Center for Drug Evaluation, National Medical Products Administration

December 2024

I. Overview

Insulin drug products include human insulin and insulin analogs. Based on their duration of action, insulin products are typically categorized into rapid-acting insulin, short-acting insulin, intermediate-acting insulin, and long-acting insulin. The first two categories are also known as prandial (mealtime) insulins, primarily used to control postprandial blood glucose levels; the latter two are known as basal insulins, mainly used to control basal blood glucose levels in the fasting state. The aforementioned insulin products can be used alone clinically or as mixtures of rapid-/short-acting insulin and intermediate-/long-acting insulin (biphasic) in various ratios, or in premixed formulations.

With the widespread application of insulin products, novel insulin formulations have gradually become a research hotspot in recent years. The primary differences among various insulin products lie in their pharmacokinetic (PK) and pharmacodynamic (PD) characteristics. The euglycemic glucose clamp technique, which can effectively exclude the influence of endogenous insulin and objectively reflect the PK and PD characteristics of exogenous insulin products, is currently internationally recognized as a reliable method for the clinical evaluation of insulin products.

Building upon the "Guideline on the Design of Clinical Studies for Once-Daily Basal Insulin Biosimilars" released by the National Medical Products Administration (NMPA) in January 2022 [1] and the "Technical Guideline for Clinical Pharmacology Studies of Biosimilars" released in February 2022 [2], and considering the characteristics of insulin products and recommendations from relevant international guidelines [3, 4], this guideline is formulated.

This guideline primarily addresses the evaluation of PK and PD characteristics of insulin products using the euglycemic glucose clamp technique. It is applicable to the clinical evaluation of insulin products. For the PK and PD study design of once-weekly long-acting insulin products, communication with the Center for Drug Evaluation is recommended.

This guideline only represents the current perspectives and understanding of the drug regulatory authority. As scientific research advances, the relevant content of this guideline will be continuously refined and updated. When applying this guideline to design and conduct studies, reference should also be made to Good Clinical Practice (GCP), International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH) guidelines, and other relevant domestic and international issued guidelines. Matters not covered in this guideline are encouraged to be discussed with the regulatory agency through communication.

II. Basic Principle of the Euglycemic Glucose Clamp Technique

Depending on the research objective, glucose clamp techniques are classified into hyperglycemic clamp, euglycemic clamp, and hypoglycemic clamp. Among these, the euglycemic glucose clamp is currently recognized as the best method for evaluating the PK and PD characteristics of insulin and is widely used in the clinical assessment of insulin products.

In a euglycemic glucose clamp study, the blood glucose-lowering effect caused by increased blood insulin concentration (via exogenous administration of the test or reference insulin preparation) is counteracted by a variably adjusted glucose infusion rate (GIR). This effectively suppresses endogenous insulin secretion while "clamping" the blood glucose level within a pre-determined target range. After administration of the test insulin product, its PK characteristics can be described by measuring the changes in plasma or serum concentration of the test drug over time. Its PD characteristics can be described by the changes in the GIR required to maintain stable blood glucose concentration within the target range during the clamp.

There are two methods for implementing the euglycemic glucose clamp: the fully automated method and the manual method. The former employs a validated automated euglycemic clamp device that continuously monitors blood glucose via a closed-loop system and automatically adjusts the GIR every minute based on a computer algorithm using the difference between the measured and target blood glucose values to maintain stable blood glucose levels. The latter typically involves manual blood sampling every 2.5 to 30 minutes (depending on expected PK/PD characteristics and insulin properties), measuring blood glucose using validated instruments, and adjusting the GIR based on the measured values and investigator experience to maintain blood glucose as close as possible to the target value.

III. Design of Euglycemic Glucose Clamp Studies

(I) Clinical Study Design

Depending on the study objectives and the action profile of the insulin product, study designs employing the clamp technique after a single subcutaneous dose or after multiple subcutaneous doses reaching steady state can be used. Blinded design should be employed in studies whenever possible. For bioequivalence (BE) studies, a crossover design is recommended. It is preferable to study both the PK and PD characteristics of the insulin product in the same clamp study.

(II) Study Population

Depending on the study objective, the study population for euglycemic clamp studies can be insulin-sensitive healthy subjects or patients with type 1 diabetes mellitus (T1DM), among others. The age of subjects is recommended to be 18-45 years, with body weight ≥50 kg for males and ≥45 kg for females, and a body mass index (BMI) of 19.0-26.0 kg/m².

1. Selecting Healthy Subjects as the Study Population:

o Advantages:1) Easier recruitment and management of subjects; 2) Shorter clamp duration and relatively simpler operation; 3) Smaller inter-individual variation in peripheral insulin sensitivity among healthy subjects.

o Disadvantages:Endogenous insulin secretion cannot be completely suppressed. Especially in poorly controlled clamp studies with significant blood glucose fluctuations, endogenous insulin secretion in subjects may vary considerably, affecting the PD data of the test insulin product.

When using healthy subjects, efforts should be made to minimize the impact of endogenous insulin secretion and hepatic glucose output on the PK and PD data of the insulin product.

2. Selecting T1DM Patients as the Study Population:

o Advantages:Since T1DM patients have severely impaired pancreatic islet cell function, endogenous insulin secretion has minimal impact on the PK and PD data of the test insulin product.

o Disadvantages:1) Prior to the study, the subject's original insulin regimen may need to be changed or adjusted based on the type of insulin product under study (e.g., long-acting, rapid-acting). This may lead to significant elevation of fasting blood glucose and potentially increase the risk of nocturnal hypoglycemia, thereby increasing risks during the study; 2) The clamp procedure is more challenging and time-consuming; 3) Inter-individual variation in insulin sensitivity among T1DM subjects may be larger, potentially affecting the PD data of the insulin product; 4) Recruitment of T1DM subjects is more difficult.

In summary, the study population should be reasonably selected based on the objective of the clamp study and the PK/PD characteristics of the insulin product under investigation.

(III) Subject Preparation

One day before the start of the euglycemic glucose clamp study, subjects should avoid strenuous exercise, avoid consuming food (e.g., alcohol, caffeinated beverages) and drugs that may affect the study drug, refrain from smoking, maintain good dietary and sleep habits, and avoid mental stress and excessive fatigue. Study participants should fast (for at least 8 hours) before the clamp study and remain fasting throughout the entire clamp procedure to avoid affecting the study results. During the clamp, subjects should rest in a supine or semi-recumbent position and maintain emotional stability.

If the subjects are T1DM patients, appropriate measures, such as changing the insulin preparation, should be taken (while ensuring subject safety) to reduce the residual effect of the last insulin injection administered prior to the study.

(IV) Selection and Implementation of Clamp Method

Currently, the euglycemic glucose clamp without intravenous insulin infusion or the hyperinsulinemic-euglycemic glucose clamp with low-dose intravenous insulin infusion is often used to evaluate the PK and PD characteristics of insulin products.

1. Selecting Healthy Subjects as the Study Population:

When studying rapid-/short-acting insulin products in healthy subjects using the euglycemic clamp withoutintravenous insulin infusion, the target blood glucose value should be strictly controlled below the subject's own fasting blood glucose level to suppress endogenous insulin secretion as much as possible [5]. If using the hyperinsulinemic-euglycemicclamp, a low-dose rapid- or short-acting insulin is continuously infused intravenously beforeadministering the test insulin product to suppress endogenous insulin secretion, while glucose is infused simultaneously to maintain blood glucose at a pre-set target value (e.g., 4.5-5.5 mmol/L). The clamp continues until the GIR stabilizes at a plateau, then the test insulin product is injected subcutaneously, and blood glucose is subsequently maintained near the target value.

When studying intermediate-/long-acting insulin products in healthy subjects, it is recommended to preferably use the euglycemic clamp withoutintravenous insulin infusion [4]. This is because in a clamp with continuous intravenous insulin infusion, prolonged insulin infusion increases peripheral tissue insulin sensitivity [6], which may alter the late-phase PD parameters of intermediate-/long-acting insulin products and overestimate the pharmacodynamic data of the test insulin.

Although using somatostatin in clamp studies can better suppress the secretion of endogenous insulin, glucagon, and growth hormone in healthy subjects, its use is generally not recommended due to tolerability issues and post-clamp hypoglycemia concerns. Furthermore, somatostatin use reduces insulin clearance by approximately 20%, artificially prolonging the duration of action of the test insulin product [4].

2. Selecting T1DM Patients as the Study Population:

If T1DM patients are used as subjects, their original insulin therapy should be adjusted appropriately in advance relative to the administration time of the test insulin product, based on the duration of action of the insulin in their original regimen, to avoid impact on the PK and PD data of the test insulin (residual effect). At least 4-6 hours before administering the test insulin product, an intravenous infusion of short- or rapid-acting insulin (and glucose solution if necessary) should be initiated to maintain blood glucose near the target value (e.g., 5.5 mmol/L). The clamp should reach a steady state at least 1 hour before administering the test insulin product. After stopping the intravenous insulin infusion, the test insulin is administered subcutaneously. After administration, the euglycemic glucose clamp is started when the blood glucose level falls more than approximately 0.28 mmol/L below the target value.

(V) Insulin Dose Selection

The dose of the test insulin product in the clamp study should preferably be within a reasonable range of clinical use. Typically, commonly used doses in studies are: rapid-/short-acting insulin 0.2-0.3 U/kg, intermediate-acting insulin 0.3-0.4 U/kg, long-acting insulin (once-daily basal insulin) 0.4-0.6 U/kg [4].

Higher insulin doses usually yield more reliable PD parameters, thereby reducing PD variability. Moreover, the blood insulin levels achieved with higher absorption doses are expected to lie on the steep part of the insulin dose-response curve, which aids in selecting appropriate doses for later dose-finding studies. However, caution is needed when extrapolating PK and PD parameters obtained from studies where the selected dose exceeds the commonly used clinical dose range for most patients with type 1 or type 2 diabetes.

(VI) Target Blood Glucose Value in Clamp Studies

For euglycemic clamp studies conducted in healthy subjects, the target blood glucose value should typically be below the subject's fasting blood glucose level (e.g., 0.3 mmol/L lower than fasting, or 10% lower than fasting) [4]. If using the hyperinsulinemic-euglycemic clamp, the target blood glucose can be chosen between 4.5-5.5 mmol/L. For clamp studies in T1DM patients, blood glucose is usually maintained around 5.5 mmol/L.

In clamp studies, an acceptable deviation range of the actual blood glucose level from the target value (e.g., 10%) should be predefined. For healthy subjects, the blood glucose level during the study must not fall below 3.3 mmol/L [7], as this may trigger the secretion of counter-regulatory hormones (e.g., adrenaline, glucagon, cortisol, growth hormone), causing a rapid and significant decrease in peripheral tissue insulin sensitivity, thereby affecting the time-action profile of the insulin under study. For T1DM patients, the blood glucose level during the study must not fall below 3.9 mmol/L.

(VII) Duration of the Clamp Study

The rationale for selecting the clamp duration should be provided.

The duration of the clamp study needs to consider the known duration of action of the test insulin product and its dose-dependency (the effect of insulin dose on its duration of action). For healthy subjects, the duration of action of the test insulin in a clamp study can be defined as the time from insulin injection until the GIR returns to baseline or a predetermined value (e.g., 0.5 mg/kg/min) and blood glucose remains stable for at least 30 minutes. For diabetic patients, it can be defined as the time until blood glucose exceeds a predefined threshold (e.g., 8.3-13.9 mmol/L) without glucose infusion and remains above for at least 30 minutes.

Typically, clamp duration for rapid-acting insulin products is 8 to 10 hours, and for short-acting insulin products is 10 to 12 hours. For intermediate-acting and long-acting insulin products (once-daily basal insulin), a clamp duration of at least 24 hours is recommended.

(VIII) Quality Assessment of Clamp Studies

The quality of the clamp study directly affects the accuracy of the PK and PD data for insulin products. The stability of blood glucose levels around the target during the clamp and the extent of suppression of blood C-peptide levels are two main indicators for assessing clamp quality. For T1DM patients, assessment of blood C-peptide levels is not involved.

In euglycemic clamp studies, every effort should be made to keep blood glucose as stable as possible within a narrow normal range to achieve two purposes:

1. Avoid the secretion of counter-regulatory hormones triggered by low blood glucose and the consequent changes in body insulin sensitivity.

2. Avoid the stimulation of endogenous insulin secretion by elevated blood glucose.

Due to factors such as varying intervals between blood glucose measurements, inherent time delays between sampling, measuring blood glucose, and adjusting GIR, and the delayed response of subject blood glucose to GIR changes, the actual blood glucose values during a clamp typically do not match the pre-set target exactly but fluctuate around it. Clamp quality can be judged by assessing the magnitude and duration of deviation of actual blood glucose values from the target [8]. Generally, the closer the actual blood glucose values are to the target, the higher the clamp quality. Short-term, small deviations from the target are inevitable and have little impact on the PD values of the test insulin. However, PD data from clamps with prolonged, large deviations from the target blood glucose are less reliable.

The following are recommended for quality assessment:

(1) Coefficient of Variation of Blood Glucose (CVBG): Evaluates the precision of the clamp, i.e., the ratio of the standard deviation to the mean of all blood glucose values measured during each clamp, reflecting the magnitude of blood glucose fluctuation. Formula: (Standard Deviation of BG / Mean BG) x 100%. Generally, the CVBG for a clamp should preferably not exceed 5% [7].

(2) Mean Absolute Relative Difference (MARD): Evaluates the accuracy of the clamp, reflecting the overall fluctuation of clamped blood glucose values from the target throughout the study [9]. Formula for MARD (%):

MARD(%) = (Σ|BG_i - Target| / Target) / n * 100%

Where BG_i is the measured blood glucose value at each time point, Target is the pre-set blood glucose target for that clamp, and n is the number of blood glucose measurements in that clamp.

The change in serum C-peptide levels after dosing compared to baseline is also an important quality control indicator for clamp studies. Insulin and C-peptide are released in equimolar amounts from pancreatic beta-cell secretory vesicles. Blood C-peptide levels are commonly used in clamp studies to assess the suppression of endogenous insulin secretion. Ideally, post-dose C-peptide levels should remain consistently below the pre-dose baseline level (the mean of 2-3 C-peptide values at different time points before dosing is typically used as the C-peptide baseline) and reach a steady state. Under physiological conditions, endogenous insulin is secreted in pulses, so post-dose C-peptide values may fluctuate slightly around the baseline, but the increase should not exceed 50% of the baseline value. If post-dose C-peptide exceeds 50% above baseline, PD data from that clamp should be included in statistical analyses with caution, and sensitivity analysis is recommended.

The study report should include assessment and discussion of data reflecting clamp quality, such as CVBG, MARD, and changes in post-dose C-peptide levels relative to baseline.

(IX) Study Endpoints and Evaluation

1. Pharmacokinetic (PK) Characteristics (using insulin BE trial as an example):

o For rapid- and short-acting insulin products, primary endpoints should be AUC(0-t), AUC(0-∞), and Cmax. Secondary endpoints may include AUC for partial meaningful time periods (relevant to the specific insulin), corresponding time to AUC, tmax, and t1/2 [4, 10].

o For intermediate-acting insulin products, primary endpoints may be AUC(0-T) and Cmax. Secondary endpoints may include AUC(0-t), AUC(0-∞), AUC for partial meaningful time periods and corresponding time, tmax, and t1/2 [4, 10].

o For long-acting insulin products, such as once-daily basal insulins, which exhibit a relatively flat PK profile with blood concentrations remaining at a relatively low level for an extended period, determining Cmax and tmax may be difficult and lack clinical significance. Therefore, AUC(0-T) should be selected as the primary endpoint, while partial period AUCs such as AUC(0-T50%) and AUC(T50%-T) serve as secondary endpoints. t1/2 should be determined if possible [1, 4, 10].

o PK Equivalence Criteria: The traditional acceptance range is recommended. The 90% confidence interval for the ratio (Test/Reference) of primary PK endpoints like AUC and Cmax should fall within 80% to 125% [11].

Bioequivalence applies to the primary PK parameters. When tmax is closely related to clinical efficacy, paired non-parametric methods can be used for difference testing of tmax. If high variability is anticipated, a replicate design study (e.g., 3-period crossover with replication of the reference product) should be considered to justify a widened acceptance range.

2. Pharmacodynamic (PD) Characteristics (using insulin BE trial as an example):

The glucose infusion rate (GIR) over time reflects the time-action profile of the insulin product. GIR data derived from the GIR-time curve can serve as PD indicators for evaluating insulin products.

o For rapid- and short-acting insulin products, primary endpoints may be GIR-AUC(0-t) and GIRmax. Other meaningful endpoints include indicators reflecting the early action of rapid/short-acting insulins, such as GIR-AUC(0-2h), tGIRmax, Tonset, and t[4]. Time parameters related to GIR can be obtained from the smoothed GIR-time curve.

o For intermediate-acting insulin products, primary endpoints may be GIR-AUC(0-T) and GIRmax. Other meaningful endpoints include tGIRmax and Tonset of action.

o For long-acting insulin products, such as once-daily basal insulins, GIR-AUC(0-T) may be selected as the primary endpoint. Other meaningful endpoints include indicators reflecting the flatness of insulin action, such as fluctuations in GIR-AUC0-6h, GIR-AUC6-12h, GIR-AUC12-18h, and GIR-AUC18-24h (typically analyzed in 25% intervals), or the hourly GIR-AUC fluctuation AUC-FLAT,GIR,T (T = [AUC above GIRmean + AUC below GIRmean] / 24 [for a 24-hour interval]) [1, 4, 12].

o PD Equivalence Criteria: The traditional acceptance range is recommended. The 90% confidence interval for the ratio (Test/Reference) of primary PD endpoint GIR should fall within 80% to 125% [11]. If a replicate design study is conducted, the within-subject variability of PD endpoints should be recorded.

IV. Safety Evaluation

In clinical trials of insulin products, besides routine safety evaluation indicators, special attention should be paid to hypoglycemic events, local injection site reactions, and immunogenicity issues. Given that euglycemic clamp studies evaluating PK/PD characteristics of insulin products often employ a single-dose clinical protocol, detecting antibody incidence and titer within a few days after a single dose is less reliable. It is recommended that immunogenicity testing be conducted in Phase III clinical trials.

V. Summary

In the clinical development of insulin products, the euglycemic glucose clamp technique is commonly used to evaluate their PK and PD characteristics. This guideline elaborates on the principles, study design, implementation, and quality assessment of euglycemic glucose clamp studies. It only represents the current review perspectives, and valuable comments and suggestions from the industry are sincerely welcomed for future improvements.

For the PK and PD study design of once-weekly long-acting insulin products, communication with the Center for Drug Evaluation is recommended.

VI. Glossary

(I) Pharmacokinetic Parameters

• AUC: Area under the plasma concentration-time curve. This parameter is an important indicator of the extent of drug absorption, reflecting the drug's exposure characteristics in the body. Since blood concentration in PK studies can only be observed up to time t, AUC is expressed in two ways: AUC(0-t) and AUC(0-∞). The former is obtained by the trapezoidal rule; the latter is calculated as: AUC(0-∞) = AUC(0-t) + (last measurable concentration / terminal elimination rate constant).

o AUC(0-t): Area under the plasma concentration curve from administration to end of clamp at time t.

o AUC(0-∞): Area under the plasma concentration curve extrapolated to infinite time.

o AUC(0-T): AUC for the time of a dosing interval (AUC from dosing to a specified time period).

o AUC(0-T50%): AUC during the first half of a dosing interval (AUC from dosing to the time when plasma concentration approaches zero, focusing on the first half of this period).

o AUC(T50%-T): AUC during the second half of a dosing interval (AUC from dosing to the time when plasma concentration approaches zero, focusing on the second half of this period).

• Cmax: Maximum plasma concentration (peak plasma concentration), the highest concentration observed after dosing. It is an important indicator reflecting the rate and extent of drug absorption in the body.

• tmax: Time to reach Cmax (time required to reach peak concentration after dosing). It reflects the rate of drug entry into the body; faster absorption leads to shorter tmax.

• t1/2: Plasma concentration half-life (time required for the plasma concentration of a drug to decrease by half from its maximum value).

(II) Pharmacodynamic Parameters

• GIR-AUC(0-t): Area under the glucose infusion rate curve from administration to end of clamp at time t.

• GIR-AUC(0-T): AUC for the time of a dosing interval (area under the glucose infusion rate curve from dosing to a specified time period after dosing).

• GIRmax: Maximum glucose infusion rate.

• tGIRmax: Time to reach maximum glucose infusion rate.

VII. References

1. National Medical Products Administration. Guideline on the Design of Clinical Studies for Once-Daily Basal Insulin Biosimilars. January 28, 2022.

2. National Medical Products Administration. Technical Guideline for Clinical Pharmacology Studies of Biosimilars. February 11, 2022.

3. FDA. Guidance for Industry: Bioavailability and Bioequivalence Studies Submitted in NDAs or INDs – General Considerations – Draft Guidance. 2014.

4. European Medicines Agency. Guideline on non-clinical and clinical development of similar biological medicinal products containing recombinant human insulin and insulin analogues [EB/OL].

5. Liu Hui, Yu Hongling, Liu Jiali, Li Jiaqi, Tan Huiwen, Yu Yerong. The effect of establishing a hyperinsulinemic platform on endogenous insulin secretion in the euglycemic clamp. Chinese Journal of New Drugs. 2018, 27(02):167-172.

6. Soop M, Nygren J, Brismar K, Thorell A, Ljungqvist O. The hyperinsulinaemic-euglycaemic glucose clamp: reproducibility and metabolic effects of prolonged insulin infusion in healthy subjects. Clin Sci (Lond). 2000;98:367-374.

7. Heinemann L. Time-action profiles of insulin preparations. Mainz: Kirchheim & Company GmbH; 2004.

8. Liu Hui, Yu Hongling, Han Lina, Zhang Siqin, Liu Jiali, Li Jiaqi, Tan Huiwen, Yu Yerong. Quality assessment of euglycemic glucose clamp studies. Journal of Sichuan University (Medical Science Edition). 2019;50(4):588-593.

9. Benesch C, Heise T, Klein O, Heinemann L, Arnolds S. How to Assess the Quality of Glucose Clamps? Evaluation of Clamps Performed With ClampArt, a Novel Automated Clamp Device. J Diabetes Sci Technol. 2015 Jul;9(4):792-800.

10. National Medical Products Administration. Technical Guideline for Bioequivalence Studies of Chemical Generic Drugs Using Pharmacokinetic Parameters as Endpoints. February 10, 2022.

11. US Food and Drug Administration. GUIDANCE DOCUMENT. Scientific Considerations in Demonstrating Biosimilarity to a Reference Product. 2015.

12. Heise T., Zijlstra E., Nosek L., Heckermann S., Plum-Mörschel L., Forst T. Euglycaemic glucose clamp: what it can and cannot do, and how to do it. Diabet. Obes. Metab. 2016;18:962-972.