Clinical Evaluation of a New, Miniaturized Biosensor for Self-monitoring of Blood Glucose

^a author for correspondence: fax 408-942-5600 e-mail calberts{at}lfsus.jnj.com

The American Diabetes Association (ADA) has estimated that in 1993 7.8 million people in the United States had type 1 or type 2 diabetes (1) . About 70% of these individuals self-monitor blood glucose, and ~2 million glucose meters were purchased in 1995–1996 (2) . The ADA's recent reduction in the diagnostic threshold for diabetes is expected to increase the number of diagnosed patients (3) . Electrochemical meter systems offer some advantages for self-monitoring: compact size, rapid test time, and small sample volume. In this study, the analytical performance of a prototype of the FastTake(TM) system was evaluated for providing accurate and reproducible blood glucose results.

The FastTake system is based on the enzymatic oxidation of glucose (glucose oxidase) using a mediator (ferricyanide) and electrochemical detection. The system produces rapid (15 s) plasma equivalent results from capillary whole blood samples over a wide range of glucose concentrations (200–6000 mg/L). The test strip contains a carbon-based working electrode, silver/silver chloride reference electrode, and dry reagents. This strip has been optimized to reduce temperature and hematocrit sensitivities, using nonconductive, partially silanized silica fillers, selected to balance the hydrophobicity/hydrophilicity of the reaction environment (4) . The silica filler forms a gelatinous layer, which restricts large sample components such as red blood cells from impairing the reaction at the electrode. The screen-printed reagents and the multilaminar construction of the test strip reduce the blood sample requirement (2.5 µL), compared with most other currently available systems.

One hundred three adults with diabetes completed the Institutional Review Board-approved, 4-week study at either Good Samaritan Hospital, Los Gatos, CA (site 1, 53 participants, sea level) or the International Diabetes Center, Lutheran Medical Center, Wheat Ridge, CO (site 2, 50 participants, 5400 ft elevation). The age of the participants ranged from 18 to 74 years. Forty-four participants had previously been diagnosed with type 1 diabetes, and 59 had been diagnosed with type 2. All participants had prior experience with self-monitoring of blood glucose (average years of monitoring, 7.1).

During each of three site visits, participants tested capillary whole blood with the FastTake system. Small volumes of blood from the same finger puncture(s) were also used to measure hematocrit (STAT-CRIT^®, Wampole Laboratories) and plasma glucose values (YSI 2700 Biochemistry Analyzer, Yellow Springs Instruments). During the study, each participant tested two FastTake test strip lots. The participants presented a wide range of blood glucose concentrations (390–4120 mg/L) and hematocrit concentrations (30–59%).

Over the range of glucose concentrations tested from both sites, FastTake correlated well with plasma YSI reference values (FastTake = 0.984YSI + 13.3 mg/L, n = 284, S_yx = 123 mg/L, r² = 0.98). By linear regression analysis, results for site 1 compared well to those for site 2. For glucose concentrations measured at site 1, FastTake correlated well with plasma glucose values measured by the YSI (FastTake = 0.987YSI - 8.4 mg/L, n = 143, S_yx = 122 mg/L, r² = 0.963). Site 2 demonstrated similar linear regression statistics (FastTake = 0.98YSI + 35.6 mg/L, n = 141, S_yx = 122 mg/L, r² = 0.975). Error grid analysis (5) (Fig. 1 a) was used to show the clinical accuracy of FastTake for both sites. Over the range of glucose tested, 97.9% of the results fell within zone A (defined as "clinically accurate"), and 2.1% fell within zone B (defined as "deviating from the reference method by >20% but would lead to benign or no treatment error"). Individually, the error grid results were very similar for the two sites, with site 1 and site 2 yielding 97.2% and 98.6%, respectively, in Zone A. For both locations, the remaining data points fell in zone B. During the studies, the clinical performance of the two test strip lots was observed to be identical based on error grid analysis (Fig. 1a ). Across the range of observed patient hematocrits (30–59%), the FastTake meter bias relative to the YSI reference was observed to be independent of hematocrit concentration (Fig. 1b ).

View larger version (29K): [in this window] [in a new window]   Figure 1. Results with FastTake meter. (a) Error grid analysis. Zone definitions: A, clinically accurate; B, deviating from the reference method by >20% but would lead to benign or no treatment error; C, deviating from the reference method by >20% and would lead to unnecessary corrective treatment errors; D, potentially dangerous failure to detect and treat blood glucose concentrations outside of desired target range; E, would result in erroneous treatment. (b) Percent bias to reference across hematocrit and glucose ranges. Glucose range (in mg/dL): +, 0–75; , 76–125; , 126–175; , 176–225; , 226–275; , 276–412.

Precision and sample volume testing were conducted at LifeScan to further evaluate the FastTake prototype. Venous whole blood samples were collected from healthy donors in heparinized tubes. Because the product is not intended for use with venous blood samples, to simulate capillary blood, the PO₂ was measured (Ciba Corning Model 278 Blood Gas System, Chiron Diagnostics) and maintained between 60 and 70 mmHg. Precision was evaluated on 2 days using 10 meters and whole blood from two donors. Over the glucose concentrations tested (~500, 800, 1500, 3000, and 5000 mg/L), within-run precision (CV) ranged from 2.7% to 5.3% (n = 20) for day 1 and 2.8% to 4.6% (n = 20) for day 2.

Sample volume testing was also conducted over 2 days by use of five meters and whole blood from two donors. Replicates (n = 3) were tested on each meter at three glucose concentrations (~700, 1800, and 3600 mg/L) with different sample volumes of 2.5, 3, 4, 5, 10, and 25 µL. Testing at 2.5 µL indicated that 95.6% of the data fell within ±20% of the reference method at the three glucose concentrations tested. For blood volumes of 3.0 µL and above, >97.8% of the laboratory data fell within ±20% of the reference.

These studies indicate that the FastTake system provides rapid, accurate, reproducible results in both the laboratory and clinical settings. Although the product is not intended for use with samples containing lower concentrations of oxygen, such as venous blood, the system performance was consistent at sea level and at the lower atmospheric oxygen testing environment encountered in the Denver clinic. System performance in a clinical setting was also found to be consistent throughout the glucose and hematocrit ranges observed.

Footnotes

LifeScan, Inc., Clinical Research M/S 3D, 1000 Gibraltar Dr., Milpitas, CA 95035

References

1. Diabetes Vital Statistics 1996. Alexandria, VA: American Diabetes Association, 1996:13..
2. 1996 Roper Starch Worldwide Diabetes Market Tracking Study. New York: Roper Starch, 1997:3..
3. Sacks D. Implications of the revised criteria for diagnosis and classification of diabetes mellitus [Editorial]. Clin Chem 1997;43:2230-2232. [Free Full Text]
4. McAleer JF, Scott D, Hall G, Alvarez-Icaza M, Plotkin EV. United States patent 5,708,247, 1998..
5. Clarke WL, Cox D, Gonder-Frederick LA, Carter W, Pohl SL. Evaluating clinical accuracy of systems for self-monitoring of blood glucose. Diabetes Care 1987;10:622-628. [Abstract]