| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
1
Departments of Drug Disposition,
2
Bioavailability and Pharmacokinetics, and
3
Protein Optimization, Lilly Research Laboratories, Eli Lilly & Company, Indianapolis, IN 46285.
a Author for correspondence. Fax 317-277-9065; e-mail bowsher_ronald_r{at}lilly.com.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
) (1)(2)(3). This inversion causes insulin lispro to
have a reduced capacity for dimerization, which promotes the
dissociation of insulin hexamers directly to monomers
(3)(4). Consequently, insulin lispro has a
faster rate of absorption, higher peak serum concentrations, and a
shorter duration of action than regular human insulin after
subcutaneous injection (2)(3)(5).
For individuals with type 1 diabetes, insulin lispro has several
advantages, including improved postprandial control of blood glucose,
reduced frequency of hypoglycemic episodes, and therapeutic convenience
compared with Humulin® R (6)(7)(8).
|
The small structural difference between insulin lispro and insulin has frustrated attempts to specifically measure insulin lispro in the presence of endogenous insulin. Accordingly, a conventional insulin RIA has been used to date for measuring serum concentrations of insulin lispro after subcutaneous administration (5)(9). However, this analytical strategy permits the assessment of insulin lispro pharmacokinetics only in terms of total (insulin plus insulin lispro) "immunoreactive insulin". The specific determination of insulin lispro offers advantages over measuring total concentrations of immunoreactive insulin and would be particularly useful in assessing the pharmacokinetics of insulin lispro in patients with insulin resistance and in documenting factitious hypoglycemia from injection of insulin lispro. However, the lack of a suitable antiserum has precluded the development of an RIA specific for insulin lispro. We therefore initiated experiments to investigate the feasibility of developing an insulin lispro specific antiserum.
After immunization of guinea pigs with insulin lispro and screening of sera for antibodies with increased affinity and selectivity for insulin lispro vs native human insulin, we prepared a monospecific polyclonal antiserum by a novel immunoadsorption strategy using despentapeptide insulin (DPI),1 an analog of human insulin lacking the five terminal amino acids at the carboxy terminus of the B chain. We used the monospecific antiserum in the successful development of a competitive RIA.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
preparation of radiolabeled peptides.
Biosynthetic insulin and insulin lispro were radoiodinated by a
conventional lactoperoxidase method (10). We isolated the
[125I]-monoiodo-Tyr(A14) peptides by preparative
C18 reversed-phase HPLC using isocratic elution with 0.20
mol/L ammonium acetate–275 mL/L acetonitrile, pH 5.5. The specific
activity of [125I]-monoiodo-Tyr(A14) insulin and insulin
lispro was routinely 
350 µCi/µg.
production of anti-insulin lispro antiserum
Immunization and antiserum generation.
The insulin lispro
immunogen was prepared by dissolving reference standard insulin lispro
(LY275585, lot no. RS0195) in 9 g/L sodium chloride at a concentration
of 2 g/L. The immunogen solution was stored as 0.5-mL aliquots in
plastic vials at -20 °C. Antibodies were produced in Duncan-Hartley
guinea pigs at Covance (Denver, PA). For the initial immunization, the
immunogen was emulsified with an equal volume of Freund's complete
adjuvant. Twelve male guinea pigs were then immunized with 0.4 mg of
the immunogen by multiple subcutaneous nuchal injections. For booster
injections, the immunogen was emulsified with an equal volume of
Freund's incomplete adjuvant. The guinea pigs received subcutaneous
booster injections of 0.2 mg at 3, 6, and 9 weeks. Beginning at 12
weeks, the guinea pigs received booster injections of 0.1 mg of
immunogen at 1-month intervals. Antibody titers were assessed by RIA in
bleeds collected 10 days after each booster injection beginning 1 month
after the initial immunization.
Antiserum immunoadsorption.
Cross-reactive insulin antibodies
were removed by immunoabsorption of the antisera using DPI coupled to
Affigel-15. DPI-Affigel-15 affinity gel was prepared according to
the manufacturer's instructions. Briefly, 16 mL of a 500 g/L slurry of
gel was washed on a scintered glass funnel with cold deionized water,
dried to a moist cake, and then transferred immediately to a
polypropylene tube containing 8 mL of a 1 g/L solution of DPI prepared
in 100 mmol/L HEPES, pH 7.5, (1 mL protein/mL gel). After sealing the
tube, we incubated the mixture on an orbital rocker at 4 °C for
~16 h. Any remaining unreacted sites were blocked by incubating the
gel with 1 mL of 1 mol/L Tris, pH 7.5, for 1 h at 20 °C. Before
use, the gel was washed extensively with 0.1 mol/L HEPES, pH 7.5, and
stored in HEPES buffer containing 0.2 g/L sodium azide.
Two pools of antisera were prepared by combining bleeds 3–7 from guinea pigs 1221 and 1222. Each lot was diluted 1:1 with assay buffer, which consisted of 100 mmol/L sodium phosphate, 20 g/L EDTA, 1 g/L sodium azide, 0.5 mL/L Tween-20, and 1 g/L BSA, adjusted to pH 7.5; the final volumes of each diluted pool was ~30 mL. The DPI-Affigel-15 gel was equilibrated with an equal volume of assay buffer for 1 h at 20 °C before antiserum adsorption. The two antisera pools from guinea pigs 1222 and 1221 were mixed with a total of 1 and 10 mL, respectively, of a 500 g/L slurry of DPI-Affigel-15 and incubated on an orbital rocker at 4 °C for 16 h. After immunoadsorption, the antisera were centrifuged, filtered through a 0.45 µm filter, combined into a single lot, and stored at -70 °C. We diluted the pooled antiserum 1:10 000 with assay buffer for use in the RIA.
insulin lispro ria
RIA solutions.
All solutions except the stock solutions of
antiserum and peptide calibrators were stored at 4 °C. The assay
buffer used to dilute the tracer and antibody contained 100 mmol/L
sodium phosphate, 20 g/L EDTA (dipotassium salt), 1 g/L sodium azide,
0.5 mL/L Tween-20, and 1 g/L BSA, and was adjusted to pH 7.5
(11). A 1 mg/L solution of insulin lispro was prepared in
assay buffer containing 30 g/L BSA and 6 g/L NaCl (sample buffer) and
was stored frozen in 1-mL aliquots in polypropylene vials at
-20 °C. Calibration curves were prepared daily, using dilutions of
this stock solution in human serum. A solution of nonimmune guinea pig
serum was prepared by reconstituting one vial of serum with 5 mL of
Alpha-Q water and diluting to final a concentration of 20 mL/L with
assay buffer. We prepared the precipitation reagent of 20 mL/L goat
anti-guinea pig IgG and 60 g/L polyethylene glycol in a buffer
containing 100 mmol/L sodium phosphate, 20 g/L EDTA, 1 g/L sodium
azide, and 0.5 mL/L Tween-20, pH 7.5.
RIA procedure.
Each binding reaction (total volume, 400 µL)
was performed in 12 x 75 mm polypropylene tube and consisted of
100 µL of radioiodinated insulin lispro tracer (50 pg/tube), 100 µL
of antiinsulin lispro antiserum (diluted 1:10 000), and 200 µL of
human serum or polyethylene glycol-treated serum (12). The
nonspecific binding was determined by replacing the antiserum with
assay buffer. After the sample was mixed, we incubated the binding
reaction at room temperature for 18–24 h. The bound and free forms of
insulin lispro were separated by addition of 100 µL of 20 mL/L
nonimmune guinea pig serum in assay buffer, followed by 1 mL of a cold
precipitation reagent. Each tube was vortex-mixed thoroughly and
incubated at 4 °C for 1 h. After each tube was centrifuged at
3000g for 15 min at 4 °C, we decanted the aqueous phase
and measured the radioactivity in the precipitate in a gamma counter. A
VAX computer was used to analyze the RIA data by a weighted
four-parameter logistic model algorithm (13). The insulin
lispro concentration in test samples was estimated from a serum
calibration curve that ranged in concentrations from 0.01 to 100 µg/L
(1.7 to 17 200 pmol/L).
assay validation
The insulin lispro RIA was validated to support bioavailability,
bioequivalence, and pharmacokinetic studies in animals and man
(14). Intra- and interassay imprecision (CVs), recovery, and
serum stability of insulin lispro were assessed by measuring the
concentrations of insulin lispro in control samples that were prepared
by adding reference standard insulin lispro to serum from healthy
adults at concentrations of 0.1–100 µg/L. Serum controls were stored
frozen in aliquots at -20 °C. We compared the cross-reactivities of
insulin lispro and the endogenous peptides, insulin, proinsulin, and
C-peptide, as well as 40 other structurally related analogs to map the
binding epitope. In assays to assess stability, we measured the
concentration of insulin lispro in each of three serum samples at three
time points in duplicate. Insulin lispro was considered to be stable if
the recovery measured by RIA was ± 20% of the nominal
concentration.
assay comparison
For this comparison we measured the "free" insulin lispro
concentrations in serum samples from a clinical study designed to
evaluate different protamine formulations of insulin lispro. This study
was conducted in accordance with the ethics principles stated in the
latest version of the Declaration of Helsinki and applicable guidelines
for good clinical practice. The protocol was approved by the local
Institutional Review Board, and each participant gave informed consent.
In this glucose clamp study, 12 individuals with type 1 diabetes were
connected to a Biostator® to maintain euglycemia.
Before starting the study, each patient received a continuous
intravenous infusion of human insulin in their contralateral arm for
~3 h to attain a stable blood glucose of 0.9 ± 0.1 g/L (90 ±
10 mg/dL). Upon subcutaneous injection of a single dose of 0.3 units/kg
of one formulation of insulin lispro, the insulin infusion was
terminated, and blood samples were collected, without anticoagulant,
for up to 19 h. To remove anti-insulin antibodies from the
patients' sera, the test samples were treated with an equal volume of
ice-cold 200 g/L polyethylene glycol, followed by centrifugation
(12). The concentrations of insulin lispro were then
measured in the supernatants by the new RIA and a validated
conventional insulin RIA (9).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
). The final antiserum, prepared by pooling the two treated
antisera, displayed a >30-fold selectivity for insulin lispro with
negligible cross-reactivity with either human insulin or proinsulin at
concentrations as high as 100 µg/L (Fig. 3
).
|
|
epitope mapping (antiserum specificity)
We evaluated the cross-reactivity of 40 structurally related
peptides in competitive binding RIAs to evaluate the specificity of the
treated antiserum (Table 1
). The percentage of cross-reactivity was calculated as the
ratio of the ED50 of insulin lispro to the
ED50 (concentration of unlabeled peptide necessary to
produce 50% displacement of radiolabeled insulin lispro) of each
peptide. As expected, modifications to the carboxy terminus of the B
chain produced the greatest disruptions in antigenicity. Modifications
at other B-chain residues did not alter cross-reactivity with the
antiserum. A marked reduction in cross-reactivity occurred when the
lysine at B28 was substituted with either an acidic or neutral
aliphatic amino acid. Any modification that removed proline from the
B29 position ablated cross-reactivity with the antiserum. The
proinsulin analog of insulin lispro was only half as potent as insulin
lispro, indicating that a free B-chain carboxy terminus is required for
full immunoreactivity. These data suggest that the insulin lispro
antigenic determinant comprises charge, steric, and secondary
structural components. The endogenous human peptides, insulin,
proinsulin, and C-peptide, all displayed negligible cross-reactivity at
physiological and pharmacological concentrations (Table 1
and Fig. 3
).
|
assay performance
Calibration curve parameters.
A typical calibration curve for
insulin lispro prepared in human serum is shown in Fig. 3
. For 12 RIAs,
the percentage of nonspecific binding was 1.9% ± 0.15% (mean ±
SE), with a maximum binding of 40.5% ± 0.45%. The slope and
ED50 were 0.99 ± 0.01 and 0.94 ± 0.04 µg/L,
respectively.
Recovery.
Intra- and interassay CVs are reported in Table 2
. The intraassay CVs were 6.1–19%, and the interassay CVs were
2.6–13%. Recoveries (Table 3
) ranged from 96% to 108% at concentrations of 0.1–25 µg/L
(n = 4–12).
|
|
Linearity.
The linearity of dilution was established with a
sample of human serum supplemented with 100 µg/L of insulin lispro
and diluted 1:5, 1:10, 1:50, 1:100, and 1:200 with human serum. The
overall mean recovery was 93.4% ± 1.8% (mean + SE; n =
10).
Stability.
Insulin lispro was stable in serum for at least 6
months when stored frozen at -20 °C, which is consistent with a
previous report of stability in serum for up to 18 months at -20 °C
(9). Insulin lispro was stable in human serum for at least 7
days at 4 °C, 24 h at room temperature, and for at least three
cycles of freezing and thawing.
assay comparison
We compared the new RIA with a conventional insulin RIA that has
been used extensively to support clinical studies of insulin lispro
(9)(15). The serum concentrations determined by
both assays are plotted against each other in Fig. 4
, with the new RIA as the y variable and the
conventional insulin RIA as the x variable. When the data
were segregated according to the concentration of immunoreactive
insulin, linear regression analysis yielded two different
relationships. When the concentration of free immunoreactive insulin
was 
0.8 µg/L (138 pmol/L), the equation was: y =
0.93x + 0.03; r = 0.87. The slope, 0.93, was
not statistically significantly different from 1.0 (P =
0.331). These data demonstrated excellent agreement between assay
results for the new method and the standard insulin RIA. Therefore,
when the majority of the immunoreactive insulin in the test sample was
insulin lispro, there was close agreement in assay results between
these two methods. However, when the concentration of free
immunoreactive insulin was >0.8 µg/L the equation was:
y = 0.52x - 0.09; r =
0.94. In this case, the slope, 0.52, differed significantly from 1.0
(P <0.001). Thus, when insulin is a major fraction of the
serum "insulin immunoreactivity", the insulin RIA detects both
insulin and insulin lispro. This produces a lack of agreement between
the RIAs, with a slope value no longer close to unity.
|
The conclusions from linear regression analysis were confirmed by a concordance analysis that evaluates agreement between paired observations (16). The concordance coefficient (rc) for the pairs of observed insulin lispro and insulin concentrations in samples with low baseline concentrations of immunoreactive insulin was 0.862, which confirms a high degree of agreement (for test of rc = 0, P <0.001). The value of rc for the samples with concentrations of immunoreactive insulin >0.8 µg/L was 0.397 (P = 0.083), which confirmed the lack of agreement between the assays for this subset of the samples.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
We selected the guinea pig for development of an insulin
lispro-specific antiserum because species specificity is recognized to
play an important role in the successful generation of anti-insulin
antibodies (17). Because all antisera displayed different
cross-reactivities with native human insulin (Fig. 2
), we developed a
novel immunoadsorption strategy, using DPI coupled to Affigel-15.
Because DPI lacks the five terminal amino acids at the carboxy terminus
of the B chain, the affinity gel is capable of removing all antibodies
against epitopes common to insulin and insulin lispro except for those
directed against the carboxy terminus of the B chain. After
immunoadsorption, the antiserum displayed a >30-fold selectivity for
insulin lispro, with negligible cross-reactivity with human insulin,
proinsulin, and C-peptide at concentrations up to 100 µg/L (Table 1
and Fig. 3
).
Cross-reactivity experiments indicated that the amino acid sequence
-Xaa-Pro-Thr-COOH (where Xaa is a basic amino acid) at B28–B30 is the
antigenic determinant. First, modifications to this region produced the
greatest disruption in antigenicity (Table 1
), and native human insulin
displays cross-reactivity. In contrast, amino acid substitutions at
residues near the amino terminus of the B chain had no effect on
antigenicity. Second, a basic amino acid, such as lysine, arginine, or
ornithine, was required at residue B28 for full cross-reactivity.
Cross-reactivity was markedly reduced when the lysine at B28 was
replaced with either an acidic or a large neutral aliphatic amino
acid. Third, a proline is required at B29, because any
modification that removes proline ablated cross-reactivity with the
antiserum. Optimal cross-reactivity was achieved when the peptide chain
terminated at B30; this indicated that a free B-chain carboxy terminus
is required for full antigenicity.
In summary, we describe the first assay that permits the specific determination of insulin lispro and is suitable for measuring insulin lispro in the presence of endogenous or pharmacological concentrations of both proinsulin and insulin in human and canine sera. We conclude the new RIA is a valid method that will be useful in future studies to assess the serum concentrations of insulin lispro.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  M. Moriyama, N. Hayashi, C. Ohyabu, M. Mukai, S. Kawano, and S. Kumagai Performance Evaluation and Cross-Reactivity from Insulin Analogs with the ARCHITECT Insulin Assay, Clin. Chem., July 1, 2006; 52(7): 1423 - 1426. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. E. Owen and W. L. Roberts Cross-Reactivity of Three Recombinant Insulin Analogs with Five Commercial Insulin Immunoassays Clin. Chem., January 1, 2004; 50(1): 257 - 259. [Full Text] [PDF]
|
|
|
|
|
|
F. S. Apple, M. Murakami, M. Panteghini, R. H. Christenson, F. Dati, J. Mair, and A. H.B. Wu International Survey on the Use of Cardiac Markers Clin. Chem., March 1, 2001; 47(3): 587 - 588. [Full Text] [PDF]
|
|
|
|
|
|
R. Sapin, V. Le Galudec, F. Gasser, M. Pinget, and D. Grucker Elecsys Insulin Assay: Free Insulin Determination and the Absence of Cross-Reactivity with Insulin Lispro, Clin. Chem., March 1, 2001; 47(3): 602 - 605. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
, before first treatment; 
, after third treatment). In contrast,
the cross-reactivity of human insulin was markedly reduced by
immunoadsorption (
, before first treatment; 
, after third
treatment). (Bottom), antiserum from guinea pig 1222 was
treated with DPI-Affigel-15 as described in Materials and
Methods. Each data point represents the mean of duplicate
determinations. Data are plotted as the ratio of the bound cpm in the
presence of unlabeled peptide (B) relative to the maximum
binding in the absence of unlabeled peptide
(B0). As shown, the cross-reactivity of insulin
lispro was unaffected by immunoadsorption (
),
both insulin (