Drug Safety 5 (Suppl. 1): 58-64, 1990 0114-5916/90/0001-0058/$3.50/0 © ADiS Press Limited All rights reserved. DSSUP1791
Safety Evaluation of Biotechnology Products G. Zbinden Institute of Toxicology, Swiss Federal Institute of Technology and University of Zurich, Schwerzenbach, Switzerland
Summary
Preclinical safety studies with biotechnology products are not only performed for regulatory purposes, but should first and foremost provide information about the potential toxic effects in patients. The initial toxicological experience, using standard testing procedures developed for drugs of small molecular weight, often gave disappointing results, and the development of antibodies against the heterologous products cast doubt upon the validity of the testing approach. In order to assess the safety of new biotechnology products, compounds must be looked at on a case by case basis. Exaggerated pharmacodynamic effects are often responsible for the major toxicological problems. For some compounds, 'intrinsic toxicity', i.e adverse effects due io the molecule per se, may playa role. With others, 'biological toxicity', i.e. the activation of physiological processes such as antigen-antibody interaction, release of mediators and cytokines, or initiation of the arachidonic acid-prostaglandin cascade, may be the cause of the observed adverse effects. Examples are given that show the importance of a good understanding of the biological mechanisms of action of toxicity observed in animals and in patients.
The original predictions of adverse effects caused by human proteins and polypeptides produced by modern biotechnology techniques were, to a large extent, incorrect: the impurities and contaminants thought by experts to present major safety problems were rapidly brought under control, thanks to improvements in manufacturing procedures. On the other hand, the hopes that human proteins and polypeptides would induce few unexpected adverse reactions remained unfulfilled. This was particularly true for the cytokines, whose therapeutic use is associated with a variety of toxic effects, ranging from trivial to dangerous (Fent & Zbinden 1987; Remick & Kunkel 1989). On the regulatory level, it was first thought that safety tests performed with conventional drugs
should also be useful for the new biotechnology products (Japanese Ministry of Health and Welfare 1984). However, industrial toxicologists soon learned that this approach, attributable to inexperience, gave disappointing results (Teelmann et al. 1986). This led to regulatory demands for more specific toxicological models, e.g. in the fields of cardiovascular, neurobehavioural and immunological toxicology (French Directorate of Pharmacy and Drugs 1984), but systematic efforts to develop and validate such procedures have not yet been made. From the chemical nature of the biotechnology products, it is evident that antibodies leading to allergic pathology or to rapid neutralisation and clearance of the test substances are induced in ani-
Toxicology of Biotechnology Products
mals. In the opinion of some toxicologists, such effects might invalidate the experimental results and cast doubt upon the justification of regulatory agencies to demand animal safety studies.
1. Current Safety Testing Approaches Human therapy with biotechnology products is in its infancy. We are still in the process of gathering experimental and clinical observations to help us develop valid safety testing concepts. However, it has now become clear that we are dealing with essentially 3 areas of concern (Zbinden 1987). Firstly, by far the most important safety problems for many compounds are those resulting from the pharmacodynamic properties. Fortunately, many of these are amenable to experimental scrutiny or rational extrapolation. The second and third areas of concern, 'intrinsic toxicity' and 'biological toxicity', are less well defined entities, and are often difficult to separate from each other. 'Intrinsic toxicity' comprises those adverse effects that are due to the molecules per se and are not a direct consequence of the pharmacodynamic actions of the substances. Toxic responses resulting from the activation of a physiological mechanism are subsumed under the designation of 'biological toxicity'. The most frequently occurring of these adverse effects are those due to an antigen-antibody reaction, but other biological pathways, e.g. those related to release of cytokines, other mediators and acute phase proteins, and activation of the complement system and the arachidonic acid-prostaglandin cascade (Remick & Kunkel 1989), must also be considered.
2. Toxicity Related to Pharmacodynamic Effects As with drugs of small molecular weight, biotechnology products may induce exaggerated pharmacodynamic responses when given at high doses. Thus, it is easy to predict that recombinant human (rh) insulin can cause hypogiycaemia, that rh tissue plasminogen activator may be associated with fibrinogenolysis and bleeding (Billings et al. 1989),
59
Table I. Adverse reactions caused by IL·2 Man Fever, chills, malaise Nausea, vomiting Diarrhoea Vascular leak syndrome Urticaria
+ +
Anaemia Thrombocytopenia Leucocytosis Eosinophilia Hypoalbuminaemia
+ + + + +
Hypotension Hepatotoxicity Nephrotoxicity Mental disturbance
+ + + +
+ +
Mouse
Rat
+
+ (+)
+ + + +
+ (+) + + +
+ (+)
+ +
Key: + = frequent, marked; (+) = slight.
that rh erythropoietin may induce an increase in red cell mass, a mobilisation of iron stores and an enlargement of highly vascularised organs (Anderson 1989), and that rh growth hormone may be diabetogenic, because of its inherent antagonistic effect against insulin. With other substances, the distinction between pharmacodynamic overshoot, lesions attributable to the substance itself ('intrinsic toxicity') and adverse reactions occurring as a consequence of a secondary biological response is difficult to make. This will be illustrated with toxicity data obtained with interleukin-2 (lL-2) [see section 3].
3. 'Intrinsic Toxicity' From their chemical structure, it is expected that biotechnology products cause few adverse reactions that are not receptor-mediated but due to direct interactions of the molecules with membranes and other cellular components. An exception to this rule appears to be the cytokines whose therapeutic use is associated with many, often dose-related, adverse responses affecting a variety of organ systems (Fent & Zbinden 1987; Remick & Kunkel 1989). As an example, toxic effects associated with the therapeutic use of IL-2 are summarised in table I
60
Drug Safety 5 (Suppi. 1) 1990
(Anderson & Hayes 1989; Anderson et al. 1988; Matory et al. 1985; Rosenberg et al. 1985; Sondel et al. 1988). Table I also shows that most adverse effects of 1L-2 seen in human subjects can be induced in laboratory animals. From this observation it is concluded that the IL-2 molecule acts directly on the target cells without the need of species-specific, receptor-mediated mechanisms. However, recent experiments have shown that some of the toxic responses in mice, particularly vascular leak syndrome and hepatic necrosis, are caused by an indirect mechanism, i.e. stimulation of, and infiltration with, an endogenous subset of IL-2 activated lymphocytes (Anderson et al. 1988).
4.
o= o
IFN pretr 2
= HSA pretr 1
• = HSA pretr 2
0
0
10
20
30
40
50
60
Time (minutes)
Fig. 1. Percentage of platelets circulating as irreversible aggregates in guinea-pigs receiving an intravenous infusion of 3 x I06U recombinant human interferon a-2a (IFN) or 5mg human serum albumin (HSA) over a 60-minute period (n = 2). Other animals (n = 2) were pretreated with either 2 intramuscular injections of 1.5 x I06U IFN (IFN pretreated) or 2.5mg HSA (HSA pretreated). No irreversible aggregates were present in animals receiving IFN I and 2, and HSA I and 2. Therefore, they are not shown in the graph.
Drug Safety 5 (Suppl. 1) 1990
62
120~~~
+= IFN 1
90
0=
80
• = IFN pretr 1
70
o=
110 100
;R ~
co
8 60
Q) a; iii
a::
IFN 2
IFN pretr 2
.. = HSA 1
50 "" = HSA 2
40
30
o = HSA pretr 1
20+-------.-------.-------.-------,-------,-------. 60 10 o 20 30 40 50
• = HSA pretr 2
Time (minutes)
Fig. 2. Platelet counts as a percentage of individual pretreatment values in guinea-pigs receiving an intravenous infusion of IFN and HSA, with and without pretreatment with IFN and HSA, respectively. Measurements were from the same animals as in figure 1.
can induce intravascular platelet aggregation and microembolism, but only in animals sensitised with the drug and having high titres of IFN antibodies. Thus, it is probable that the effect is due to antigenantibody reaction and unlikely to occur in nonsensitised humans.
5. Scientific Challenges The example ofIL-2 (section 3) shows that routine toxicological studies were adequate for detecting many of the adverse effects later found in humans. On the other hand, the standard approach failed dismally with the interferons, probably because of lack of appropriate receptors in the target organs of the animals. It is predictable that similar failures could occur with other classes of biotechnology products. For this reason, it has been proposed that primates and even subhuman primates should be used for toxicological investigations of biotechnology products (Schellekens et al. 1984). This approach is as yet unproved, has obvious economic disadvantages and raises serious ethical questions (Schellekens & v.d. Meide 1983; Teelmann et al. 1986). As an alternative, it is often suggested that toxicological studies should be con-
ducted with homologous biotechnology products originating from the species in which the experiments are performed. This might indeed be advantageous in certain circumstances, e.g. when an animal safety study appears to be compromised by high titres of neutralising antibodies, when negative results in toxicity studies may be due to lack of specific receptors, and when a toxic response that could be due to an antigen-antibody interaction directed against the human material is observed. However, the efforts involved in such studies are enormous and can only be justified in exceptional cases. The same is true for the recent suggestion to use transgenic animals carrying the gene of the human biotechnology product as toxicological models. Another possible approach is the development of tests that measure well-defined, specific properties of the biotechnology products, preferably those that are pertinent to the toxicological problem. In this area, in vitro models may often be useful. For example, neurobehavioural disturbances such as mental and motor slowing, loss of memory, disorientation and distinct electroencephalographic changes have often been observed in patients treated with IFN (Rohatiner et al. 1983). These abnormalities cannot be duplicated in ani-
Toxicology of Biotechnology Products
mals, but distinct and dose-dependent effects on the electrophysiological process of the brain can be demonstrated in vitro. For example, in hippocampal slice cultures obtained from newborn mice, r murine IFN--y enhanced excitability of CA3 pyramidal cells and increased the number of spontaneous action potentials. This was followed by a disappearance of synaptic inhibition. The effect lasted for several hours and was never fully reversible (M. Miiller, personal communication). Similar observations, i.e. enhancement of spontaneous electrical activity and a marked shortening of the latency of evoked electrical activity, were made with explants from cerebral and cerebellar cortex of newborn rats and kittens. Only human but not rat IFN was active on the cat neurons (Calvet & Gresser 1979). Further studies must show whether in vitro nerve cell cultures can be used to detect neurotoxic effects of other cytokines, e.g. those seen with IL-2 in man (Sosman et al. 1988). Finally, the importance of basic investigations of the biomechanisms of biotechnology products must be underlined. From this knowledge, potential areas of toxicological concern can be identified, and specific in vivo and clinical studies can be suggested to either confirm or reject the hypothesis. As an example, the relationship between IFN--y and expression of major histocompatibility complex (MHC) antigens is mentioned. It was found that this cytokine enhances the expression of class I and class II MHC antigens in many cell types in vitro and in many tissues in vivo (Skokiewicz et al. 1985). Of particular interest is the appearance of MHC class II molecules on the surface of 'inappropriate' cells (these antigens are normally present only in B-Iymphocytes, some macrophages and Langerhans cells). T helper cells require these antigens as restriction elements for peptide recognition, a key step in the activation of T helper cells. The presence of class II MHC antigens may then form the basis for the development of autoimmune disease (Bottazzo et al. 1983; Todd et al. 1988). The observations that treatment of newborn mice with mouse IFN caused an autoimmune-type nephropathy (Gresser 1983) and that the development of naturally occurring autoimmune disease of New
63
Zealand Black (NZB) mice was accelerated by treatment with homologous IFN (Sergiescu et al. 1979) are in agreement with the hypothesis.
6. Conclusions Experimental and clinical experience with a limited number of drugs produced by modern biotechnology techniques has shown that standard safety studies with conventional laboratory animals can, but often do not, predict adverse reactions occurring in humans. For this reason, toxicity testing of these compounds requires a flexible approach, taking into consideration all available chemical, pharmacological and biochemical information. Firstly, the physiological functions of the products must be characterised, and the pharmacodynamic spectrum must be analysed. To study potential toxicity after administration of exaggerated doses, conventional animal test procedures should first be used. If no effects are seen, specially developed animal models may be tried. In addition, studies designed to discover species differences (e.g. receptor binding experiments) may be added, and, in exceptional cases, model experiments with homologous biotechnology products may be performed. When toxic effects are discovered in animal studies, it is necessary to investigate their mechanism. Are the changes related to a pharmacodynamic effect, are physiological mechanisms set in motion (e.g. release of mediators or acute phase proteins, activation of the arachidonic acid-prostaglandin cascade etc.), or is the injury caused by a direct effect of the molecule on certain cellular or subcellular targets? For the investigation of such questions, in vitro test systems that can assess specific toxicological mechanisms have already proved to be useful and are likely to become even more important in the future. The development of antibodies by the animals against the human materials needs to be carefully watched, and the potential allergic nature of all lesions developing in toxicity tests must be kept in mind. Rigorous application of standard toxicological protocols is certain to yield much useless or equiv-
64
Drug Safety 5 (Suppl. 1) 1990
ocal information. But a close interplay between flexible toxicological investigations tailored to the special characteristics of each drug and early human trials, with careful monitoring of patients by clinical pharmacologists, is the best guarantee for a rapid acquisition of relevant safety data with minimal risks for the subjects.
Acknowledgement Recombinant human interferon-a was a gift from F. Hoffmann-La Roche, AG, Basel, Switzerland, and r murine IFN-,.. was obtained from Dr G.R. Adolph, Ernst B6hringer Institut fOr Arzneimittelforschung, Vienna, Austria. The technical assistance of Mrs L. Grimm, Mrs H. Spichiger and Mrs V. Streit is gratefully acknowledged.
References Anderson Le. Safety evaluation of recombinant human erythropoietin. American Association of Science Meeting San Francisco, CA, January 1989. Abstract, p. 58, 1989 Anderson TD, Hayes TJ. Toxicity of human recombinant interleukin-2 in rats. Pathologic changes are characterized by marked lymphocytic and eosinophilic proliferation and multisystem involvement. Laboratory Investigation 60: 331-346, 1989 Anderson TD, Hayes TJ, Gately MK, Bontempo JM, Stern LL, et al. Toxicity of human recombinant interleukin-2 in the mouse is mediated by interleukin-activated lymphocytes. Separation of efficacy and toxicity by selective lymphocyte subset depletion. Laboratory Investigation 59: 598-612, 1988 Billings RE, Finkle BS, Hotchkiss A. Safety assessment of activase (recombinant human tissue plasminogen activator). American Association of Science Meeting San Francisco, CA, January 1989. Abstract, p. 58, 1989 Bottazzo GF, Pujol-Borrell R, Hanafusa T. Role of aberrant HLADR expression and antigen presentation in induction of endocrine autoimmunity. Lancet 2: 1115-1119, 1983 Cal vet MC, Gresser I. Interferon enhances the excitability of cultured neurones. Nature 278: 558-560, 1979 Fent K, Zbinden G. Toxicity of interferon and interleukin. Trends in Pharmacological Sciences 8: 100-105, 1987 French Directorate of Pharmacy and Drugs. Recommendation concernant Ie protocole toxicologique des interferons pour l'obtention d'une autorisation de mise sur Ie marche. March 12-16, 1984 Gresser I. Interferon-induced disease. In De Maeyer E, Schellekens H (Eds) The biology of the interferon system, p. 363, Elsevier Scientific Publishers, Amsterdam, 1983 Japanese Ministry of Health and Welfare. Notification on application data for rDNA drugs. Notification No 243 of the Pharmaceutical Affairs Bureau, March 30, 1984 Maca RD, Fry GL, Hoak Je. New method for detection and quantitation of circulating platelet aggregates. Microvascular Research 4: 453-457, 1972 Matory YL, Chang AE, Lipford IJI EH, Braziel R, Hyatt CL, et al. Toxicity of recombinant human interleukin-2 in rats following intravenous infusion. Journal of Biological Response Modifiers 4: 377-390, 1985 Mirro Jr J, Kalwinsky D, Whisnant J, Weck P, Chesney C, et al.
Coagulopathy induced by continuous infusion of high doses of human Iymphoblastoid interferon. Cancer Treatment Report 69: 315-317,1985 Panem S, Yilcek J. Antibodies to interferon in man. In De Maeyer E & Schellekens H (Eds) The biology of the interferon system, pp. 369-378, Elsevier Science Publishers BY, Amsterdam, 1983 Remick DG, Kunkel SL. Toxic effects of cytokines in vivo. laboratory Investigation 60: 317-319, 1989 Rohatiner AZS, Prior PF, Burton AC, Smith AT, Balkwill FR, et al. Central nervous system toxicity of interferon. British Journal of Cancer 47: 419-422, 1983 Rosenberg SA, Lotze MT, Muul LM, Leitman S, Chang AE, et al. Observations on the systemic administration of autologous Iymphokine-activated cells and recombinant interleukin-2 to patients with metastatic cancer. New England Journal of Medicine 313: 1485-1492, 1985 Schellekens H, v.d. Meide PH. Animal studies with interferon. In De Maeyer E & Schellekens H (Eds) The biology of the interferon system, pp. 409-418, Elsevier Science Publishers BY Amsterdam, 1983 Schellekens H, de Reus A, v.d. Meide PH. The chimpanzee as a model to test the side effects of human interferons. Journal of Medical Primatology 13: 235-245, 1984 Scott GM. Interferon: pharmacokinetics and toxicity. Philosophical Transactions of the Royal Society London B 299: 91-107, 1982 Sergiescu D, Cerutti I, Efthymiou E, Kahan A, Chany e. Adverse effects of interferon treatment on the life span of NZB mice. Biomedicine 31: 48-51, 1979 Skoskievicz MJ, Colvin RB, Schneeberger EE, Russell PS. Widespread and selective induction of major histocompatibility complex-determined antigens in vivo by gamma interferon. Journal of Experimental Medicine 162: 1645-1664, 1985 Sondel PM, Kohler PC, Hank JA, Moore KH, Rosenthal NS, et al. Clinical and immunological effects of recombinant interleukin-2 given by repetitive weekly cycles to patients with cancer. Cancer Research 48: 2561-2567, 1988 Sosman JA, Kohler PC, Hank JA, Moore KA, Bechhofer R, et al. Repetitive weekly cycles of interleukin-2. II. Clinical and immunological effects of dose, schedule and addition of indomethacin. Journal ofthe National Cancer Institute 80: 14511461, 1988 Teelmann K, Hohbach C, Lehmann H. Preclinical safety testing of species-specific proteins produced with recombinant DNAtechniques. Archives of Toxicology 59: 195-200, 1986 Todd JA, Acha-Orbea H, Bell JI, Chao N, Fronek Z, et al. A molecular basis for MHC class II-associated autoimmunity. Science 240: 1003-1009, 1988 Yelcovsky HG, Federlin KF. Insulin-specific IgG and IgE antibody response in type I diabetic subjects exclusively treated with human insulin (recombinant DNA). Diabetes Care 5 (Suppl. 2): 126-128, 1982 Zbinden G. Biotechnology products intended for human use, toxicological targets and research strategies. In Graham CE (Ed.) Preclinical safety of biotechnology pr.oducts intended for human use, pp. 143-159, Alan R. Liss, Inc., New York, 1987 Zbinden G. Effects of recombinant human alpha interferon in a rodent cardiotoxicity model. Toxicology Letters, in press Zbinden G, Grimm L. Thrombogenic effects of xenobiotics. Archives of Toxicology (Suppl. 8): 131-141, 1985
Author's address: Dr G. Zbinden, Institute of Toxicology, Swiss Federal Institute of Technology and University of Zurich, Schorenstrasse 16, CH-8603 Schwerzenbach, Switzerland.
Safety Evaluation of Biotechnology Products G. Zbinden Institute of Toxicology, Swiss Federal Institute of Technology and University of Zurich, Schwerzenbach, Switzerland
Summary
Preclinical safety studies with biotechnology products are not only performed for regulatory purposes, but should first and foremost provide information about the potential toxic effects in patients. The initial toxicological experience, using standard testing procedures developed for drugs of small molecular weight, often gave disappointing results, and the development of antibodies against the heterologous products cast doubt upon the validity of the testing approach. In order to assess the safety of new biotechnology products, compounds must be looked at on a case by case basis. Exaggerated pharmacodynamic effects are often responsible for the major toxicological problems. For some compounds, 'intrinsic toxicity', i.e adverse effects due io the molecule per se, may playa role. With others, 'biological toxicity', i.e. the activation of physiological processes such as antigen-antibody interaction, release of mediators and cytokines, or initiation of the arachidonic acid-prostaglandin cascade, may be the cause of the observed adverse effects. Examples are given that show the importance of a good understanding of the biological mechanisms of action of toxicity observed in animals and in patients.
The original predictions of adverse effects caused by human proteins and polypeptides produced by modern biotechnology techniques were, to a large extent, incorrect: the impurities and contaminants thought by experts to present major safety problems were rapidly brought under control, thanks to improvements in manufacturing procedures. On the other hand, the hopes that human proteins and polypeptides would induce few unexpected adverse reactions remained unfulfilled. This was particularly true for the cytokines, whose therapeutic use is associated with a variety of toxic effects, ranging from trivial to dangerous (Fent & Zbinden 1987; Remick & Kunkel 1989). On the regulatory level, it was first thought that safety tests performed with conventional drugs
should also be useful for the new biotechnology products (Japanese Ministry of Health and Welfare 1984). However, industrial toxicologists soon learned that this approach, attributable to inexperience, gave disappointing results (Teelmann et al. 1986). This led to regulatory demands for more specific toxicological models, e.g. in the fields of cardiovascular, neurobehavioural and immunological toxicology (French Directorate of Pharmacy and Drugs 1984), but systematic efforts to develop and validate such procedures have not yet been made. From the chemical nature of the biotechnology products, it is evident that antibodies leading to allergic pathology or to rapid neutralisation and clearance of the test substances are induced in ani-
Toxicology of Biotechnology Products
mals. In the opinion of some toxicologists, such effects might invalidate the experimental results and cast doubt upon the justification of regulatory agencies to demand animal safety studies.
1. Current Safety Testing Approaches Human therapy with biotechnology products is in its infancy. We are still in the process of gathering experimental and clinical observations to help us develop valid safety testing concepts. However, it has now become clear that we are dealing with essentially 3 areas of concern (Zbinden 1987). Firstly, by far the most important safety problems for many compounds are those resulting from the pharmacodynamic properties. Fortunately, many of these are amenable to experimental scrutiny or rational extrapolation. The second and third areas of concern, 'intrinsic toxicity' and 'biological toxicity', are less well defined entities, and are often difficult to separate from each other. 'Intrinsic toxicity' comprises those adverse effects that are due to the molecules per se and are not a direct consequence of the pharmacodynamic actions of the substances. Toxic responses resulting from the activation of a physiological mechanism are subsumed under the designation of 'biological toxicity'. The most frequently occurring of these adverse effects are those due to an antigen-antibody reaction, but other biological pathways, e.g. those related to release of cytokines, other mediators and acute phase proteins, and activation of the complement system and the arachidonic acid-prostaglandin cascade (Remick & Kunkel 1989), must also be considered.
2. Toxicity Related to Pharmacodynamic Effects As with drugs of small molecular weight, biotechnology products may induce exaggerated pharmacodynamic responses when given at high doses. Thus, it is easy to predict that recombinant human (rh) insulin can cause hypogiycaemia, that rh tissue plasminogen activator may be associated with fibrinogenolysis and bleeding (Billings et al. 1989),
59
Table I. Adverse reactions caused by IL·2 Man Fever, chills, malaise Nausea, vomiting Diarrhoea Vascular leak syndrome Urticaria
+ +
Anaemia Thrombocytopenia Leucocytosis Eosinophilia Hypoalbuminaemia
+ + + + +
Hypotension Hepatotoxicity Nephrotoxicity Mental disturbance
+ + + +
+ +
Mouse
Rat
+
+ (+)
+ + + +
+ (+) + + +
+ (+)
+ +
Key: + = frequent, marked; (+) = slight.
that rh erythropoietin may induce an increase in red cell mass, a mobilisation of iron stores and an enlargement of highly vascularised organs (Anderson 1989), and that rh growth hormone may be diabetogenic, because of its inherent antagonistic effect against insulin. With other substances, the distinction between pharmacodynamic overshoot, lesions attributable to the substance itself ('intrinsic toxicity') and adverse reactions occurring as a consequence of a secondary biological response is difficult to make. This will be illustrated with toxicity data obtained with interleukin-2 (lL-2) [see section 3].
3. 'Intrinsic Toxicity' From their chemical structure, it is expected that biotechnology products cause few adverse reactions that are not receptor-mediated but due to direct interactions of the molecules with membranes and other cellular components. An exception to this rule appears to be the cytokines whose therapeutic use is associated with many, often dose-related, adverse responses affecting a variety of organ systems (Fent & Zbinden 1987; Remick & Kunkel 1989). As an example, toxic effects associated with the therapeutic use of IL-2 are summarised in table I
60
Drug Safety 5 (Suppi. 1) 1990
(Anderson & Hayes 1989; Anderson et al. 1988; Matory et al. 1985; Rosenberg et al. 1985; Sondel et al. 1988). Table I also shows that most adverse effects of 1L-2 seen in human subjects can be induced in laboratory animals. From this observation it is concluded that the IL-2 molecule acts directly on the target cells without the need of species-specific, receptor-mediated mechanisms. However, recent experiments have shown that some of the toxic responses in mice, particularly vascular leak syndrome and hepatic necrosis, are caused by an indirect mechanism, i.e. stimulation of, and infiltration with, an endogenous subset of IL-2 activated lymphocytes (Anderson et al. 1988).
4.
o= o
IFN pretr 2
= HSA pretr 1
• = HSA pretr 2
0
0
10
20
30
40
50
60
Time (minutes)
Fig. 1. Percentage of platelets circulating as irreversible aggregates in guinea-pigs receiving an intravenous infusion of 3 x I06U recombinant human interferon a-2a (IFN) or 5mg human serum albumin (HSA) over a 60-minute period (n = 2). Other animals (n = 2) were pretreated with either 2 intramuscular injections of 1.5 x I06U IFN (IFN pretreated) or 2.5mg HSA (HSA pretreated). No irreversible aggregates were present in animals receiving IFN I and 2, and HSA I and 2. Therefore, they are not shown in the graph.
Drug Safety 5 (Suppl. 1) 1990
62
120~~~
+= IFN 1
90
0=
80
• = IFN pretr 1
70
o=
110 100
;R ~
co
8 60
Q) a; iii
a::
IFN 2
IFN pretr 2
.. = HSA 1
50 "" = HSA 2
40
30
o = HSA pretr 1
20+-------.-------.-------.-------,-------,-------. 60 10 o 20 30 40 50
• = HSA pretr 2
Time (minutes)
Fig. 2. Platelet counts as a percentage of individual pretreatment values in guinea-pigs receiving an intravenous infusion of IFN and HSA, with and without pretreatment with IFN and HSA, respectively. Measurements were from the same animals as in figure 1.
can induce intravascular platelet aggregation and microembolism, but only in animals sensitised with the drug and having high titres of IFN antibodies. Thus, it is probable that the effect is due to antigenantibody reaction and unlikely to occur in nonsensitised humans.
5. Scientific Challenges The example ofIL-2 (section 3) shows that routine toxicological studies were adequate for detecting many of the adverse effects later found in humans. On the other hand, the standard approach failed dismally with the interferons, probably because of lack of appropriate receptors in the target organs of the animals. It is predictable that similar failures could occur with other classes of biotechnology products. For this reason, it has been proposed that primates and even subhuman primates should be used for toxicological investigations of biotechnology products (Schellekens et al. 1984). This approach is as yet unproved, has obvious economic disadvantages and raises serious ethical questions (Schellekens & v.d. Meide 1983; Teelmann et al. 1986). As an alternative, it is often suggested that toxicological studies should be con-
ducted with homologous biotechnology products originating from the species in which the experiments are performed. This might indeed be advantageous in certain circumstances, e.g. when an animal safety study appears to be compromised by high titres of neutralising antibodies, when negative results in toxicity studies may be due to lack of specific receptors, and when a toxic response that could be due to an antigen-antibody interaction directed against the human material is observed. However, the efforts involved in such studies are enormous and can only be justified in exceptional cases. The same is true for the recent suggestion to use transgenic animals carrying the gene of the human biotechnology product as toxicological models. Another possible approach is the development of tests that measure well-defined, specific properties of the biotechnology products, preferably those that are pertinent to the toxicological problem. In this area, in vitro models may often be useful. For example, neurobehavioural disturbances such as mental and motor slowing, loss of memory, disorientation and distinct electroencephalographic changes have often been observed in patients treated with IFN (Rohatiner et al. 1983). These abnormalities cannot be duplicated in ani-
Toxicology of Biotechnology Products
mals, but distinct and dose-dependent effects on the electrophysiological process of the brain can be demonstrated in vitro. For example, in hippocampal slice cultures obtained from newborn mice, r murine IFN--y enhanced excitability of CA3 pyramidal cells and increased the number of spontaneous action potentials. This was followed by a disappearance of synaptic inhibition. The effect lasted for several hours and was never fully reversible (M. Miiller, personal communication). Similar observations, i.e. enhancement of spontaneous electrical activity and a marked shortening of the latency of evoked electrical activity, were made with explants from cerebral and cerebellar cortex of newborn rats and kittens. Only human but not rat IFN was active on the cat neurons (Calvet & Gresser 1979). Further studies must show whether in vitro nerve cell cultures can be used to detect neurotoxic effects of other cytokines, e.g. those seen with IL-2 in man (Sosman et al. 1988). Finally, the importance of basic investigations of the biomechanisms of biotechnology products must be underlined. From this knowledge, potential areas of toxicological concern can be identified, and specific in vivo and clinical studies can be suggested to either confirm or reject the hypothesis. As an example, the relationship between IFN--y and expression of major histocompatibility complex (MHC) antigens is mentioned. It was found that this cytokine enhances the expression of class I and class II MHC antigens in many cell types in vitro and in many tissues in vivo (Skokiewicz et al. 1985). Of particular interest is the appearance of MHC class II molecules on the surface of 'inappropriate' cells (these antigens are normally present only in B-Iymphocytes, some macrophages and Langerhans cells). T helper cells require these antigens as restriction elements for peptide recognition, a key step in the activation of T helper cells. The presence of class II MHC antigens may then form the basis for the development of autoimmune disease (Bottazzo et al. 1983; Todd et al. 1988). The observations that treatment of newborn mice with mouse IFN caused an autoimmune-type nephropathy (Gresser 1983) and that the development of naturally occurring autoimmune disease of New
63
Zealand Black (NZB) mice was accelerated by treatment with homologous IFN (Sergiescu et al. 1979) are in agreement with the hypothesis.
6. Conclusions Experimental and clinical experience with a limited number of drugs produced by modern biotechnology techniques has shown that standard safety studies with conventional laboratory animals can, but often do not, predict adverse reactions occurring in humans. For this reason, toxicity testing of these compounds requires a flexible approach, taking into consideration all available chemical, pharmacological and biochemical information. Firstly, the physiological functions of the products must be characterised, and the pharmacodynamic spectrum must be analysed. To study potential toxicity after administration of exaggerated doses, conventional animal test procedures should first be used. If no effects are seen, specially developed animal models may be tried. In addition, studies designed to discover species differences (e.g. receptor binding experiments) may be added, and, in exceptional cases, model experiments with homologous biotechnology products may be performed. When toxic effects are discovered in animal studies, it is necessary to investigate their mechanism. Are the changes related to a pharmacodynamic effect, are physiological mechanisms set in motion (e.g. release of mediators or acute phase proteins, activation of the arachidonic acid-prostaglandin cascade etc.), or is the injury caused by a direct effect of the molecule on certain cellular or subcellular targets? For the investigation of such questions, in vitro test systems that can assess specific toxicological mechanisms have already proved to be useful and are likely to become even more important in the future. The development of antibodies by the animals against the human materials needs to be carefully watched, and the potential allergic nature of all lesions developing in toxicity tests must be kept in mind. Rigorous application of standard toxicological protocols is certain to yield much useless or equiv-
64
Drug Safety 5 (Suppl. 1) 1990
ocal information. But a close interplay between flexible toxicological investigations tailored to the special characteristics of each drug and early human trials, with careful monitoring of patients by clinical pharmacologists, is the best guarantee for a rapid acquisition of relevant safety data with minimal risks for the subjects.
Acknowledgement Recombinant human interferon-a was a gift from F. Hoffmann-La Roche, AG, Basel, Switzerland, and r murine IFN-,.. was obtained from Dr G.R. Adolph, Ernst B6hringer Institut fOr Arzneimittelforschung, Vienna, Austria. The technical assistance of Mrs L. Grimm, Mrs H. Spichiger and Mrs V. Streit is gratefully acknowledged.
References Anderson Le. Safety evaluation of recombinant human erythropoietin. American Association of Science Meeting San Francisco, CA, January 1989. Abstract, p. 58, 1989 Anderson TD, Hayes TJ. Toxicity of human recombinant interleukin-2 in rats. Pathologic changes are characterized by marked lymphocytic and eosinophilic proliferation and multisystem involvement. Laboratory Investigation 60: 331-346, 1989 Anderson TD, Hayes TJ, Gately MK, Bontempo JM, Stern LL, et al. Toxicity of human recombinant interleukin-2 in the mouse is mediated by interleukin-activated lymphocytes. Separation of efficacy and toxicity by selective lymphocyte subset depletion. Laboratory Investigation 59: 598-612, 1988 Billings RE, Finkle BS, Hotchkiss A. Safety assessment of activase (recombinant human tissue plasminogen activator). American Association of Science Meeting San Francisco, CA, January 1989. Abstract, p. 58, 1989 Bottazzo GF, Pujol-Borrell R, Hanafusa T. Role of aberrant HLADR expression and antigen presentation in induction of endocrine autoimmunity. Lancet 2: 1115-1119, 1983 Cal vet MC, Gresser I. Interferon enhances the excitability of cultured neurones. Nature 278: 558-560, 1979 Fent K, Zbinden G. Toxicity of interferon and interleukin. Trends in Pharmacological Sciences 8: 100-105, 1987 French Directorate of Pharmacy and Drugs. Recommendation concernant Ie protocole toxicologique des interferons pour l'obtention d'une autorisation de mise sur Ie marche. March 12-16, 1984 Gresser I. Interferon-induced disease. In De Maeyer E, Schellekens H (Eds) The biology of the interferon system, p. 363, Elsevier Scientific Publishers, Amsterdam, 1983 Japanese Ministry of Health and Welfare. Notification on application data for rDNA drugs. Notification No 243 of the Pharmaceutical Affairs Bureau, March 30, 1984 Maca RD, Fry GL, Hoak Je. New method for detection and quantitation of circulating platelet aggregates. Microvascular Research 4: 453-457, 1972 Matory YL, Chang AE, Lipford IJI EH, Braziel R, Hyatt CL, et al. Toxicity of recombinant human interleukin-2 in rats following intravenous infusion. Journal of Biological Response Modifiers 4: 377-390, 1985 Mirro Jr J, Kalwinsky D, Whisnant J, Weck P, Chesney C, et al.
Coagulopathy induced by continuous infusion of high doses of human Iymphoblastoid interferon. Cancer Treatment Report 69: 315-317,1985 Panem S, Yilcek J. Antibodies to interferon in man. In De Maeyer E & Schellekens H (Eds) The biology of the interferon system, pp. 369-378, Elsevier Science Publishers BY, Amsterdam, 1983 Remick DG, Kunkel SL. Toxic effects of cytokines in vivo. laboratory Investigation 60: 317-319, 1989 Rohatiner AZS, Prior PF, Burton AC, Smith AT, Balkwill FR, et al. Central nervous system toxicity of interferon. British Journal of Cancer 47: 419-422, 1983 Rosenberg SA, Lotze MT, Muul LM, Leitman S, Chang AE, et al. Observations on the systemic administration of autologous Iymphokine-activated cells and recombinant interleukin-2 to patients with metastatic cancer. New England Journal of Medicine 313: 1485-1492, 1985 Schellekens H, v.d. Meide PH. Animal studies with interferon. In De Maeyer E & Schellekens H (Eds) The biology of the interferon system, pp. 409-418, Elsevier Science Publishers BY Amsterdam, 1983 Schellekens H, de Reus A, v.d. Meide PH. The chimpanzee as a model to test the side effects of human interferons. Journal of Medical Primatology 13: 235-245, 1984 Scott GM. Interferon: pharmacokinetics and toxicity. Philosophical Transactions of the Royal Society London B 299: 91-107, 1982 Sergiescu D, Cerutti I, Efthymiou E, Kahan A, Chany e. Adverse effects of interferon treatment on the life span of NZB mice. Biomedicine 31: 48-51, 1979 Skoskievicz MJ, Colvin RB, Schneeberger EE, Russell PS. Widespread and selective induction of major histocompatibility complex-determined antigens in vivo by gamma interferon. Journal of Experimental Medicine 162: 1645-1664, 1985 Sondel PM, Kohler PC, Hank JA, Moore KH, Rosenthal NS, et al. Clinical and immunological effects of recombinant interleukin-2 given by repetitive weekly cycles to patients with cancer. Cancer Research 48: 2561-2567, 1988 Sosman JA, Kohler PC, Hank JA, Moore KA, Bechhofer R, et al. Repetitive weekly cycles of interleukin-2. II. Clinical and immunological effects of dose, schedule and addition of indomethacin. Journal ofthe National Cancer Institute 80: 14511461, 1988 Teelmann K, Hohbach C, Lehmann H. Preclinical safety testing of species-specific proteins produced with recombinant DNAtechniques. Archives of Toxicology 59: 195-200, 1986 Todd JA, Acha-Orbea H, Bell JI, Chao N, Fronek Z, et al. A molecular basis for MHC class II-associated autoimmunity. Science 240: 1003-1009, 1988 Yelcovsky HG, Federlin KF. Insulin-specific IgG and IgE antibody response in type I diabetic subjects exclusively treated with human insulin (recombinant DNA). Diabetes Care 5 (Suppl. 2): 126-128, 1982 Zbinden G. Biotechnology products intended for human use, toxicological targets and research strategies. In Graham CE (Ed.) Preclinical safety of biotechnology pr.oducts intended for human use, pp. 143-159, Alan R. Liss, Inc., New York, 1987 Zbinden G. Effects of recombinant human alpha interferon in a rodent cardiotoxicity model. Toxicology Letters, in press Zbinden G, Grimm L. Thrombogenic effects of xenobiotics. Archives of Toxicology (Suppl. 8): 131-141, 1985
Author's address: Dr G. Zbinden, Institute of Toxicology, Swiss Federal Institute of Technology and University of Zurich, Schorenstrasse 16, CH-8603 Schwerzenbach, Switzerland.
