested in duplicate.
Visual Outlier Evaluation: The laboratory evaluated retest results for "visual outliers." If found, additional testing was performed.
Average Calculation: The average of valid retest results (excluding visual outliers) was calculated.
Risk Assessment: If the average was below the high-risk cutoff, a risk assessment was performed by the laboratory. This assessment included review of other newborn screening enzyme results, sample quality, transfusion status, health status, transit time, gestational age, family history, and age at sample collection.
Referral for Confirmatory Testing: Babies still considered at high risk after the risk assessment were referred for confirmatory diagnostic methods.
Presumed Normal: Babies with an average above the high-risk cutoff after retest, or those not referred after risk assessment, were presumed normal.

This is a complex, stepwise adjudication process, rather than a simple 2+1 or 3+1 consensus. It involves a combination of algorithmic cutoffs, repeat testing, visual inspection, and comprehensive clinical review.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study was done

No, an MRMC comparative effectiveness study was not performed for the SEEKER System. This device is a quantitative enzymatic activity assay for newborn screening, not an imaging-based diagnostic aid or an AI system intended to be used directly by human readers in an interpretive capacity. Therefore, the concept of "how much human readers improve with AI vs without AI assistance" does not apply in the context of this device. The SEEKER System performs the assay autonomously using digital microfluidic technology.

6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) was done

Yes, the primary function of the SEEKER System is standalone (algorithm only) in measuring enzyme activities. The results are reported as quantitative values (micromoles of 4-MU produced / liter of blood / hour of incubation).

However, the "human-in-the-loop" component comes into play with the interpretation of these quantitative results within the newborn screening program's workflow:

The device generates raw data, which is then processed by its software (Spot Logic).
The laboratory applies specific cutoff values (high risk, borderline), which were adjusted and refined during the study period by human experts based on observed performance and factors like age.
Human judgment is involved in the "risk assessment" process for borderline cases.
Human decision-making leads to referral for confirmatory testing.

So, while the initial measurement is standalone, the screening decision (positive/negative/retest/referral) is a combination of the device's quantitative output and a multi-step human-defined protocol.

7. The Type of Ground Truth Used

The ground truth used for determining true positives and true negatives in the clinical study was primarily based on confirmatory diagnostic methods and clinical follow-up/outcomes data.

For positive cases, the "True Positives" are those confirmed by "other confirmatory diagnostic methods." This implies biochemical, genetic, or clinical diagnoses consistent with the respective lysosomal storage diseases.
For negative cases, the "Presumptive False Positives" are those referred but not confirmed, or those deemed normal after risk assessment. The absence of reported false negatives through the active surveillance program and the assumption that early-onset cases would be clinically identified also contribute to the ground truth for true negatives (i.e., those who screened negative and did not later develop symptoms or receive a diagnosis).

8. The Sample Size for the Training Set

The document does not explicitly delineate a "training set" in the context of typical machine learning or AI models with a distinct number of samples. This device is an enzymatic assay system, not an algorithm that undergoes iterative training on a dataset.

However, the "cutoffs" used by the laboratory for interpreting the results can be thought of as parameters that were "tuned" or "trained" based on observational data:

"A preliminary study was performed by the public state health laboratory prior to the pilot/pivotal studies to establish initial cutoff values. Approximately 13,000 presumed normal de-identified specimens and 29 known affected specimens were analyzed using the proposed device."
"The cutoff values were adjusted during the course of the study in order to reduce the false negative and false positive rate and also to take into account seasonal changes..."

These initial 13,029 specimens and the ongoing refinement using data from the 48,813 newborns in the pilot phase could be considered analogous to a "training" or "development" dataset for establishing the operational parameters (cutoffs) of the screening algorithm.

9. How the Ground Truth for the Training Set was Established

For the "training" data (preliminary study and pilot phase used for cutoff adjustments):

Known affected specimens (29 samples): These likely had their diagnosis confirmed by established diagnostic methods (e.g., enzyme assays, genetic testing) or clinical presentation from relevant patient populations prior to being used in the preliminary study.
Presumed normal de-identified specimens (13,000 samples): These were likely samples from healthy newborns collected and de-identified, implicitly considered "normal" based on the absence of a disease diagnosis through routine clinical channels.
Ongoing Adjustment: During the pilot and pivotal phases, the clinical outcomes (confirmatory diagnoses, absence of symptoms) informed the continuous adjustment of cutoffs, acting as a feedback mechanism for refining the test's interpretive ground truth. The "active surveillance program" for false negatives also served to validate the accuracy of the negative screening decisions.

Ask a Question

Ask a specific question about this device

Page 1 of 1