`Received November 16, 1995
`Accepted February 27, 1996
`Local Tolerance of Subcutaneous Injections
`
`© 1996 J. Phann. Pharmacol.
`
`J. FRANSSON•t AND A. ESPANDER - JANSSON•
`
`*Pharmacia and Upjohn, S -112 87 Stockholm, tDepartment of Pharmaceutics, Faculty of Pharmacy, Uppsala
`University, Box 570, S-751 23 Uppsala, Sweden
`
`Abstract
`Human insulin-like growth factor I (hlGF-1) has several possible clinical applications. Because subcutaneous
`administration of the drug can cause pain, local tolerance to injection of different formulations with or without
`hlGF-l has been investigated in man using isotonic saline solution as reference.
`The formulations, made isotonic with NaCl, ranged in pH from 6 to 7 with phosphate buffer concentrations
`of 5 to 50 mM. The local tolerance after injection was assessed as injection pain on a visual analogue scale, pain
`duration and local tolerance (redness, paleness and oedema). The discomfort at the injection site was lowest
`with 10 mM phosphate, pH 7. Injection of buffer at pH 6 (50 mM phosphate) caused significantly more pain
`than using 10 mM phosphate, whereas the pain at pH 6 using 10 mM phosphate did not differ significantly from
`that experienced on injection of the solution at pH 7 using either IO or 50 mM phosphate. hIGF-1 itself did not
`seem to cause pain.
`We concluded that for subcutaneous injections al non-physiological pH, the buffer strength should be kept as
`low as possible to avoid pain upon injection. We also hypothesize that when a non-physiological pH must be
`used for stability reasons, a lower buffer strength enables more rapid normalization of the pH al the injection
`site.
`
`Human insulin-like growth factor (hIGF-1) has insulin-like and
`growth-promoting effects and there are possibilities for future
`clinical use of recombinant hIGF-1 in a wide variety of
`applications. hlGF-1 is most stable at pH 6 (Fransson et al
`1994) and a parenteral formulation at this non-physiological
`pH was developed for subcutaneous administration. The severe
`pain experienced by several patients in early clinical trials on
`injection with hIGF-1 has been investigated in this study. In
`numerous studies with other drugs (lpp et al 1990; Zindel
`1989; Gazzaniga et al 1993) subjects have reported pain upon
`subcutaneous injection. Factors that might cause pain include
`the injection volume and the speed of injection (Frenken et al
`1994), osmolality (Doenicke et al 1992), the pH of the
`formulation (Barnet & Kapp 1992), injection site (Ipp et al
`1990), the size and quality of the injection needle (Coley et al
`1987), the presence of irritating substances (Frenken et al
`1993) and the temperature of the solution (Ross & Soltes
`1995).
`The purpose of buffers in pharmaceutical formulations is (o
`maintain a stable pH, usually that at which the drug is most
`stable. The ability of a buffer to maintain a pH value is
`dependent on the pK. of the buffer, on the pH and on the
`concentration (Flynn 1980). We hypothesi.zed that the injection
`pain could be reduced if a formulation with a lower buffer
`capacity was used for hlGF-1.
`Several clinical trials on analgesics have shown that pain
`scales can be a reliable and objective method of assessing the
`analgesic properties of drugs (Banos et al 1989). It is generally
`accepted that visual analogue scales are more sensitive and
`accurate than other measures, thus being generally advised in
`the evaluation of the intensity of pain (Langley & Sheppard
`1985).
`
`Correspondence: J. Fransson, Phannacia and Upjohn, S-112 87
`Stockholm, Sweden.
`
`The aim of this study was to evaluate how pH, buffer con(cid:173)
`centration and presence of hJGF-1 affect local tolerance to
`subcutaneous injection of the solution.
`
`Materials and Methods
`
`Study design
`The investigation was designed as a double-blind, randomized,
`IO x IO Graeco-Latin square with the three factors: subject
`(1- 10), site of injection (1- 10) and injection order number (I(cid:173)
`I 0). Eight formulations and two identical reference solutions
`containing 150 mM NaCl were compared and were balanced in
`respect of the site and of the injection number in the sequence
`of ten injections given to each subject. All injections were to
`be in 4 x 4 cm regions on the lower left and right arms, which
`are relatively sensitive to pain.
`A published IO x IO Graeco-Latin square (Peng 1967) was
`taken as a basis (Table I). The ten rows of this square corre(cid:173)
`spond to subjects, the ten columns correspond to injection
`sites, the ten Latin letters A-J correspond to formulations, and
`the ten numerical indices correspond to the injection number in
`the sequence of injections given to each subject. The assign(cid:173)
`ments of subjects to rows, of injection sites 10 columns. and of
`letters A-J to formulations, were performed randomly. The
`resulting assignments were then transferred to personal maps
`specifying the site for each injection in each subject.
`
`Formulations
`Although phosphate is not ideal at pH 6, because of its low
`buffer capacity at this pH, sodiwn phosphate was chosen as
`buffer, because citrate buffer causes p3in (Frenken et al 1993).
`The compositions of the eight formulations and the reference
`solution are listed in Table 2. The formulations were prepared
`by mixing disodiurn phosphate, monosodiurn phosphate and
`
`FRESENIUS EXHIBIT 1047
`Page 1 of 4
`
`
`
`LOCAL TOLERANCE OF SUBCUTANEOUS fNJECTIONS
`Table I. The Graeco-Latin square experimental design. The columns corTespond to the injection sites, the rows corTespond
`to the subjects and the indices corTcspond to the injection number.
`
`1013
`
`Row
`(subject no.)
`
`Column
`(site of injection)
`
`6
`8
`10
`3
`7
`9
`5
`4
`2
`I
`
`8
`
`A,
`) 7
`16
`f1 0
`H•
`G9
`Ds
`Bs
`C2
`E1
`
`5
`
`Es
`B2
`1,
`J,
`G,o
`Hs
`A9
`c6
`D1
`F•
`
`7
`
`89
`Fa
`C1
`12
`J,
`Aio
`H6
`D1
`E.
`Gs
`
`3
`
`H1
`C9
`Gs
`o.
`13
`12
`8 10
`E,
`fs
`~
`
`4
`
`C,o
`H,
`D9
`As
`Es
`
`,.
`
`)3
`f 2
`G6
`8 1
`
`JO
`
`) 4
`D,o
`H2
`E9
`Ba
`F6
`Is
`G1
`A1
`c,
`
`6
`
`16
`ls
`E,o
`H1
`f9
`Cs
`G1
`A•
`B,
`Di
`
`9
`
`Ds
`E6
`F,
`G,
`A2
`83
`c.
`Hs
`1,o
`J9
`
`2
`
`f3
`a.
`As
`86
`C1
`D,
`Ei
`J,o
`H9
`Is
`
`G2
`A3
`84
`Cs
`D6
`E1
`F,
`)9
`Js
`H,o
`
`sodium chloride in water for injection to give isotonic solu(cid:173)
`tions of different pH and buffer concentration. hIGF-1 was
`added to some of the solutions by ultrafiltration.
`
`Study pcpulation
`Ten healthy adult male volunteers, aged 18-40 and sensitive to
`injection pain, were recruited to the study. All potential par(cid:173)
`ticipants were injected with 0·2 ml o f a vehicle known to
`inflict pain on injection (pH 6, 50 mM phosphate) over 20 s at
`a pre-study medical examination. At 30 s after injection each
`subject assessed the injection pain on a I 00 mm horizontal
`visual analogue scale (0 mm = no pain, I 00 mm= severe
`pain). Subjects that marked > 20 mm on this scale were eli(cid:173)
`gible for inclusion in the study.
`
`Injection procedure
`After randomization, all subjects received a total of ten
`
`injections (0·2 ml injection - 1) . The ten injections, repre(cid:173)
`senting the eight formulations and the two sodium chloride
`references, were injected subcutaneously using a 0·5 ml 28 G
`Microfine syringe (Beckton and Dickinson, USA) with a fixed
`cannula. The solutions were stored at 2-8°C until the day of
`use, when they were equilibrated to room temperature before
`injection. All injections were given subcutaneously over 20 s
`by one person.
`For each subject each of the ten injections w~s made every
`15 min in a specified order and site on the lower left and right
`
`Table 2. The pH and concentrations of sodium phosphate buffer,
`sodnun chlonde and hlGF-1 m the formulations.
`
`Formulation
`
`pH
`
`Phosphate
`(mM)
`
`Sodium chloride
`(mM)
`
`hlGF-1
`(mg ml - ')
`
`A
`B
`C
`D
`E
`F
`G
`H
`1•
`J•
`
`60
`6-0
`7-0
`7-0
`6-0
`60
`6-0
`7-0
`8
`8
`
`50
`JO
`50
`10
`5
`50
`10
`50
`0
`0
`
`112
`145
`112
`145
`ISO
`112
`145
`112
`150
`150
`
`0
`0
`0
`0
`0
`5
`5
`5
`0
`0
`
`•Reference solutions
`
`arms. The subjects filled in assessment sheets after each
`injection. In order to estimate the local tolerance to injection,
`the variables evaluated were: injection pain 30 s after each
`injection, assessed on a I 00 mm visual analogue scale
`(0 mm = no pain, I 00 mm = severe pain); duration of injection
`pain, measured with a stop-watch given to each subject; local
`tolerance in tenns of redness, paleness and oedema, assessed
`by one of the staff on a yes or no scale; and, for some subjects,
`description of the injection pain in words (e.g. burning, itch(cid:173)
`ing).
`An erroneous exchange of injection 2 between subjects 7
`and 8 occurred during the study. Subject 7 did not receive
`formulation I and received formulation B twice, whereas
`subject 8 received formulation I twice and did not receive
`formulation B. All other subjects received all the formulations
`once. This imbalance was taken into account in the statistical
`handling of the data (see below).
`
`Blood glucose
`On treatment with hlGF-1 there is a minor risk of hypogly(cid:173)
`caemia. At the doses used in this study, this was considered to
`be very unlikely because the injections were given to non(cid:173)
`fasting subjects. In order to minimize the possibilities of
`hypoglycaemia, non-fasting glucose was measured before the
`first injection and after the last, and food was supplied during
`the study day.
`The concentration of glucose was measured with a Reflolux
`photometer (Boehringer Mannheim, Germany). Blood samples
`were taken from the finger tips from all subjects on the study
`day. The glucose levels were measured before the injections,
`and 0·5 and 3 h after the last injection.
`
`Statistics and data analysis
`The primary variable in the statistical analysis consisted of the
`visual analogue scale assessments of pain at each injection.
`Comparison of the eight formulations and two references was
`based on a graphical version of Tukey's T-procedure adapted
`to a 10 x 10 Graeco-Latin square design. Comparison between
`the effects of the formulation on the visual analogue scale
`responses could then be made in terms of so-called LSMEANS
`which, briefly, estimated what the means would have been if
`the design had been balanced. A slightly conservative variant
`of Tukey's T-procedure was used, namely to replace the
`
`FRESENIUS EXHIBIT 1047
`Page 2 of 4
`
`
`
`1014
`
`J. FRANSSON AND A. ESPANOER-JANSSON
`
`standard error of the difference between any two LSMEANS
`by the maximum 7-237 of all such standard errors. The 5%
`upper quartile of the studentized range distribution with
`parameter k = 9 (fonnulations) and 64 degrees of freedom is
`equal to 4.54 (Hochberg & Tamhane 1987). The ' uncertainty'
`intervals based on Tukey's T-procedure to be used in the
`graphical display, each have endpoints of the
`form
`LSMEAN± 4,54 x 7,237/8, i.e. LSMEAN ± 11 ·6. The for-
`mat comparisons of the formulation effects using Tukey's T-
`procedure were considered to be approximate in view of the
`visual analogue scale used. The residual plot indicated, as
`expected, that the variability in the visual analogue scale
`scores was relatively small at low levels, although the statis-
`tical model used actually assumes constant variability. This
`was compensated somewhat by the use of a slightly con-
`servative variant of Tukey's T procedure, with somewhat too
`large 'uncertainty' intervals.
`
`Results and Discussion
`
`As shown in Fig. I the different formulations caused different
`amounts of injection pain. The LSMEANS of the visual ana(cid:173)
`logue scale scores and the end-points of the corresponding
`'uncertainty' intervals are shown in the figure. Non-over(cid:173)
`lapping intervals indicate significant (at the 5% level) differ(cid:173)
`ences between the effects of the formulations. Even though
`there were quite large inter-individual differences, as shown by
`Table 3, pH 6, 50 mM phosphate formulations clearly caused
`more injection pain than pH 6, 10 mM phosphate formulations.
`It was shown that injection of hlGF-1, (pH 6, 50 mM phos(cid:173)
`phate), caused injection pain in at least 90% of the subjects in
`the study ( > 10 mm on visual analogue scale), whereas only
`300/4 of the subjects in the study marked more than 10 mm on
`the visual analogue scale on injection with a formulation of
`hlGF-1 of the same pH but with only 10 mm phosphate. Fur(cid:173)
`ther reduction in buffer concentration to 5 mm phosphate did
`not reduce pain further. Because formulations with or without
`hlGF-1 caused similar amounts of injection pain, it is con(cid:173)
`cluded that hIG F-1 itself did not cause injection pain.
`
`60
`
`50
`
`e
`.s 40
`
`(J
`
`Q)
`iii
`"'
`OJ 30
`~
`C)
`0 .;
`C: .,
`.; 20
`>
`
`~
`Cl)
`
`10
`
`0
`
`a
`
`<>
`
`0
`
`a
`
`<>
`
`0
`
`0
`
`a
`
`IJ
`
`0
`
`a
`
`0
`
`0
`
`0
`
`0
`
`a a
`
`a a
`
`<>
`<>
`0 ...... "'T'"_'T""" _ _ __,_~:..-~---'T"""--,.,....
`F
`G
`B
`A
`D
`C
`l +J H
`E
`Formulation
`
`FIG. I. Assessment of injection pain by visual analogue scale (0).
`Results fTom statistical evaluation of ' uncertainty intervals', upper (0)
`and lower limit (◊) for simultaneous comparisons of fonnulations at
`an approximate significance level of 5% based on Tukey 's T-method.
`The formulations are described in Table 2.
`
`An hlGF-1 preparation of pH 7 is not feasible because of the
`lower stability of hIGF-1. This pH caused less injection pain
`than pH 6, however. At pH 6, a lower buffer concentration
`resulted in less discomfort, possibly because a higher buffer
`concentration results in a slower change in solution pH at the
`injection site. The frequency and intensity of occurring red(cid:173)
`ness, paleness and oedema at the injection site also decreased
`on reducing the buffer concentration or increasing the pH.
`A non-physiological pH changes the chemical equilibrium
`between physiological buffers in subcutis and cutis tissue. This
`change causes transport of potassium ions which depolarize the
`nerve endings, causing pain (Guyton 1991). The magnitude
`
`Subject
`number
`
`I
`2
`3
`4
`5
`6
`7
`8
`9
`10
`
`Table 3. Summary of the assessments of pai n on a visual analogue scale (mm) for each
`formulation versus subject number. The fonnulations are described in Table 1.
`
`Formulation
`
`A
`
`B
`
`C
`
`D
`
`E
`
`F
`
`G
`
`H
`
`l + J
`
`29·0
`2·0
`1-0
`45·0
`56-0
`26·0
`18,0
`
`20-0
`47-0
`
`16-0
`11-0
`1-0
`28·0
`0-0
`7-0
`10•5
`
`0 0
`0·0
`
`64·0 -· 0-0
`
`0-0
`0-0
`0·0
`19-0
`15-0
`2·0
`5-0
`
`12·0
`12-0
`
`0-0
`7-0
`IO·0
`30·0
`7-0
`6-0
`2-0
`28-0
`00
`73-0
`
`163
`
`14•0
`14-0
`3-0
`33-0
`38-0
`23-0
`2·0
`17-0
`3-0
`4,0
`
`13-0
`57-0
`0-0
`33·0
`78-0
`21-0
`13·0
`75-0
`41 -0
`53-0
`
`15- 1
`
`38-4
`
`2·0
`21 -0
`5-0
`17·0
`60,0
`8-0
`0-0
`2·0
`7-0
`0·0
`
`12-2
`
`11 -0
`0-0
`7-0
`15-0
`0 -0
`9-0
`3-0
`2-0
`27•0
`9·0
`
`8-3
`
`0-0
`0-0
`1-0
`20·5
`8,5
`2·0
`O·O
`10,3
`29-5
`1-5
`
`7-2
`
`Mean
`
`30-9
`
`9.7
`
`6-5
`
`•subject number 8 did not receive formulation B.
`
`FRESENIUS EXHIBIT 1047
`Page 3 of 4
`
`
`
`LOCAL TOLERANCE OF SUBCUTANEOUS INJECTIONS
`
`!015
`
`and duration of the depolarization would be dependent on the
`pH and buffer concentration. In other words, the induced
`depolarization of the nerve endings is more quickly nonnali.zed
`at lower buffer concentrations.
`Pain is difficult to evaluate, partly because of its subjective
`character and the complex feelings that pain evokes. The
`subjective sensation of pain differs greatly within individuals,
`as is obvious from the large inter-individual differences in
`visual analogue scale scores in Table 3.
`The Graeco-Latin square design used in this study was
`chosen in order to obtain as much information as possible from
`a limited number of subjects. This design requires complete
`balance in order to achieve convincing results on completion
`of the study. In this study, however, imbalance occurred as
`stated at the end of the section on injection procedure. It was,
`therefore, necessary to modify the statistics somewhat on
`calculation of the results.
`There is a significant difference in pain, measured on a
`visual analogue scale, caused by injections at pH 6 and pH 7
`when the buffer concentration is high (50 mM). The pain
`experienced at pH 6 can, however, be reduced substantially by
`reducing the buffer concentration. It is the buffer capacity of
`the solution (i.e. resistance to pH changes) that determines how
`stable the pH value will be when the solution is injected into
`the tissue. The results from this study show that because the
`hlGF-1 formulation of pH 6 in IO mM phosphate results in less
`pain at the injection site it is a more suitable hIGF-1 for(cid:173)
`mulation of than that of pH 6 in 50 mM phosphate. As the
`formulations with hIGF-1 and the corresponding vehicles cause
`a similar degree of injection pain, it is shown that the pain is
`not caused by the hIGF-1 itself.
`
`Acknowledgements
`We thank Associate Professor Bertil Karlmark and Dr Marta
`Ryde for their expertise, Dr Ebba Florin-Robertsson for valu(cid:173)
`able comments during the preparation of the manuscript, and
`Dr Olivier Gilbaud for the statistical analysis.
`
`References
`
`Banos, J. E., Bosch, F., Canellas, M., Bassols, A., Ortega, F., Bigorra,
`J. (1989) Acceptability of visual analogue scales in the clinical
`setting: a comparison with verbal rating scales in postoperative pain.
`Meth. Find. Eitp. Clin. Phannacol. II: 123-127
`Barnett, T. A., Kapp, D. S. (1992) Reduction of pain and local
`complications when buffered lidocaine is used as a local anesthetic
`in conjunction with hyperthennia treatments: results of randomized
`trial. Int. J. Radial. Oncol. Biol. Phys. 23: 585-591
`Coley, R. M., Butler, C. D., Beck, B. I., Mullane, J.P. (1987) Effect of
`needle size on pain and hematoma formation with subcutaneous
`injection of heparin sodium. Clin. Phann. 6: 725-727
`{)Qenicke, A., Nebauer, A. E., Hoemecke, R., Mayer, M., Roizen, M.
`F., ( 1992) Osmolalities of propylene glycol-containing drug for(cid:173)
`mulations for parenteral use. Should propylene glycol be used as a
`solvent? Anesth. Analg. 75: 431-435
`Flynn, G. L. (1980) Parenteral fundamentals: buffers-pH control with
`pharmaceutical systems. J. Paren. Drug. Assoc. 34: 139-162
`Fransson, J., Florin-Robertsson, E., Moore, D. (1994) Solution contain(cid:173)
`ing IGF-1, WO patent 94/15584
`Frenken, L. A., van Lier, H.J., Jordans, J. G., Leunissen, K. M., van
`Leusen, R., Verstappen, V. M., Koene, R. A. (1993) Identification of
`the component part i.n an epoetin alfa preparation that causes pain
`after subcutaneous injection. Am. J. Kidney Dis. 22: 553-556
`Frenken, L. A., van Lier, H. J., Koene, R. A. (1994) Analysis of the
`efficacy of measures to reduce pain after subcutaneous administra(cid:173)
`tion of epoetin alfa. Nephrol. Dial. Transplant 9: 1295-1298
`Gazzaniga, G. M .• Angelim, G. , Pastorino, G. , Santuro, E., Lucchini,
`M., Dal-Pra, M. L. (1993) Enoxaprin in the prevention of deep
`venous thrombosis after major surgery. Int. Surg. 78: 271-275
`Guyton, A. C. (1991) Textbook of Medical Physiology. W. B.
`Saunders, London
`Hochberg, Y., Tamhane, A. C. (1987) Multiple Comparison Proce(cid:173)
`dures. Wiley, New York, pp 96-97
`lpp, M. M., Gold, R., Goldbach, M., Maresky, D. C., Saunders, N.,
`Greenberg, S .• Davy, T. (1990) Adverse reactions to diphtheria,
`tetanus, pertussis-polio vaccination at 18 months of age: effect of
`injection site and needle length. Pediatrics 85: 134-135
`Langley, G. B., Sheppard, H. (1985) The visual analogue scale: its use
`in pain measurement. Rheumatol. Int. 5: 145-148
`Peng, K. C. ( 1967) The Design and Analysis of Scientific Experiments.
`Addison-Wesley, Reading, p. 104
`Ross, S., Soltes, D. (1995) Heparin and haematoma: does ice make a
`difference? J. Adv. Nurs. 21: 434-439
`Zindel, L. R. (1989) Debut of a somatostatin analogue: octreotide in
`review. Conn. Med. 53: 741- 744
`
`FRESENIUS EXHIBIT 1047
`Page 4 of 4
`
`