Concentrations of Intact Insulin Concurs With FDA and EMA Standards When Measured by HPLC in Different Parts of the Distribution Cold Chain

Abstract

Background:

The article by Carter and Heinemann raised serious concerns about the concentrations of insulin in vials being sold in US pharmacies. To study the claims made in the manuscript, we reviewed Novo Nordisk data on insulin concentration.

Methods:

Insulin concentrations within vials from three different sources along the distribution chain were evaluated utilizing currently accepted US Pharmacopeia methodology: (1) insulin content and stability based on production batches covering 7 years of insulin production, (2) insulin content in samples returned to Novo Nordisk over the last three years in the United States, and (3) data from eight years of independent EMA testing.

Results:

The data demonstrated that without exception (1) insulin quality based on stability data was maintained, even in scenarios that stressed the normal recommendations for temperature storage conditions, (2) insulin content from the last three years of samples returned to Novo Nordisk from patients in the United States (233 vials) was within USP requirements recognized by FDA, and (3) ten years of independent EMA sampling of products obtained at wholesalers and pharmacies across the EU confirmed compliance (n = 43).

Conclusions:

The study by Carter and Heinemann utilized an LC-MS technique, which has not been validated for the quantification of insulin in pharmaceutical preparations. It appears likely that their findings are the result of the method utilized and not due to decreased insulin content in samples. However, recognizing the importance of maintaining Insulin content from production to the patient, Novo Nordisk supports continued evaluation of insulin distributed to pharmacies and patients utilizing validated techniques compliant with international pharmacopeias.

Around 6 million people with diabetes use insulin either alone or in combination with an oral medication to manage their diabetes in the United States.¹ Substantial variance in the quality (ie, concentration or bioactivity) of insulin, a lifesaving treatment for diabetes, could be harmful to patients requiring insulin therapy. A recent study by Carter and Heinemann raised doubts about the quality of insulin in the supply chain when they evaluated intact insulin concentration in 18 vials randomly purchased in pharmacies in the United States.² The study used a highly sensitive method, liquid chromatography–mass spectrometry (LC-MS), which has however not been validated to correspond to the US Pharmacopeia and globally accepted method to quantify insulin.³ They reported intact insulin concentrations ranging from 13.9 to 94.2 U/ml (mean 40.2 U/ml) with no vial meeting the minimum US Pharmacopeia (USP) standard of not lower than 95 U/ml (recognized by the US Food and Drug Administration, FDA). The low concentration of insulin observed in the study was attributed to potential loss of content in the cold supply chain² and, more specifically, after transport from the manufacturer.

In response to these results, patient organizations such as American Diabetes Association (ADA) have posted a statement concerning the publication.⁴ They have raised concerns over the small sample size of 18 vials as well as the methodology used to assess the insulin concentration. In addition, ADA has reviewed and confirmed that the insulin manufacturers strictly adhere to and document quality control procedures throughout the supply chain to meet the USP standards recognized by the FDA to ensure safety, potency and efficacy of insulin.⁴

On the background of the article by Carter and Heinemann and given the importance of insulin quality on patient outcomes, we evaluated insulin concentrations in vials from three different sources: (1) we analyzed insulin content and stability based on production batches covering 7 years of production, (2) we measured insulin content from the last three years of insulin samples returned to Novo Nordisk from patients in the United States because of concerns expressed by the patients, and (3) we reviewed data from 10 years of independent European Medicines Association (EMA) sampling and testing of products.

Methods

Analysis of Insulin Concentration and Stability of Production Batches Covering Seven Years of Production

Analysis of insulin concentration and stability of production batches covering seven years of production of NPH insulin: Insulatard® (named Novolin® N in the United States; Novo Nordisk A/S, Bagsværd, Denmark) and regular human insulin Actrapid® (named Novolin® R in the United States; Novo Nordisk A/S).

For Insulatard, the analyses are based on stability data from 88 batches covering 7 years of production (n = 761 tested samples), and for Actrapid the analyses are based on stability data from 85 batches covering 7 years of production (n = 855 tested samples).

Actrapid was analyzed for impact on insulin concentration under the following two conditions:

• As per label: Stored for 30 months at 5°C and handled as per procedure during the cold-chain transport, that is, inspected, labeled and packed, and transported at 5 ± 3°C and during pharmacy storage. Samples were left at 30 C for 6 weeks to mimic in-use impact. The degradation due to excursions from 5 ± 3°C is included in the specification limits.

• Stressed conditions: Stored for 30 months at 5°C and handled as per procedure during the cold-chain transport, that is, inspected, labeled and packed, and transported at 5 ± 3°C and during pharmacy storage. Samples were left for additional 4 weeks in 30°C to mimic non recommended handling conditions and then left at 30°C for 6 weeks to mimic in-use impact. The degradation due to excursions from 5 ± 3°C is included in the specification limits.

Insulatard was analyzed for impact on insulin concentration under the following two conditions:

• As per label: Stored for 30 months at 5°C and handled as per procedure during the cold-chain transport and the pharmacy storage and left at 30°C for additional 6 weeks to mimic in-use impact.

• Stressed conditions: Stored for 30 months at 5°C and handled as per procedure during the cold-chain transport and the pharmacy storage as well as left for 4 more weeks in 30°C to mimic non recommended handling conditions and then left at 30°C for additional 6 weeks to mimic in-use impact.

Insulin concentrations were measured using reverse phase–high pressure liquid chromatography (RP-HPLC) according to regulatory standards (European Pharmacopeia, United States Pharmacopeia, Japanese Pharmacopeia: 3.7.17 method description for identity, assay and related products of Human Insulin).^3,5,6 This method requires minimal sample preparation.

A total of 4 µL of 6 M hydrochloric acid per mL is added to the insulin preparation to be examined followed by 10 minutes rotation mixing prior to RP-HPLC analysis. Acidification ensures dissolution of all insulin crystals and dissociates insulin hexamers into monomers for both reference and test sample. The chromatographic separation is selective and there is no interference between excipient components and insulin peak in the chromatogram. The insulin peak in the test sample is easily quantified against insulin peak in the chromatogram of the reference sample. This method has been correlated to bioactive insulin through extensive testing for decades and has been accepted as the standard analysis method for accurate and precise measurement of bioactive insulin in insulin drug formulations by the different regulatory authorities and the US Pharmacopeia.^3,5,6

Analysis of Last 3 Years of Returned Insulin Samples From US Patients

To evaluate the insulin concentration in the vials used by patients, we assessed insulin samples returned by patients in the United States. A detailed analysis was performed in Denmark on 233 samples of Actrapid and Insulatard. Insulin concentrations were measured using RP-HPLC as utilized for the production batches.

The European Medicines Agency

In accordance with European Parliament regulation, EMA has the responsibility to coordinate and supervise the quality of the medicinal products by requesting testing of compliance with their authorized specifications by the official medicinal control laboratories or by a laboratory a member state has designated for that purpose.⁷ In accordance with this the EMA organizes an annual Sampling and Testing Programme for Centrally Authorised Products. Some of these products included insulin containing products from Novo Nordisk. The products were selected for inclusion using a risk based methodology. The EMA purchased Novo Nordisk insulins from pharmacies and wholesalers, and tested the samples to confirm compliance with insulin content specification. We have collected this information from EMA from 2007 to 2017 for Insulatard (Novo Nordisk A/S), Actrapid (Novo Nordisk A/S), Levemir® (Novo Nordisk A/S), NovoMix® (Novo Nordisk A/S), Mixtard® 30 (Novo Nordisk A/S), NovoRapid® (Novo Nordisk A/S), and Tresiba® (Novo Nordisk A/S).⁸

Results

The evaluation of the impact from shelf life storage, cold-chain transport and pharmacy handling, and in-use conditions on insulin concentrations of Actrapid and Insulatard is based on data from stability studies of insulin production batches covering 7 years of production. These were followed for stability at 3 different temperatures. For Insulatard the analyses are based on stability data from 88 batches covering 7 years of production (n = 761 tested samples), and for Actrapid the analyses are based on stability data from 85 batches covering 7 years of production (n = 855 tested samples). The data demonstrated that even at scenarios that stressed the normal recommendations for temperature storage conditions, the resulting impact on insulin concentrations, that is, 4.0 IU/ml and 1.4 IU/ml for Actrapid and Insulatard, remained within the acceptable limits outlined in the FDA standards (Table 1). None of the samples were outside FDA specifications. The analysis was based on stability data from 85 or more insulin production batches covering 7 years of production.

Table 1.

Estimated Impact on Insulin Concentration of Actrapid and Insulatard: in an “As per Label” Scenario and an “Extreme Outside Label” Scenario.

Product	Scenario	Full shelf life storage impact	+ cold-chain transport and pharmacy handling impact	+ in-use impact	= resulting assay impact
Actrapid	As per label	30 months at 5°C	As per handling procedure	6 weeks at 30°C	Mean: –3.1 IU/ml 95% CI: –4.0 IU/ml
	Extreme outside Label	30 months at 5°C	As per handling procedure +4 weeks at 30°C	6 weeks at 30°C	Mean: –4.0 IU/ml 95% CI: –4.9 IU/ml
Insulatard	As per label	30 months at 5°C	As per handling procedure	6 weeks at 30°C	Mean: –1.0 IU/ml 95% CI: –1.9 IU/ml
	Extreme outside label	30 months at 5°C	As per handling procedure + 4 weeks at 30°C	6 weeks at 30°C	Mean: –1.4 IU/ml 95% CI: –2.3 IU/ml

The estimation is based on stability studies data from insulin production batches covering 7 years of production of Actrapid and Insulatard.

The method validation for Actrapid showed a repeatability of 0.1361% relative standard deviation (RSD) and an intermediate precision of 0.2529% RSD (Table 2). The method validation for Insulatard showed a repeatability of 0.2332% RSD and an intermediate precision of 0.3762% RSD (Table 2).

Table 2.

Precision Data for Actrapid and Insulatard.

			Result of analysis (units/mL), Assay
Actrapid 100 Units/mL	Mean		99.4632
	% RSD (%)	Repeatability	0.1361
		Intermediate precision	0.2529
	95% confidence interval % RSD	Repeatability	0.0877-0.2997
		Intermediate precision	0.1659-0.5267
Insulatard 100 units/mL	Mean		99.7729
	% RSD (%)	Repeatability	0.2332
		Intermediate precision	0.3762
	95% confidence interval % RSD	Repeatability	0.1502-0.5134
		Intermediate precision	0.2505-0.7499

An analysis of insulin returned to Novo Nordisk by patients in the United States due to concerns about its effectiveness for the period January 5, 2015, to January 5, 2018, for Actrapid (Novo Nordisk A/S) and Insulatard (Novo Nordisk A/S) revealed that in the 233 vials returned from the US patients in the last 3 years showed an average assay concentration of 98.5 IU/ml with a standard deviation of 6.7 and 2.4; respectively (Table 3). The vials were returned to Novo Nordisk by individual patients in the United States over concerns that a particular vial or penfill® of insulin was ineffective in controlling blood glucose levels or the patient noted a crack in the insulin vial. There has been no recent increase in the number of vials returned to Novo Nordisk over the last several years and no safety signal relating to lack of efficacy of fast or long-acting human insulin in the United States has been identified when assessing the adverse event database over last 3 years.

Table 3.

Analysis of Last 3 Years of Returned Insulin Samples (233 Vials) From US Patients.

Product	Average assay (IU/ml)	SD	Sample size
Actrapid	98.5	6.7	58
Insulatard	98.5	2.4	175
All products above	98.5	3.9	233

Only five vials, four of Insulatard and one of Actrapid, fell below 95 U/ml. The four vials on Insulatard had at least 50% of insulin removed and had evidence of inhomogeneous extraction (inadequate mixing). The contents of the vial of Actrapid had evidence of contamination with another liquid. All data were included in the analyses above, as also indicated by the standard deviations.

The data from EMA testing of insulin content also confirmed compliance with required specification for 7 different insulins from 2007 to 2017 (Table 4).

Table 4.

EMA Sampling and Testing of Novo Nordisk Products.

Year	Products sampled	Assay test result	Specification
2011	Actrapid	98.4-100.4 IU/ml	95-105 IU/ml
2014	Insulatard	3.42-3.53 mg/ml	95-105 IU/ml or 3.29-3.64 mg/ml
2007	Levemir	2410-2502 nmol/ml	2261-2520 nmol/ml (94.2-105 %)
2009	NovoMix 30	581-601 nmol/ml	564-630 nmol/ml (94.0-105 U/ml)
2010	Levemir	2282-2434 nmol/ml	2280-2520 nmol (95.0-105 %)
2014	Actraphane 50 Mixtard 30	100.96-102.74 IU/ml 101.60-103.23 IU/ml	95-105 IU/ml
2014	Levemir	2338-2444 nmol/ml	2280-2520 nmol/ml (95.0-105 %)
2014	NovoRapid	594.6-599.6 nmol/ml	570-630 nmol/ml (95-105 U/ml)
2014	Tresiba	1190.6-1208.9 nmol/ml	1140-1260 nmol/ml (95.0-105 %)
2017	NovoMix 30 Penfill	566-595 nmol/ml	564-630 nmol/ml (94.0-105 U/ml)

Discussion

Our data on insulin production quality confirm that insulin delivered to pharmacies and when in use by patients meets the required regulatory standards and provides no evidence of concern about insulin content. Our results include analyses of insulin samples that were returned to Novo Nordisk “for cause” and were outside the recommended cold chain during their return by shipping to Novo Nordisk. These are exactly the type of samples that might be predicted to contain degraded insulin, since transportation of these samples occurs via post, but insulin content met specifications. Our results are further supported with 100% compliance in EMA quality controls of Novo Nordisk insulins over the last 10 years.

Carter and Heinemann implicated the cold supply chain for the low insulin concentrations observed in their study. The cold supply chain is the temperature controlled supply chain for insulins. However, a previous study from Chandler et al modeled the impact of transit temperature excursions on the potency of insulin (regular, NPH, and 70/30) and demonstrated that the potency would be maintained in accordance with US Pharmacopeia⁹ under the conditions that they tested.

The low insulin concentration observed by Carter and Heinemann, may be explained by the analytical method utilized in the study. Insulin content was measured using quadrapole time-of-flight mass spectrometer along with LC-MS.² LC-MS is a very sensitive and specific method and can quantify insulin in human plasma from 0.2 to 25 ng/mL (0.45 pmol/L to 56.17 pmol/L).¹⁰ However, the sample preparation in this method requires extensive dilution that poses a significant risk for loss of the analyte. In addition, sample preparation may result in insufficient dissociation of insulin hexamers causing a false decrease of insulin content in the assay. It is also important to note that the QTOF/LC-MS method used in the study was not validated to quantify insulin as outlined by regulatory authorities. This has also been corroborated in the response letter to the Carter and Heinemann publication from Eli Lilly¹¹ and the response letter from the ADA, Juvenile Diabetes Research Foundation (JDRF), and the Helmsley Charitable trust.¹²

In our study, the RP-HPLC method that is the recognized standard for assessing insulin content was utilized. This assay requires minimum sample preparation and results in 100% dissociation of insulin hexamers to monomers. This ensures insulin content is quantified with high precision and robustness, as demonstrated in our results. Furthermore, the HPLC assay has been widely validated for determining bio-potency.^13-15 RP-HPLC methodology has been used to quantify insulin potency in insulin drug products since the late 1980s and has been acknowledged by the regulatory authorities to provide a reliable quantification of insulin content and biopotency.¹⁵ Due to high precision and robustness of the RP-HPLC based methods, these have been adopted in all major pharmacopeia monographs for human insulin since the 1990s.^3,5,6

Conclusion

The data reported here provides a comprehensive analysis of insulin concentrations in commercially available samples. We also measured insulin concentrations in a much larger sample size compared to the study by Carter and Heinemann. However, the methodology used to analyze insulin was different between the two studies. We also did not just focus on samples of insulins obtained from US pharmacies. It is important to note that if there was a significant generalized decrease in insulin potency as mentioned by Carter and Heinemann, there should have been a significant increase in adverse event reporting for insulin. However, no increase in adverse event reporting has been observed which also suggests that the findings were a result of the methods utilized in this study.