C-peptide Correction Method to Determine Exogenous Insulin Levels in

Pharmacokinetic Studies Using Technosphere Insulin

Anders H. Boss, Mark T. Marino, James P. Cassidy, Robert A. Baughman, Peter C. Richardson
MannKind Corporation, Valencia, CA, United States

TI Insulini,j = Measured Insulini,j Insulini,j

An example of the C-peptide insulin relationship is displayed in the figure
on the right. Conclusion: We suggest that this novel method may be used to
replace determination of individuals C-peptide elimination experimentally.

0
112

115

123

124

Insulin (IU/mL)

C-peptide concentrations allow the estimation of an individuals endogenous

insulin concentration from the equation above. Exogenous insulin is then
calculated by subtracting the endogenous insulin from the total insulin
measured as per the equation below.

10

30

All insulin and C-peptide levels were evaluated by Bio Analytical Research Corporation (BARC), Lake Success, NY.
The insulin method was an ElectroChemiLuminescence assay (ECLIA) with a lower limit of detection of 0.5 U/mL. The
C-peptide method was a competitive ChemiLuminescence assay with a limit of sensitivity of 0.5 ng/mL.
Insulin and C-peptide samples were collected prior to the administration of insulin or GLP-1 and up to 8 hours after
administration. Insulin and C-peptide values prior to and more than 6 hours after insulin administration were used to
evaluate the insulin:C-peptide relationship. All insulin and C-peptide values were used for analysis in subjects who
received GLP-1. Time points before insulin administration and greater than 6 hours after insulin administration were
used to ensure that only endogenous insulin was present. In previous studies, it has been demonstrated that TI is rapidly
absorbed (Tmax ~14 minutes) and rapidly eliminated, similar to intravenous insulin (T ~60 minutes) suggesting that
98% of TI was cleared from the plasma.
The insulin:C-Peptide relationship was evaluated by a mixed effect linear regression model. The fixed effect was the
population mean slope of the insulin:C-peptide relationship () and the intercept. The random effects were each
subjects deviation from slope (i) and the measurement error (i,j). The measurement error was assumed to be
independent, normally distributed, and comprised of an additive and constant coefficient of variation error terms.
In this equation, i represents the i-th individual and j represents the j-th administration of insulin or GLP-1 in a crossover
manner. The analysis was run in NONMEM, version VI level 2.0 on the equation below on a HP xw4550 workstation
and the data were graphed in R version 2.7.1. The statistical model output from NONMEM included:

20
10
0
101

102

106

108

30
20
10
0
2

Cpeptide (ng/mL)

INTRODUCTION

Population or mean estimates for the insulin:C-peptide relationship ()

Interindividual variance for the insulin:C-peptide relationship (i)
Estimates for the additive and constant coefficient variation error terms
Estimates of standard error for each of the above estimates
Estimate for each individuals insulin:C-peptide relationship (i)
The goodness of fit was determined by the objective function
Equation 1
(extended least squares), the number of significant digits for the
parameter estimates, and the visual assessment of the fit of the Insulini,j = Intercepti + Slopei x C-peptidei,j
data and residuals.

R E S U LT S

C-peptide is co-secreted with insulin from the pancreatic beta () cell in a one-to-one molar ratio. The relationship
between C-peptide and insulin was first described by Steiner and Oyer in 19671, and the peptide structure of proinsulin, insulin, and C-peptide was described by Chance, et al. in 1968.2 The interactions between C-peptide and insulin
have been extensively explored with a focus on evaluating C-peptide secretion and kinetics as a marker for insulin
secretion and pancreatic functioning. C-peptide correction methods have been described, but were based on basal
fasting insulin and C-peptide levels and an experimental procedure in humans which utilized somatostatin.3 Some
authors have expressed doubt that a useful relationship between plasma insulin and C-peptide would be developed
due to inherent differences in the kinetic properties of insulin and C-peptide.4
Evaluating the pharmacokinetic (PK) profile of exogenously administered human insulin from new devices and
formulations is difficult for many reasons. Evaluation of subjects with type 1 diabetes requires clinical research centers
that can perform the study and maintain good clinical care. Clinical research centers which perform hyperinsulinemic
euglycemic clamp studies using rapid acting analogues are often plagued by an inability to completely suppress
endogenous insulin secretion in healthy and type 2 subjects. Studies have been done using somatostatin suppression and
are associated with adverse effects.
For these studies, each subjects own C-peptide:insulin ratio was used during the administration of oral meals or during
an insulin clamp procedure using a rapid acting insulin analogue, which can be distinguished from native human insulin in
serum assays. In several clinical trials of Technosphere Insulin (TI) Inhalation Powder and a single trial of Technosphere
inhaled GLP-1, a strong linear relationship was demonstrated between simultaneously measured C-peptide and insulin
levels which varied between individuals. This relationship was observed only with endogenous insulin, either prior to or
several half-lives after the administration of insulin or GLP-1. It was used to back-calculate the amount of insulin
resulting from either exogenous administration or endogenous secretion during administration of exogenous insulin. This
method can be useful for evaluation of PK profiles of new formulations and delivery systems for human insulin. The
intraindividual variation in the C-peptide:insulin relationship may also be a useful technique to assess beta cell function,
as well as hepatic insulin extraction.

Study 1: 12 healthy volunteers

The predicted values for , the variance of i, the error terms, the standard
error for each parameter, and the estimate for each subjects insulin:C-peptide
relationship (i) are listed in Table 1. The model was fit with Equation 1. The
intercept was not statistically distinct from 0. In the modeling procedure, the
overall fit of the data was not found to improve with a term included for
interindividual variation which indicated that the intercept was most likely close
to 0 for all study subjects. The estimate for is expressed as U/ng. When
converting to molar equivalents, the ratio becomes ~0.08, indicating that the
molar ratio of C-peptide to insulin is approximately 12.5. This is consistent with
previously reported values of fasting and postprandial C-peptide:insulin ratios.

Table 1

Study 1

Parameter Estimates of the Insulin C-peptide

Relationship in Healthy Volunteers

Parameter

Estimate ( SE)

(U/ng)

3.80 (0.286)

Intercept

-0.120 (0.255)

j (U/ng) variance

0.57

Additive error

0.69

Constant C.V. error

16.4%

The interindividual variability of the estimates ranged from 2.76 to 5.21 U/ng, almost a two-fold difference in the
C-peptide to insulin relationship. This is shown in Figure 1 which displays all C-peptide and insulin data for each subject
and the regression line showing the linear relationship. Total insulin concentrations during the administration of
exogenous insulin were determined by calculating the estimated endogenous insulin for each subject from their
respective insulin:C-peptide relationship and then subtracting that amount from the total insulin. The results of this
method are shown in Figure 2.
Figure 1. Insulin vs C-peptide plotted by
subject demonstrating variation in the slope of
the relationship between insulin and C-peptide.

Figure 2. Total and exogenous insulin levels

as determined by C-peptide correction. Total
insulin (red), exogenous insulin (blue).
Insulin With and Without Cpeptide Correction
by Subject and Visit

Insulin and Cpeptide Relationship

Data and Regression Line by Subject
2

132

133

0123

0123

0123

0123

0123

0123

VISIT 2 VISIT 4 VISIT 4 VISIT 4 VISIT 4 VISIT 4 VISIT 4 VISIT 4 VISIT 4 VISIT 4 VISIT 4 VISIT 4
101
102
106
108
112
115
123
124
128
132
133
136

136

250
30

AND

M AT E R I A L S

200
150

20

100
10

Two studies of inhaled insulin were conducted using hyperinsulinemic-euglycemic clamp procedures with a rapid acting
analogue, one with healthy subjects (MKC-TI-114) and one with subjects who had COPD (MKC-TI-015). One trial
studied healthy subjects given a meal challenge (MKC-TI-141). The single trial for GLP-1 (MKC-253-002) studied
subjects with type 2 diabetes.

0
112

115

123

VISIT 3 VISIT 3 VISIT 3 VISIT 3 VISIT 3 VISIT 3 VISIT 3 VISIT 3 VISIT 3 VISIT 3 VISIT 3 VISIT 3
101
102
106
108
112
115
123
124
128
132
133
136

124

30
20
10

250

Insulin (U/mL)

Technosphere particles are a proprietary composition of FDKP (fumaryl diketopiperazine) that is useful in the
formulation of peptides for delivery by inhalation. Peptides adhere to the surface of FDKP particles and are inhaled
via a patient-activated, high resistance device without propellant. FDKP rapidly dissolves at physiologic pH and
releases the peptide which is absorbed via the deep lung producing the rapid appearance of the peptide in
the plasma.

50

Insulin (IU/mL)

Data on C-peptide and insulin are from 4 clinical trials: 3 studies involved the administration of TI Inhalation Powder to
healthy volunteers and subjects with COPD, and 1 study evaluated subjects with type 2 diabetes who inhaled GLP-1
administered to via the Technosphere platform.

Study 2: 18 Subjects with COPD and 20 Subjects without COPD

716

218

Table 2

Study 2

Parameter Estimates of the Insulin C-peptide

Relationship in Subjects With and Without COPD

201

717

222

202

723

508

203

805

513

207

806

517

209

807

716

814

605

608

210

211

609

708

212

215

218

712

216

40
30

(U/ng)

4.05 (0.259)

Intercept

-1.34 (0.23)
0.278

Additive error

0.152

Constant C.V. error

20.6%

200
150
100
50
0

0
101

102

106

VISIT 2 VISIT 2 VISIT 2 VISIT 2 VISIT 2 VISIT 2 VISIT 2 VISIT 2 VISIT 2 VISIT 2 VISIT 2 VISIT 2
101
102
106
108
112
115
123
124
128
132
133
136

108

20
10

105

112

113

115

117

120

125

138

157

4 8 th E A S D A n n u a l M e e t i n g S e p t e m b e r 2 9 O c t o b e r 2 , 2 0 0 9 V i e n n a , A u s t r i a P o s t e r P r e s e n t a t i o n 5 8 0

Total insulin

0123

0123

Time (hrs)

0123

0123

Exogenous insulin

210

609

211

708

212

712

215

0
2 4 6 810

2 4 6 810

112

2 4 6 810

2 4 6 810

113

0123

2 4 6 810

Cpeptide (ng/mL)

115

117

0123

120

0123

125

138

157

0123

167

0123

Time (hrs)

Total insulin

Exogenous insulin

Figure 5. Insulin vs C-peptide plotted by

subject demonstrating variation in the slope of
the relationship between insulin and C-peptide.
Insulin and Cpeptide Relationship
Data and Regression Line by Subject
0 1 2 3 4 5
16

17

0 1 2 3 4 5
18

19

20

200
150
100
50
0
11

12

13

14

15

10

200
150
100
50
0

150
100
50
0
1

32.9%

200
150
100
50
0
0 1 2 3 4 5

0 1 2 3 4 5

0 1 2 3 4 5

Cpeptide (ng/mL)
A Phase 1, single-center, open-label, randomized, crossover design
clinical trial in healthy normal volunteers was
conducted to evaluate the bioavailability of TI
Figure 6. Insulin vs C-peptide plotted by subject demonstrating the
variation in the slope of the relationship between insulin and C-peptide.
inhalation powder with Gen2B Inhaler
compared to a MedTone
Relationship Between Insulin and C-peptide by Individual
0
5
10 15
0
5
10 15
0
5
10 15
Inhaler Model C after a meal.

Forty-eight subjects were administered a meal

and had suitable insulin and C-peptide
concentrations for analysis. The predicted
values for , the variance of i, the error
terms, the standard error for each parameter
and the estimate for each individuals
Study 4

Parameter Estimates of the Insulin C-peptide

Relationship in Healthy Volunteers

Estimate ( SE)

j (U/ng) variance

0.06

Intercept variance

0.21

Diabetologia . 2009;52(Supp.1).

100

200

300

100

200

1115

1125

1134

1022

1025

1086

1101

300

150
100
50
0

150
100
50
0
1001

1002

1007

1015

150
100
50
0
100

200

300

Total insulin

Time (min)

100

200

300

Exogenous insulin

CONCLUSIONS

35.1%

150
100
50
0

150
100
50
0

150
100
50
0

1160

1162

1163

1166

1169

1170

1126

1134

1142

1144

1145

1153

1157

1108

1111

1113

1115

1118

1122

1125

1083

1085

1086

1087

1100

1101

1105

1057

1058

1060

1066

1067

1078

1079

1015

1016

1022

1025

1034

1047

1051

1001

1002

1003

1007

1011

1012

1014

150
100
50
0

150
100
50
0

150
100
50
0

150
100
50
0
0

10

15

The relationship between C-peptide and insulin has been explored since the mechanism of insulin secretion was
first elucidated. The focus has been on how to use the pharmcokinetic profile of C-peptide to understand and
predict pancreatic insulin secretion. Because C-peptide does not undergo hepatic uptake after secretion into the
portal vein while insulin is extracted by the liver, its plasma profile levels might be useful to determine the
pancreatic response to various stimuli which enhance insulin secretion. Although many studies have been done
and progress has been made using C-peptide, the use of simultaneous plasma C-peptide and insulin values for
evaluation has not been extensively studied. Clearance mechanisms for insulin and C-peptide are very different.
Insulin undergoes a first pass effect in the liver and subsequent elimination which has been described by
several authors as mono- or bi-exponential (one-compartment or two-compartment models). C-peptide is not
extracted to a significant degree by the liver before it undergoes bi-exponential elimination (two-compartment
model). Several authors have noted these differences and stated that analyzing the relationship would be
limited because of their different kinetic profiles7, or can only be approximated when insulin and C-peptide
levels are changing in response to a stimulus.8 Many studies which evaluated insulin and C-peptide used the
administration of bolus glucose in a step-function manner which rapidly induced insulin and C-peptide release.
This is not consistent with insulin and C-peptide secretions observed with the administration of a meal to the GI
tract. While the alteration in first-phase insulin secretion from a glucose infusion is diminished in subjects with
type 2 diabetes compared to healthy subjects, some authors question how this may relate to insulin secretion
secondary to a meal in both populations.9 Whether the methodology in these studies may have affected the
relationship is unknown. This method may not have been used because of its perceived limited utility.

200

Study 4: 11 healthy volunteers given insulin

and a meal

C.V. error

0
1105

40

20

Constant C.V. error

Table 4

Total and Exogenous Insulin Levels by C-peptide Correction

216

40

8.68 (0.454)

0123

209

608

60

10

-2.99 (0.758)

0123

207

605

20

20

(U/ng)

203

517

814

80

167

30

Intercept

Cpeptide (ng/mL)

513

807

Figure 7. Total and exogenous insulin levels as determined by C-peptide

correction. Total insulin (red), Exogenous insulin (blue).

80

150

806

60

100
50

508

202

105

Twenty subjects with type 2 diabetes received Technosphere Inhalation

Powder as a vehicle control, GLP-1 or exenitide in a crossover trial in
both the fed and fasted states. C-peptide and insulin results from each
separate administration were pooled for this analysis. While the
intercept was statistically distinct from 0 in the modeling procedure, the
overall fit of the data was not found to improve with a term included for
interindividual variation indicating that the intercept was most likely
similar for all study subjects. The estimate for in this study is
significantly higher than that estimated Table 3
Study 3
Parameter Estimates of the Insulin C-peptide
for Study 1. The different insulin:
Relationship in Subjects With Type 2 Diabetes
C-peptide relationships in Study 3 may
Estimate ( SE)
Parameter
reflect differences in hepatic insulin
23 (1.23)
(U/ng)
extraction between the two studies or
-9.89 (1.23)
Intercept
populations. Because no exogenous
0.121
insulin was given in this study, there was j (U/ng) variance
Additive error
15.5
no need to use this method.

Parameter

200

201

250

30

222

805

10

40

j (U/ng) variance

723

20

Estimate ( SE)

Parameter

717

(CONTD)

insulin:C-peptide relationship (i) are listed in

Table 4. The model was fit with Equation 1
with the addition that the intercept was
statistically distinct from 0 in the modeling
procedure. The overall fit of the data was
found to improve with a term included for
inter-individual variation for slope and
intercept. This is also clearly shown in
Figure 6 which displays all of the C-peptide
and insulin data for each subject along with
the regression line through the data showing
the linear relationship. Total insulin
concentrations during the administration of
exogenous insulin were done by calculating
the estimated endogenous insulin for each
individual from their respective
insulin:C-peptide relationship and then
subtracting that amount from the total insulin.
The results of this method are shown for
subjects from one visit (Figure 7).

Thirty-seven subjects had suitable jnsulin and C-peptide concentrations for analysis. The predicted values for , the
variance of i, the error terms, the standard error for each parameter, and the estimate for each subjects insulin
C-peptide relationship (i) are listed in Table 2. The model was fit with Equation 1. While the intercept was statistically
distinct from 0 in the modeling procedure, the overall fit of the data was not found to improve with a term included for
interindividual variation
Figure 3. Insulin vs C-peptide plotted by
Figure 4. Total and exogenous insulin levels
indicating that the
subject demonstrating the variation in the
as determined by C-peptide correction. Total
intercept was most likely
slope of the relationship between insulin
insulin (red), exogenous insulin (blue).
similar for all subjects in
and C-peptide.
the study (Figure 3). The
Insulin With and Without Cpeptide Correction
Insulin and Cpeptide Relationship
estimate for was lower
by Subject and Visit
Data and Regression Line by Subject
0123
0123
0123
2 4 6 810
2 4 6 810
2 4 6 810
than that observed in
subjects with type 2
80
40
60
diabetes in Study 3
30
40
20
(Figure 4). The lower
20
10
0
C-peptide levels may
0
80
reflect differences in
40
60
hepatic insulin extraction
30
40
20
between the two studies
20
10
0
or populations.
0

Study 3: 20 subjects with type 2 diabetes

128

METHODS

R E S U LT S

(CONTD)

Insulin (U/mL)

R E S U LT S

(CONTD)

Insulin (U/mL)

Insulini,j = Intercepti + Slopei x C-peptidei,j

M AT E R I A L S

Insulin (U/mL)

Background and aims: Determining the concentration time profile of exogenous-administered insulin in pharmacokinetic
studies usually requires the use of patients with poor or absent -cell function (type 1 diabetes mellitus), labeled insulin,
or insulin clamp studies. These methodologies are difficult and limit the ability to explore the kinetics of novel insulins in
broad populations. Although baseline C-peptide corrections have been used, they are not robust or precise. Materials
and methods: We report here the use of repeated intraindividual correlations between insulin and C-peptide to assess
exogenous insulin contribution in individuals with intact -cell function analyzed in several clinical trials (69 subjects from
3 trials). C-peptide and insulin are secreted in a 1:1 molar ratio by the pancreas, but while insulin is removed by the
liver, C-peptide is not. The clearance, and ultimate ratio, differs among individuals. Data for the analysis were obtained
when exogenous insulin was cleared from the plasma (ie, prior to dosing and 6 h or more after dosing with TI). The
relationship was analyzed with a linear mixed effect model where the fixed
Insulin and Cpeptide Relationship
effect was the population mean for intercept and slope and the random
Data and Regression Line by Subject
2
4
6
2
4
6
effect was the individual deviation from those means. Results: As expected,
128
132
133
136
a general linear relationship was found to exist between C-peptide and
30
insulin, but the slope varied by individual.
20

AND

Insulin (U/mL)

METHODS

Insulin (U/mL)

ABSTRACT

10

15

C-peptide (ng/mL)

10

15

10

15

We developed this method to specifically address the problem of distinguishing the pharmacokinetic profiles of
exogenously administered human insulin from endogenous human insulin. This method has proven to be useful in
understanding and describing the pharmacokinetic profile of inhaled insulin. Baseline levels of insulin are
corrected to very low levels. Endogenous levels of insulin diminish during exposure to exogenous insulin and
return towards normal when the administered exogenous insulin has been cleared from the body. This has
facilitated the study of bioavailability, bioequivalence, and pharmacokinetics/pharmacodynamics using healthy
volunteers who are given meals sufficient to prevent hypoglycemia. These studies can be performed safely and
rapidly provide useful and accurate data during the development of a human insulin product. Of further interest
is the C-peptide:insulin relationship in the context of severity and stage of disease, and the evaluation of hepatic
insulin extraction in relationship to meals, disease, and other therapies.

REFERENCES
1. Steiner DF, Oyer PE. The Biosynthesis Of Insulin And A Probable Precursor Of Insulin By A Human Islet Cell Adenoma. Proc
Natl Acad Sci U S A. 1967;Feb;57(2):473480.
2. Chance RE, Ellis RM, Bromer WW. Porcine Proinsulin: Characterization and Amino Acid Sequence. Science. 1968;
161(3837):165-167.
3. DR Owens, ed. Human Insulin: Clinical pharmacological studies in normal man. Lancaster, UK: MTP Press; 1986.
4. Polonsky KS, Rubenstein AH. C-Peptide as a measure of the secretion and hepatic extraction of Insulin. Diabetes. 1984;
33:486-494.
5. Humphriss DB, Stewart MW, Berrish TS, Barriocanal LA, Trajano LR, Ashworth LA, Brown MD, Miller M, Avery PJ, Alberti KG,
Walker M. Multiple metabolic abnormalities in normal glucose tolerant relatives of NIDDM families. Diabetologia. 1997;
40:11851190.
6. Sakai T, Ohneda A., Nihei J, Kobayashi, T. Analysis of Insulin Secretion Based On Changes In Plasma Insulin And 0-Peptide In
Man. Tohoku J Exp Med. 1982;138(4),427-440.
7. Polonsky KS, Rubenstein AH. C-Peptide as a measure of the secretion and hepatic extraction of Insulin. Diabetes. 1984;
33:486-494.
8. DR Owens, ed. Human Insulin: Clinical pharmacological studies in normal man. Lancaster, UK: MTP Press; 1986:138.
9. Caumo A, Luzi L. First-phase Insulin secretion: does it exist in real life? Considerations on shape and function. Am J Physiol
Endocrinol Metab. 2004; 287:E371E385.

Contact: M a r k M a r i n o 2 0 1 . 9 8 3 . 5 2 3 8 MannKind Corporation 61 S o u t h P a r a m u s R o a d P a r a m u s , N J 0 7 6 5 2