In Response to Letters to the Editor From the American Diabetes Association and Eli Lilly in Regard to: Insulin Concentration in Vials Randomly Purchased in Pharmacies in the United States: Considerable Loss in the Cold Supply Chain

Dear Editor,

We are grateful for the opportunity to respond to comments from our colleagues at the American Diabetes Association (ADA) and Eli Lilly regarding the results of our insulin concentration measurements published in the December online-first issue of this journal. We are in complete agreement that our preliminary findings do warrant further investigation as it is clearly in the interests of people with diabetes worldwide to have clarity in relation to this topic.

Alan W. Carter and Lutz Heinemann

Response to letters to the editor from Peterson et al from the ADA ¹ and Connery and Martin from Eli Lilly ² regarding our publication “Insulin Concentration in Vials Randomly Purchased in Pharmacies in the United States: Considerable Loss in the Cold Supply Chain.”

Overall, the statement “we have serious reservations about the reported results and question whether they accurately describe the actual potency of pharmacy-obtained insulin” encapsulates the current debate raised by our article. In response, we do share these reservations; in our original article we did recognize some important limitations but nevertheless using the approach described we have reported new data that raise clinically relevant concerns about the potency of insulin at the end of the cold supply chain. We all agree that this warrants further investigation. We are pleased to see that this will be happening and the reality is that we owe this to people with diabetes.

However, we would also like to respond to some of the specific concerns raised.

Comments by the ADA

• 1. Statement: “The number of samples studied is very small (18 vials total), and the methodology for sample handling and preparation is minimally described.”

• Response: We acknowledged the limitations of our methodology in our publication, as well as during a public presentation at the Diabetes Technology Meeting in November 2017 (see the report by Close Concerns at https://www.closeconcerns.com/knowledgebase/r/b6d0dea7?utm_source=Closer+Look+Subscribers+2017&utm_campaign=d539e768e2-2018-01-26_%28HTML_LINKS%29_JDRF_Mission_Su01_26_2018&utm_medium=email&utm_term=0_c55d924bf1-d539e768e2-411755437).

• Moving forward, we are preparing a “White Paper” with details of all information/aspects that are relevant with respect to the results of our evaluation. When completed this paper will be made available online.

• 2. Statement: “More important, there are serious concerns about whether the results match real-world experience and whether appropriate testing methods were used.”

• Response: We believe that real-world experiences do, in fact, suggest that there may be problems with maintaining insulin potency. Since the publication of our data, there has been considerable discussion on social media about this topic. A number of people with diabetes and physicians have contacted us directly with unsolicited comments related to their personal experience of episodic “unpredictable” insulin action. However, we agree that this has not been proven and that further research on this topic is warranted.

• 3. Statement: “The results are inconsistent with our experience of using pharmacy-obtained insulin in controlled clinical studies, where the potency is clearly in line with the labeled 100 U/ml.”

• Response: There are obviously important differences between participation in clinical trials and real-world experiences. Given the above statement, it might be helpful to see data that confirm the statement that the potency is “clearly in line” with the labeled concentration in insulin vials picked up at end user pharmacies given that handing and transport may differ markedly from research sites.

• 4. Statement: “Such alterations in potency would clearly be evident to patients, especially those frequently checking their blood glucose levels or using continuous glucose monitoring.”

• Response: For the majority of people living with diabetes using insulin, achieving optimum glucose control is inherently difficult as evidenced by population-based A1c levels and the continued clinical problem of hypoglycemia despite advances in insulin formulation and delivery systems. The question at hand is whether some of this difficulty arises from the insulin vial itself. This potential confounder can only be addressed in properly powered and carefully controlled clinical studies (ie, glucose clamp studies). When unexpected problems with glucose control arise, how often do clinicians recommend throwing away the current insulin vial and using a new one or how often does an individual with diabetes do this? Such cases most often will not be reported or documented in a uniform way, leading to the risk of considerable underreporting. There is no independent resource to which the individual can report such complaints without considerable effort.

• Furthermore, when vials with doubtful insulin activity are sent back to the manufacturer (no publicly available data exist about how often this happens), the usual response in practically every case is that “the concentration is according to the specifications.” This is invariably reported using a standard format without providing any specific details on how the vial in question may or may not have been evaluated. We believe that this can create an impression that the insulin user must have done something wrong or the answer lies elsewhere (which we agree is possible). Remember that measurements to evaluate a vial of “bad” insulin utilize the manufacturer’s standard high-performance liquid chromatography (HPLC) methodology and that this methodology may not reflect the vial’s actual biologic insulin activity. From our point of view, this discrepancy between the clinical observations raised by people with diabetes/physicians and the measurement results reported by the manufacturer deserves a more complete explanation performed by an independent party.

• 5. Statement: “Clinically we rarely receive complaints other than during extremes in temperature when insulin stability will be problematic.”

• Response: We are very aware of the impact of extreme temperatures on insulin stability. In our study no fibrils were visible in the vials that we examined. However, what happens if insulin storage is too cold but not frozen during the transport and motion of the cold chain? Changes to the ratio of the mixture of insulin monomers, dimers, and hexamers might occur in the vials, which also would support formation of stable insulin dimers. Minute structural changes might show up that impair insulin action and probably impact the insulin activity without displaying fibrils. We see the need to evaluate such potentially clinically relevant aspects by using appropriate methods, rather than with an older HPLC standard that may need to be updated.

• 6. Statement: “A more quantitative physiologic approach to assess potency would be to perform euglycemic clamp studies to assess insulin action with different insulin vials.”

• Response: We fully agree. There are clear reasons why the approval of novel and biosimilar insulins by regulatory agencies is based on extensive clinical trials with glucose clamps rather than the results of analytical laboratory work. It would be of interest to perform head-to-head glucose infusion rate studies by independent sites that simultaneously compare vials that have identified insulin anomalies following transport with insulin vials shipped directly from the manufacturer. Such studies might help to clarify this topic.

• 7. Statement: “We also question the test methodology (their reference 2). Originally designed to identify modified insulins in plasma, it faces difficult challenges as a quantitative method and has not been validated for measurement of the large amount of insulin in commercial vials.”

• Response: It is absolutely valid to question the test methodology. Additional evaluations would clarify this.

• 8. Statement: “Insulin self-aggregates, sticks to container and column surfaces, and is not very soluble in organic solvents, reducing recovery.”

• Response: We are fully aware of such issues and have taken great care in this respect. A key point of our measurements is that we handled the United States Pharmacopeia (USP) human insulin standard in the very same manner as the samples drawn from the insulin vials. The only significant difference was that the USP standard had not been subjected to the prolonged storage and handling of the cold chain pharmacy vials and the unlikely confounding influence that insulin excipients might have in commercial insulin formulations.

• 9. Statement: “It would be important to compare freshly-manufactured and cold supply chain samples using the authors’ method, manufacturers’ methods, dual antibody immunoassays, protein content, and lyophilized weight.”

• Response: This would be the ideal approach, and we highly support performing such evaluations to solve an issue we did not expect to arise.

• 10. Statement: “We have reviewed and confirmed the quality control procedures that are strictly adhered to and documented by each manufacturer to meet U.S. Food and Drug Administration (FDA) standards to ensure the safety, potency and efficacy of insulin throughout the supply chain.”

• Response: We are clearly aware of such procedures; however, is it not worth verifying as our research was designed to determine how effective such procedures are at the point of delivery? Standard commercial product release testing is one thing; data from randomly collected samples are something different. If an insulin manufacturer has data from such evaluations available (obtained with which measurement method?), they should share this with the scientific community.

• 11. Statement: “It seems highly unlikely that the insulin could degrade by any significant amount, certainly not the 75% degradation consistently found by Carter and Heinemann for regular human insulin.”

• Response: Other things in the past also appeared to be unlikely and were challenged by scientists/medical associations, until shown to be real (eg, the relevance of helicobacter pylori). It would be fine for us to be proven wrong! Nevertheless, at this time our data are the data and an open discussion is needed about the clinical relevance of our observations.

• 12. Statement: “We fear that there is significant risk of harm, either in the form of patient anxiety or of outright noncompliance or misuse if patients incorrectly believe that their insulin’s potency is substantially less than it is supposed to be.”

• Response: We are unaware of any data supporting this statement. Alternatively, and as outlined above, it appears as if a number of people with diabetes (and clinical colleagues) already have doubts about their insulin potency. Going forward we suggest performing a surveillance about this topic to gain more widespread knowledge through independent research – this may be something that the ADA could lead on?

• 13. Statement: “To address this concern, the American Diabetes Association is planning to work with the Juvenile Diabetes Research Foundation and the Helmsley Charitable Trust to commission an independent laboratory analysis of samples obtained from manufacturers and from retail pharmacies.”

• Response: This is highly appreciated and necessary.

• 14. Statement: “We will assess insulin concentration using several laboratory methods, including that used by the authors. We hope to determine whether these findings represent a real problem in the potency of insulin available at the retail level, or if the reported results are in some way erroneous.”

• Response: This is clearly the best approach from our point of view. We are happy to make publicly available all the information that is available to us. The current approach of simply measuring the insulin concentration in insulin vials again with the standard HPLC method might be giving a false impression that everything is ok, as it may be overlooking additional relevant information.

Comments by Eli Lilly

• 1. Statement: “Testing method, this was not developed or validated to measure insulin concentration in the drug product.”

• Response: The HPLC method has been the FDA accepted standard method for more than 30 years, but does this method provide all relevant information? The original goal of our efforts was to map insulin for any small differences from random samples taken from the supply chain using latest liquid chromatography/mass spectrometry equipment. The first step was to compare content of the “intact” monomeric form to a known USP standard. However, we discovered that the vials we purchased at different pharmacies did not appear to have the human insulin concentrations expected. The need for others to validate the method employed by us and conduct further evaluation was stated in our conclusions.

• 2. Statement: “The USP methods have been specifically developed and validated per International Council for Harmonization (ICH) for the accurate and precise measurement of insulin concentration in insulin drug product formulations.”

• Response: This is also correct. However, the standard method may not measure structural variations in insulin that are acquired during shipment and storage. Small changes in the secondary structure that may be relevant to biological activity might not be detected by this method. We believe our method accounts for changes introduced on the way from the manufacturer to the pharmacies by exposure to temperature extremes or vibration during transport from the manufacturer to the pharmacies. The method employed by us is focused on measurement of “intact” monomeric insulin. The reference samples we used for comparison conformed to the USP standard listed weight of 0.0347 mg/unit for human insulin to prepare the standard solution and appeared to contain only the monomeric (bioactive) form.

• 3. Statement: “The method referenced in the article subjected the well-characterized insulin formulations to unnecessary and complex manipulations, which significantly and negatively affect the solubility of insulin in the original formulations.”

• Response: From our point of view this statement is not scientifically based. The USP standard HPLC method requires the samples be exposed to a very strong acid (hydrochloric acid) in a timed preparation step to force dissolution of all (stable) dimers in the formulation and any insulin remaining “stuck” to the container walls. In contrast, with our method the samples were not subjected to such aggressive pretreatment. After preparation, the USP human insulin standard was subjected to the same “manipulations” as all the samples we tested. Compared to the USP human insulin standard solution that was freshly prepared at 100 U/ml concentration, the insulin vials purchased at different pharmacies should have an identical insulin concentration. The only difference between the USP standard and samples was the length of time the pharmacy samples spent under cold or hot temperatures during shipment and storage. A possible (albeit unlikely) confounder could be the effect(s) of other formulation excipients on the assay system. The USP human insulin standard did not contain the same buffers and preservatives as the commercial formulations. We see a need for head-to-head comparisons between both methods (and other methods) to evaluate whether detection of intact monomeric insulin accurately reflects the bioactive dose.

• 4. Statement: “Lilly uses robust quality management systems and product control strategies to meet or exceed the high standards required by global regulatory agencies and expected by those who prescribe and use our products.”

• Response: We have never challenged the production quality of Lilly; our concerns involve what may be happening to the insulin during transport in the “Cold Supply Chain.” Can companies such as Eli Lilly provide data from their evaluations of the insulin concentration with the standard HPLC method in insulin vials purchased randomly and without prior notification in different pharmacies across the United States? Assuming that the integrity of insulin formulations can be influenced by suboptimal storage/transport, would it not be appropriate to verify the findings reported in our paper?

• To our knowledge, each insulin manufacturer provides clear instructions how the insulin should be stored/transported. However, it is not clear how well insulin transport is monitored and documented (eg, by means of temperature logs) in daily practice. We see a risk that insulin may be exposed to temperatures outside the recommended range, depending on the proximity of a given insulin vial to the cooling element. Even a temperature probe/thermometer that is centrally located may not represent vials that are close to the cooling element. Variance in supply chain temperatures may enhance the tendency for self-aggregation or degradation. To date, we know of no insulin vials collected randomly without prior notification of the manufacturer or the shipping company that have been evaluated across the United States at, for example, different stages of the cold supply chain. Data previously presented by insulin manufacturers from such studies from samples taken outside the United States showed no issues; however, only the standard HPLC method that might not detect issues regarding biologic activity was applied to measure “insulin concentration.”

• 5. Statement: “These controls ensure the solubility and stability of insulin over the shelf life of the product when stored under the stated label and in-use conditions.”

• Response: This is our assumption as well. Our concerns are in regard to what happens with the product on the way to and at the pharmacy. Do the vials delivered to patients in any way impair their glucose control? This should be studied with scrutiny in controlled and systematic evaluations. It would be of great benefit for Eli Lilly and other insulin manufacturers to scientifically evaluate our findings. Only providing a statement that this is the case is not seen as sufficient proof.

• 6. Statement: “Wholesalers and pharmacies that distribute insulin must hold a pharmaceutical distribution license issued by local government agencies in the U.S., which requires appropriate cold chain storage and distribution practices.”

• Response: Again, we are fully aware of this. It would be helpful to know the methods by which monitoring is conducted. Can Lilly provide information about the steps taken to ensure that temperature (and possibly vibration) are consistently monitored in the multiple vehicles and methods by which insulin in transported, as well as what advice is given to pharmacies for refrigerator cooling. It would be good to have additional evidence.

• 7. Statement: “Additionally, regulatory agency surveillance programs are conducted periodically and we are not aware of any concerns arising from testing.”

• Response: This statement is somewhat vague with no more detailed information and data provided. The references refer to measurements made in the EU, not in the United States.

• 8. Statement: “If studies are pursued to further clarify this issue, every effort should be made to ensure the validity of the analytical results before drawing conclusions related to the post-manufacturing cold supply chain.”

• Response: We fully agree. The aim of our publication was to highlight that our measurements (“the data are the data”) suggest that this more recent and sensitive method indicates that the insulin concentration in vials handed over to patients do not contain the concentration of intact monomeric insulin that they are supposed to have. One alternative approach might be to form a panel of independent and corporate researchers to outline a method that will then be funded to evaluate the validity of an appropriate evaluation method or methods.

• 9. Statement: “This can be ensured by using a certified laboratory that has successfully demonstrated proficiency in executing the standard and accepted FDA reference methods (USP) for determining insulin concentration in drug product prior to analyzing any investigational samples.”

• Response: MRIGlobal (where our measurements were performed) had the lab capabilities and ISO certifications for conducting such measurements. These measurements may show that the insulin concentration is present according to expectations, that is, close or at 100 IU/ml; however, the question is, is all the insulin present “bioactive”? To evaluate this, different methods/types of measurement should also be employed. Other current analytical methods allow evaluation of the structure of the insulin(s) present in a vial of the formulation. The biological activity of a given insulin formulation can be appropriately evaluated in animal models (originally insulin potency measurements were performed with rabbits) or by euglycemic glucose clamp studies in humans.

General Comments

• We welcome the supportive views of the ADA and also welcome further feedback. We are encouraged by the combined foundations’ stated intent to explore this issue further with well-designed and conducted studies, ideally in both the US and European supply chains.

• We took great care in handling vials and uniform preparation of all samples.

• We have the impression that the many aspects of insulin transport and usage requires more attention by all stakeholders.

• We see the need for temperature tracking from start to finish in the cold chain as an important part of a consumer-level quality assurance system.

• We believe we share a collective responsibility to determine any impact of the supply chain on known and unknown variables of insulin configuration and concentration to the point of self-administration of insulin by the patients. This holds true for other biologics as well.

• We think that evaluating quality of insulin should not be limited to insulin concentration measurements by forced disassociation alone. Evaluation for any structural changes that may occur between time of manufacture and at the end of the cold chain should also be conducted. Sanofi published data about differences between insulin glargine copies from different manufacturer sources. ³ These evaluations would best be done by independent institutions.

• We believe that reports from patients and clinicians regarding suboptimal insulin activity should be handled more systematically in a formal system. Patients with diabetes who depend upon insulin to survive deserve certainty that the vials they receive at pharmacies have full biologic activity.