Abstract
Adequacy of insulin concentration in commercially available insulin formulations has recently been challenged. We therefore repeatedly evaluated insulin content and stability of 58 insulin vials containing 5 different insulin formulations (human insulin, standard/faster-acting insulin aspart, insulin lispro, and insulin glargine) over a period of 85 days. High-resolution mass spectrometry was used to quantify intact monomeric insulin in glass vials and plastic pump cartridges exposed to three different temperatures (4°C, 22°C, 37°C), simulating real-life conditions. In all cases, measured insulin concentration was in accordance with FDA and European Medicines Agency (EMA) requirements without evidence of chemical instability.
Background
To ensure optimal treatment and patient safety, rigorous quality control and assurance of therapeutics is of utmost importance. According to FDA requirements, each manufacturer must guarantee a concentration of 95–105 IU/mL of intact insulin in U-100 formulations. Manufacturers recommend storing unopened vials at 2°C–8°C and keep used vials for up to 28 days, while avoiding exposure to extreme temperatures and sunlight.
A recently published study by Carter and Heinemann examined 18 randomly purchased vials of NPH and regular insulin using liquid chromatography-mass spectrometry (LC-MS) and found mean concentrations of 40.2 IU/mL with levels ranging from 13.9 to 94.2 IU/mL. 1 The authors attributed the findings to inappropriate handling along the distribution chain. Their report raised concerns among people with diabetes and health care providers, and controversy among manufacturers, clinicians, and biochemists. Follow-up studies using nuclear magnetic resonance (NMR) spectroscopy and high-pressure liquid chromatography (HPLC) demonstrated that insulin content was maintained along the supply chain, thereby complying with FDA and EMA requirements. 2,3
Determination of insulin concentration is usually performed using HPLC, the FDA accepted standard method. While HPLC is highly accurate, analysis times are usually more than 15 min and therefore limit throughput. 4 Alternatively, mass spectrometry (MS) is very well suited for peptide analysis, 5 and there are numerous accounts of LC-MS being a highly sensitive and specific for detection of insulin and its analogs. 6,7 In an attempt to challenge the findings of Carter and Heinemann, we set out to develop a high-throughput (1 min/sample) high-resolution MS method to quantify insulin content and stability in vials.
Methods
Collection of insulin formulations
A total of 58 U-100 vials comprising insulin aspart (Novorapid®), faster-acting insulin aspart (Fiasp®), human insulin (Actrapid®), (all Novo Nordisk, Bagsvaerd, Denmark), insulin glargine (Lantus®; Sanofi-Aventis, Paris, France), and insulin lispro (Humalog®; Eli Lilly, Indianapolis, IN) were purchased from the hospital pharmacy (unopened; total of 31) or were obtained from wards of University Hospital Bern or patients (used; total of 27). All vials were stored according to the specifications of the manufacturer until measurement. DANA R pump cartridges (Diabecare, Seoul, South Korea) were filled with Fiasp, Novorapid, and Humalog (unopened; in triplicate). Standards for human insulin, bovine insulin, lispro, and glargine were obtained from Sigma-Aldrich (Buchs, Switzerland) and aspart from LGC Standards (Teddington, United Kingdom). Hydrochloric acid and acetonitrile were purchased form Merck (Zug, Switzerland) and formic acid from Thermo Fisher Scientific (Rheinach, Switzerland). All reagents were used in the highest obtainable quality.
To assess potential changes in insulin content and stability over time, formulations were sampled from vials and cartridges on days 1, 2, 3, 4, 5, 8, 9, 10, 11, 12, 15, 16, 17, 54, and 85 after receipt from the hospital pharmacy upon storage at either 4°C, 22°C, and 37°C in temperature-controlled environments with electronic temperature monitoring (either fridge or incubators).
Determination of insulin content
Mass spectrometric measurements were performed on a Q Exactive Orbitrap (Thermo Fisher Scientific) equipped with an electrospray ionization source. Measurements were performed in the range m/z 800–1500 at R = 70,000 in positive ion mode. Samples were injected using the autosampler and pump modules of a Vanquish system (Thermo Fisher Scientific).
As we only analyzed drug formulations of minor complexity and high concentrations in this study, we decided to forego the HPLC step and directly analyze the vial contents by MS for increased throughput. Such flow injection setups, which omit a preseparation of analytes, are widely used in metabolomics and lipidomics. 8,9
Reference compounds were dissolved in 0.01 M hydrochloric acid and diluted using 33% acetonitrile in water to desired concentrations. External and internal calibration was performed using a six-point calibration curve between 0.2 and 54 μmol/L. Internal standard was added to all insulin therapeutics, followed by 50-fold dilution using 33% acetonitrile to generate monomeric insulin, resulting in a theoretical insulin concentration of 12 μmol/L in polypropylene 96-well plates (Eppendorf, Switzerland). Samples were then stored in the autosampler at 10°C and measured within 1 h after dilution. All samples and calibrants contained bovine insulin at a final concentration of 2 μmol/L as an internal standard.
Statistical analyses
Change in insulin content was determined using linear regression. A significant decline in concentration was considered a slope of the regression line significantly different from zero, evidenced by P < 0.05. Statistical analysis was performed using R (R Foundation for Statistical Computing, Vienna, Austria) and GraphPad Prism (GraphPad Software, La Jolla, CA).
Results
For all compounds in the present analysis, the quantification of monomeric insulin using our mass spectrometric method showed linear response over the entire calibration range (R 2 ranging from 0.9948 to 0.9988), accuracy of between 96% and 104%, and precision between 2.4% and 3.8% (Supplementary Table S1).
The measured insulin content of the 58 vials comprising 5 different drug formulations fell in the required concentration range of within 5% of the nominal value. No systematic difference between unopened vials (obtained from the pharmacy) and already used vials (obtained from hospital wards and patients) was observed (see Supplementary Figure S4). Likewise, the time-course experiment, during which the insulin content of the different formulations stored at different temperatures (4°C, 22°C, or 37°C) was investigated, yielded results according to specification. An example of such a longitudinal assessment at different temperatures is shown in Supplementary Figure S1 for Humalog. The statistical evaluation of the long-term content assessment for all compounds stored in vials is summarized in Table 1.
Concentration Assessment of Insulin Formulations Stored in Original Glass Vials (Novorapid, Fiasp, Humalog, Actrapid, and Lantus) and Cartridges (Novorapid, Fiasp, and Humalog) at Three Different Temperatures
Mean is average concentration over all measurements (15 time points over 86 days). CV = coefficient of variations over all measurements. P = P-value of the regression line (concentration vs. time). A P-value >0.05 indicates no change in concentration over time.
Screening for degradation products throughout the study turned out negative. An example of a theoretical degradation product resulting from deamidation is illustrated in Supplementary Figure S2.
Discussion
The present study repeatedly determined the insulin content and chemical stability of 58 insulin vials containing 5 different analogs over a period of almost three months. During the study period, glass vials and plastic cartridges were exposed to three different temperatures (4°C, 22°C, and 37°C), simulating real-life conditions. Despite stressed conditions (storage duration and temperature exposure beyond manufacturer's recommendations) the measured insulin content was in accordance with FDA and EMA requirements, and we did not find any evidence of chemical instability.
Despite the use of a similar analytical approach, our findings are inconsistent with the results obtained by Carter and Heinemann. 1 Conversely, our data are in line with results by Moses et al. who reported insulin concentrations within the specified range for hundreds of different vials using HPLC, which is currently the recognized FDA standard. 3 Even though our MS assay exhibits similar accuracy when compared to HPLC (standard deviations ranging from 2.4% to 6.7% for the HPLC method and 3.9%–5.1% for the MS method as evidenced in Supplementary Fig. S1), the sample-to-sample time of the MS assay is 1 min compared with at least 15 min for HPLC (Supplementary Fig. S3). Consequently, the MS assay may have potential for large-scale application.
Our results further support the suggestion brought forward by others 2,3 that there were potential issues with the analytical method used by Heinemann and Carter. We can only speculate on the reasons for the discrepancies, which could include steps during sample dilution (precluded in our case by using an internal standard) or using a low-resolution mass spectrometer. Our results also conform with recent findings obtained by NMR spectroscopy. 2 When compared with NMR and HPLC, MS confers the benefit of offering automation and high-throughput analysis. Yet, all discussed analytical techniques (HPLC, MS, NMR) suffer from the same limitation: biological activity, which could potentially be impacted through changes in the tertiary protein structure or protein aggregation upon storage, cannot be assessed. Determination of actual biologic insulin activity requires use of specialized assays to monitor insulin receptor activity and binding or the evaluation of glycemic effects in animal models or humans using the hyperinsulinemic euglycemic clamp method. 10
Conclusions
In summary, the present high-resolution MS study demonstrated that insulin content in vials, even when exposed to suboptimal temperature conditions, remained within the acceptable limits specified by FDA and EMA without evidence of degradation. From a technical point of view, we established flow-injection high-resolution MS as an attractive tool for high-throughput determination of insulin concentrations. The method that allows accurate measurement of compound concentrations in 1 min is capable of directly detecting degradation such as deamidation or oxidation and therefore represents an alternative to HPLC-based methods.
Footnotes
Acknowledgments
The authors thank the University Institute of Clinical Chemistry/Center of Laboratory Medicine, Clinical Metabolomics Facility, Inselspital, Bern University Hospital for support in logistics and instrumentation.
Author Contributions
L.B. and M.G. designed the study. F.B. and M.G. performed the experiments, collected, processed, and analyzed the data. L.B. and M.G. wrote the article. B.V. and C.S. critically reviewed the article. L.B. and M.G. are the guarantors of this work, as such, had full access to all the data in the study, and take the responsibility for the integrity of the data and the accuracy of the data analysis.
Author Disclosure Statement
No competing financial interests exist.
Funding Information
The work was supported by the UDEM Scientific Fund. B.V. was supported by the Foundation “Fonds pour la Recherche Thérapeutique,” Puilly, Switzerland.
Supplementary Material
Supplementary Figure S1
Supplementary Figure S2
Supplementary Figure S3
Supplementary Figure S4
Supplementary Table S1
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
