【翻译】Development trends for monoclonal antibody cancer therapeutics
丁香园论坛
3230
Evolution of production methods
The total number of anticancer mAbs in clinical study by companies has risen steadily over time. In fact, the rate at which new oncology mAbs entered clinical study sponsored by commercial organizations has more than tripled since the 1980s, from 4.3 per year in the 1980s to 8.3 per year in the 1990s and 13.3 per year for 2000–2005. One reason for the increased investment in mAbs was the evolution of the discovery technology. The initial method for producing mAbs, first described in 1975 (Ref 6), involved use of mouse-derived hybridomas. However, in humans, murine mAbs were commonly (although not always) ineffective as cancer therapeutics. Studies revealed that murine candidates often had short circulating half-lives7 and patients frequently developed antibodies to the mouse-derived proteins, which limited their utility8, 9. In addition, only certain murine mAb isotypes have been shown to effectively bind to and activate elements of the human immune system, thereby triggering cytotoxic effector functions10. In practice, murine mAbs were mostly limited to acting as targeting agents for radioactive elements or cytotoxins that might kill targeted tumour cells.
Advances in genetic engineering over the years have provided numerous ways to design mAbs that are more robust and efficacious compared with the original murine versions11. Techniques suitable for generating chimeric and humanized mAbs that contain protein sequence from both human and murine sources were developed in the mid- to late 1980s (Refs 12–14). In addition, human mAbs derived from phage display and transgenic mice became available in the 1990s (Refs 15, 16). It should be noted that human mAbs were also made from human hybridomas, but this production method was not commercially viable because the cell lines did not reliably produce sufficient quantities of the desired mAb.
Technological advances have had a clear impact on the trends in mAb category in clinical study in the 1980s, 1990s and so far this decade (Fig. 1). Murine candidates constituted 86% of the total in clinical study in the 1980s. No single category has dominated since the 1990s because of the availability of a variety of mAb categories. The rate at which humanized mAbs entered clinical study increased dramatically between the 1980s and the 1990s (from 0.1 to 2.8 mAbs per year). Because human mAb discovery technology lagged behind that for humanized mAbs, the rate that human mAbs entered clinical study was still slow in the 1990s (0.9 mAbs per year). However, since 2000, humanized and human mAbs have been entering clinical study at approximately the same rate (4.3 versus 4.5 mAbs per year, respectively).
Figure 1 | Categories of monoclonal antibody cancer therapeutics entering clinical study during 1980–1989, 1990–1999 and 2000–2005.
Modes of action
mAbs are potentially capable of multiple functions. Efficacious anticancer mAbs must bind to an appropriate antigen in quantities sufficient to mediate a disease-relevant response. When tumour cell-surface antigens are targeted, the response should ultimately lead to destruction of the targeted cell. The mode of destruction can be direct (for example, via conjugated radioactive isotopes or toxins, or antibody triggered apoptosis) or indirect (for example, by activation of immune system components or blockade of critical receptors). In the case of soluble antigens, the mAb should sequester the target. The complexity of the mAb varies with the mode of action — immunoconjugates must be constructed to carry radioactive elements, cytotoxins or cytokines, whereas unmodified mAbs will suffice in other cases.
Conjugates. Conjugated mAbs can increase the specificity of chemo- or radiation therapy and improve the efficacy of immunotherapy, but have some drawbacks; they are more difficult to manufacture and may have greater safety issues compared with their naked counterparts. Despite this, immunoconjugates of various kinds constituted 44% of the total anticancer mAbs in clinical study to date (Fig. 2).
Figure 2 | Modes of action for immunoconjugate cancer therapeutics in clinical study, 1980–2005.
Most immunoconjugates have been designed to carry a radioactive isotope (radio-immunoconjugates) or a toxin (immunotoxins)17, 18, 19. However, the percentage of radiolabelled candidates constructed from the four main categories of mAbs has shown considerable variability (Fig. 2). Nearly equal percentages of murine and chimeric mAbs were radiolabelled (33% and 32%, respectively), but fewer humanized and only a small fraction of human mAbs were radiolabelled (22% and 5%, respectively). This is probably due, at least in part, to the lack of success with radiolabelled products in the 1980s and the timing of mAb production method development. Humanized and human anticancer mAbs did not enter the clinic in large numbers until the 1990s and 2000, well after clinical studies revealed the limitations of radio-immunoconjugates as cancer therapeutics (for example, mAbs did not deliver effective radiation doses, especially to solid tumour sites20, 21, complex chemistry was often required for conjugation, and there were potentially toxic effects on normal tissues). The two radioisotopes most commonly conjugated to mAbs studied in the clinic were Y-90 (2.7-day half-life) and I-131 (8.0-day half-life). Three radiolabelled mAbs — two murine, Y-90 ibritumomab tiuxetan (Zevalin; Biogen Idec) and I-131 tositumomab (Bexxar; Corixa/GSK), and one chimeric, I-131 ch-TNT (Shanghai Medipharm Biotech) — have been approved for non-Hodgkin's lymphoma or lung cancer, but only 14% of anticancer mAbs currently in the clinic are radio-immunoconjugates.
Immunotoxins composed of mAbs that are conjugated to a wide variety of either protein or small-molecule cytotoxins have also been studied in the clinic18. The protein cytotoxins used in this way include various modified versions of Pseudomonas exotoxin, Staphylococcus enterotoxin, neocarzinostatin, and the plant-derived molecules ricin (and ricin A chain) and gelonin. Small molecules incorporated in therapeutic immunotoxins included vinblastine, methotrexate, doxorubicin, calicheamicin and maytansine derivatives (auristatin immunoconjugates entered the clinic in 2006, so are not included in our analysis). Drugs are attached to mAbs through chemical linkers, whereas protein immunotoxins can be produced through either chemical conjugation or genetic engineering22. Development of both types of immunotoxins has posed challenges because the protein toxins can be immunogenic and the drug toxins may lack potency at the doses delivered to the tumour site17. In both cases the effectiveness of the product is reduced. Further potential limitations in efficacy and safety arise with chemically conjugated immunotoxins, which might be subject to decay of the linker used to join the antibody to the toxin. Another limitation in the choice of targets and antibodies arises with immunotoxins, which require internalization in order to have a cytotoxic effect on tumour cells. Of 44 immunotoxins that entered clinical study from 1980 to 2005, 27 (61%) were protein–mAb conjugates and 17 (39%) were drug–mAb conjugates, although an equal number of drug and protein immunoconjugates (6 candidates each) are now in the clinic. Only one immunotoxin, gemtuzumab ozogamicin (Mylotarg; UCB/Wyeth), a humanized mAb conjugated to calicheamicin, has been approved to date and 14% of mAbs now in the clinic (including fragments) are immunotoxins.
Overall, for anticancer mAbs, the trend has been towards unmodified mAbs; the percentage of mAb immunoconjugates studied in the clinic decreased from 56% to 49% to 31% in the 1980s, 1990s and 2000–2005, respectively. Of the current pipeline candidates, unmodified mAbs outnumber immunoconjugates by more than two to one, but improvements in key features such as potency or linker stability might spur additional study of immunoconjugates in the future.
Immune system activation. Unmodified mAbs make up just over half (56%) of the mAbs included in the data set. Some characteristics of unmodified mAbs, such as average half-life, vary with the isotype23, which is determined by the crystallizable-fragment (Fc) region. mAbs can function through more than one mechanism, but a common primary mode of action is the destruction of targeted cells through activation of components of the human immune system. For example, after binding to a target, mAbs can recruit effector cells such as natural killer cells, macrophages or neutrophils, or activate complement to destroy the target3, 24, 25, 26, 27. These two modes of action are referred to as antibody-dependent cell cytotoxicity (ADCC) and complement-dependent cytotoxicity (CDC), respectively, and are mediated through the Fc portion of mAbs. Some mAb isotypes (human IgG1 and IgG3, and murine IgG2a) are potentially more competent than others at inducing ADCC and CDC3, 26, 28, probably owing to their differential affinities for Fc receptors29. It should be noted that chimeric, humanized and human mAbs all have human Fc regions.
Of the 115 unmodified mAbs, 97 were full-sized and specific for a single target. mAb fragments with no Fc region, which cannot induce ADCC or CDC, and bispecific mAbs, which are designed to work through an alternate primary mechanism, were excluded. The isotype was identified for 81 (84%) of the 97 mAbs, so although comprehensive data were not available, general trends could be discerned. Of the unmodified mAbs with a known isotype, 77 (95%) were IgG, whereas 4 (5%) were IgM, which function primarily through CDC. The majority (80%) of the 61 IgG mAbs with human Fc regions (murine mAbs excluded) were IgG1. These observations suggest that commercial interest has focused on chimeric, humanized and human mAbs that might elicit ADCC and CDC, although they could also have been designed to have other effects (such as receptor blockade). In fact, although all seven currently approved unmodified anticancer mAbs are IgG1 and potentially have the ability to function through ADCC or CDC mechanisms, only two — rituximab (Rituxan; Biogen Idec) and alemtuzumab (Campath; Genzyme) — are believed to use ADCC and CDC as their primary mode of action. The other five products function through alternative primary modes of action (blockade of growth factor–receptor interaction, receptor downmodulation and inhibition of signalling). Again, it should be noted that mAbs are not necessarily restricted to a single mode of action. For example, trastuzumab (Herceptin; Genentech) has been reported to induce ADCC as a possible alternative or additional mechanism to human epidermal growth factor receptor 2 (HER2) downmodulation30. Similarly, besides eliciting ADCC and CDC mechanisms upon binding to CD20+ cells, rituximab can also affect intracellular calcium levels and thereby induce cell death through apoptosis31.
Although murine mAbs were largely abandoned as cancer therapeutics when humanized and human candidates became available, there is evidence that murine IgG2a mAbs could potentially induce ADCC and CDC26, 29, and that murine IgG may also activate human effector function32. Seven of the sixteen unmodified murine IgG mAbs in the data set were identified as IgG2a, four mAbs were IgG3, four were IgG1, and one mAb was known only to be IgG. No correlation was observed between the isotype of the naked murine mAbs and the available information on the primary reason for termination of development (for example, safety, efficacy or commercial). The only approved unmodified murine anticancer mAb, edrecolomab, was IgG2a, but it did not prove to be sufficiently efficacious5 and was discontinued.
Alternative modes of action of unmodified mAbs. As already noted above, in addition to ADCC and CDC modes of tumour-cell killing, unmodified mAbs can also exert their primary effects on tumour cells in less direct ways. These include blockade of relevant receptors and inhibition of critical growth factors, both of which could inhibit proliferation of targeted cells and prevent the advance of cancerous disease. Three approved unmodified mAbs (all IgG1) are believed to function primarily by blocking epidermal growth factor receptor (EGFR, also known as HER1) — cetuximab (Erbitux; ImClone/Bristol-Myers Squibb), nimotuzumab (TheraCIM; YM BioSciences) and panitumumab (Vectibix; Amgen). These products inhibit binding of EGF and subsequent activation of the EGF receptor33, 34. The in vivo mechanism of action of trastuzumab has not yet been resolved; downmodulation of HER2-mediated signalling and increased tumour-cell apoptosis have been proposed in addition to recruitment of ADCC35, 36. One approved unmodified mAb — bevacizumab (Avastin; Genentech/Roche) — binds to the ligand vascular endothelial growth factor (VEGF) and ultimately inhibits tumour angiogenesis37.
Several mAbs in the pipeline use additional primary modes of action such as induction of programmed cell death or immunomodulation. For example, agonistic antibodies to the death receptor TRAILR1 (HGS-ETR1; Human Genome Sciences) directly trigger apoptosis of tumour cells38. In addition, two human mAbs (CP-675,206; Pfizer, and ipilimumab; Medarex) that target cytotoxic T lymphocyte-associated antigen 4 (CTLA4) are undergoing Phase III studies in melanoma patients. Instead of acting directly on tumour cells, these mAbs target molecules that interfere with the body's response to tumours39. CTLA4 engagement on T cells can suppress the immune response to cancer, so blocking this molecule may allow a more effective antitumour response40.
The total number of anticancer mAbs in clinical study by companies has risen steadily over time. In fact, the rate at which new oncology mAbs entered clinical study sponsored by commercial organizations has more than tripled since the 1980s, from 4.3 per year in the 1980s to 8.3 per year in the 1990s and 13.3 per year for 2000–2005. One reason for the increased investment in mAbs was the evolution of the discovery technology. The initial method for producing mAbs, first described in 1975 (Ref 6), involved use of mouse-derived hybridomas. However, in humans, murine mAbs were commonly (although not always) ineffective as cancer therapeutics. Studies revealed that murine candidates often had short circulating half-lives7 and patients frequently developed antibodies to the mouse-derived proteins, which limited their utility8, 9. In addition, only certain murine mAb isotypes have been shown to effectively bind to and activate elements of the human immune system, thereby triggering cytotoxic effector functions10. In practice, murine mAbs were mostly limited to acting as targeting agents for radioactive elements or cytotoxins that might kill targeted tumour cells.
Advances in genetic engineering over the years have provided numerous ways to design mAbs that are more robust and efficacious compared with the original murine versions11. Techniques suitable for generating chimeric and humanized mAbs that contain protein sequence from both human and murine sources were developed in the mid- to late 1980s (Refs 12–14). In addition, human mAbs derived from phage display and transgenic mice became available in the 1990s (Refs 15, 16). It should be noted that human mAbs were also made from human hybridomas, but this production method was not commercially viable because the cell lines did not reliably produce sufficient quantities of the desired mAb.
Technological advances have had a clear impact on the trends in mAb category in clinical study in the 1980s, 1990s and so far this decade (Fig. 1). Murine candidates constituted 86% of the total in clinical study in the 1980s. No single category has dominated since the 1990s because of the availability of a variety of mAb categories. The rate at which humanized mAbs entered clinical study increased dramatically between the 1980s and the 1990s (from 0.1 to 2.8 mAbs per year). Because human mAb discovery technology lagged behind that for humanized mAbs, the rate that human mAbs entered clinical study was still slow in the 1990s (0.9 mAbs per year). However, since 2000, humanized and human mAbs have been entering clinical study at approximately the same rate (4.3 versus 4.5 mAbs per year, respectively).
Figure 1 | Categories of monoclonal antibody cancer therapeutics entering clinical study during 1980–1989, 1990–1999 and 2000–2005.
Modes of action
mAbs are potentially capable of multiple functions. Efficacious anticancer mAbs must bind to an appropriate antigen in quantities sufficient to mediate a disease-relevant response. When tumour cell-surface antigens are targeted, the response should ultimately lead to destruction of the targeted cell. The mode of destruction can be direct (for example, via conjugated radioactive isotopes or toxins, or antibody triggered apoptosis) or indirect (for example, by activation of immune system components or blockade of critical receptors). In the case of soluble antigens, the mAb should sequester the target. The complexity of the mAb varies with the mode of action — immunoconjugates must be constructed to carry radioactive elements, cytotoxins or cytokines, whereas unmodified mAbs will suffice in other cases.
Conjugates. Conjugated mAbs can increase the specificity of chemo- or radiation therapy and improve the efficacy of immunotherapy, but have some drawbacks; they are more difficult to manufacture and may have greater safety issues compared with their naked counterparts. Despite this, immunoconjugates of various kinds constituted 44% of the total anticancer mAbs in clinical study to date (Fig. 2).
Figure 2 | Modes of action for immunoconjugate cancer therapeutics in clinical study, 1980–2005.
Most immunoconjugates have been designed to carry a radioactive isotope (radio-immunoconjugates) or a toxin (immunotoxins)17, 18, 19. However, the percentage of radiolabelled candidates constructed from the four main categories of mAbs has shown considerable variability (Fig. 2). Nearly equal percentages of murine and chimeric mAbs were radiolabelled (33% and 32%, respectively), but fewer humanized and only a small fraction of human mAbs were radiolabelled (22% and 5%, respectively). This is probably due, at least in part, to the lack of success with radiolabelled products in the 1980s and the timing of mAb production method development. Humanized and human anticancer mAbs did not enter the clinic in large numbers until the 1990s and 2000, well after clinical studies revealed the limitations of radio-immunoconjugates as cancer therapeutics (for example, mAbs did not deliver effective radiation doses, especially to solid tumour sites20, 21, complex chemistry was often required for conjugation, and there were potentially toxic effects on normal tissues). The two radioisotopes most commonly conjugated to mAbs studied in the clinic were Y-90 (2.7-day half-life) and I-131 (8.0-day half-life). Three radiolabelled mAbs — two murine, Y-90 ibritumomab tiuxetan (Zevalin; Biogen Idec) and I-131 tositumomab (Bexxar; Corixa/GSK), and one chimeric, I-131 ch-TNT (Shanghai Medipharm Biotech) — have been approved for non-Hodgkin's lymphoma or lung cancer, but only 14% of anticancer mAbs currently in the clinic are radio-immunoconjugates.
Immunotoxins composed of mAbs that are conjugated to a wide variety of either protein or small-molecule cytotoxins have also been studied in the clinic18. The protein cytotoxins used in this way include various modified versions of Pseudomonas exotoxin, Staphylococcus enterotoxin, neocarzinostatin, and the plant-derived molecules ricin (and ricin A chain) and gelonin. Small molecules incorporated in therapeutic immunotoxins included vinblastine, methotrexate, doxorubicin, calicheamicin and maytansine derivatives (auristatin immunoconjugates entered the clinic in 2006, so are not included in our analysis). Drugs are attached to mAbs through chemical linkers, whereas protein immunotoxins can be produced through either chemical conjugation or genetic engineering22. Development of both types of immunotoxins has posed challenges because the protein toxins can be immunogenic and the drug toxins may lack potency at the doses delivered to the tumour site17. In both cases the effectiveness of the product is reduced. Further potential limitations in efficacy and safety arise with chemically conjugated immunotoxins, which might be subject to decay of the linker used to join the antibody to the toxin. Another limitation in the choice of targets and antibodies arises with immunotoxins, which require internalization in order to have a cytotoxic effect on tumour cells. Of 44 immunotoxins that entered clinical study from 1980 to 2005, 27 (61%) were protein–mAb conjugates and 17 (39%) were drug–mAb conjugates, although an equal number of drug and protein immunoconjugates (6 candidates each) are now in the clinic. Only one immunotoxin, gemtuzumab ozogamicin (Mylotarg; UCB/Wyeth), a humanized mAb conjugated to calicheamicin, has been approved to date and 14% of mAbs now in the clinic (including fragments) are immunotoxins.
Overall, for anticancer mAbs, the trend has been towards unmodified mAbs; the percentage of mAb immunoconjugates studied in the clinic decreased from 56% to 49% to 31% in the 1980s, 1990s and 2000–2005, respectively. Of the current pipeline candidates, unmodified mAbs outnumber immunoconjugates by more than two to one, but improvements in key features such as potency or linker stability might spur additional study of immunoconjugates in the future.
Immune system activation. Unmodified mAbs make up just over half (56%) of the mAbs included in the data set. Some characteristics of unmodified mAbs, such as average half-life, vary with the isotype23, which is determined by the crystallizable-fragment (Fc) region. mAbs can function through more than one mechanism, but a common primary mode of action is the destruction of targeted cells through activation of components of the human immune system. For example, after binding to a target, mAbs can recruit effector cells such as natural killer cells, macrophages or neutrophils, or activate complement to destroy the target3, 24, 25, 26, 27. These two modes of action are referred to as antibody-dependent cell cytotoxicity (ADCC) and complement-dependent cytotoxicity (CDC), respectively, and are mediated through the Fc portion of mAbs. Some mAb isotypes (human IgG1 and IgG3, and murine IgG2a) are potentially more competent than others at inducing ADCC and CDC3, 26, 28, probably owing to their differential affinities for Fc receptors29. It should be noted that chimeric, humanized and human mAbs all have human Fc regions.
Of the 115 unmodified mAbs, 97 were full-sized and specific for a single target. mAb fragments with no Fc region, which cannot induce ADCC or CDC, and bispecific mAbs, which are designed to work through an alternate primary mechanism, were excluded. The isotype was identified for 81 (84%) of the 97 mAbs, so although comprehensive data were not available, general trends could be discerned. Of the unmodified mAbs with a known isotype, 77 (95%) were IgG, whereas 4 (5%) were IgM, which function primarily through CDC. The majority (80%) of the 61 IgG mAbs with human Fc regions (murine mAbs excluded) were IgG1. These observations suggest that commercial interest has focused on chimeric, humanized and human mAbs that might elicit ADCC and CDC, although they could also have been designed to have other effects (such as receptor blockade). In fact, although all seven currently approved unmodified anticancer mAbs are IgG1 and potentially have the ability to function through ADCC or CDC mechanisms, only two — rituximab (Rituxan; Biogen Idec) and alemtuzumab (Campath; Genzyme) — are believed to use ADCC and CDC as their primary mode of action. The other five products function through alternative primary modes of action (blockade of growth factor–receptor interaction, receptor downmodulation and inhibition of signalling). Again, it should be noted that mAbs are not necessarily restricted to a single mode of action. For example, trastuzumab (Herceptin; Genentech) has been reported to induce ADCC as a possible alternative or additional mechanism to human epidermal growth factor receptor 2 (HER2) downmodulation30. Similarly, besides eliciting ADCC and CDC mechanisms upon binding to CD20+ cells, rituximab can also affect intracellular calcium levels and thereby induce cell death through apoptosis31.
Although murine mAbs were largely abandoned as cancer therapeutics when humanized and human candidates became available, there is evidence that murine IgG2a mAbs could potentially induce ADCC and CDC26, 29, and that murine IgG may also activate human effector function32. Seven of the sixteen unmodified murine IgG mAbs in the data set were identified as IgG2a, four mAbs were IgG3, four were IgG1, and one mAb was known only to be IgG. No correlation was observed between the isotype of the naked murine mAbs and the available information on the primary reason for termination of development (for example, safety, efficacy or commercial). The only approved unmodified murine anticancer mAb, edrecolomab, was IgG2a, but it did not prove to be sufficiently efficacious5 and was discontinued.
Alternative modes of action of unmodified mAbs. As already noted above, in addition to ADCC and CDC modes of tumour-cell killing, unmodified mAbs can also exert their primary effects on tumour cells in less direct ways. These include blockade of relevant receptors and inhibition of critical growth factors, both of which could inhibit proliferation of targeted cells and prevent the advance of cancerous disease. Three approved unmodified mAbs (all IgG1) are believed to function primarily by blocking epidermal growth factor receptor (EGFR, also known as HER1) — cetuximab (Erbitux; ImClone/Bristol-Myers Squibb), nimotuzumab (TheraCIM; YM BioSciences) and panitumumab (Vectibix; Amgen). These products inhibit binding of EGF and subsequent activation of the EGF receptor33, 34. The in vivo mechanism of action of trastuzumab has not yet been resolved; downmodulation of HER2-mediated signalling and increased tumour-cell apoptosis have been proposed in addition to recruitment of ADCC35, 36. One approved unmodified mAb — bevacizumab (Avastin; Genentech/Roche) — binds to the ligand vascular endothelial growth factor (VEGF) and ultimately inhibits tumour angiogenesis37.
Several mAbs in the pipeline use additional primary modes of action such as induction of programmed cell death or immunomodulation. For example, agonistic antibodies to the death receptor TRAILR1 (HGS-ETR1; Human Genome Sciences) directly trigger apoptosis of tumour cells38. In addition, two human mAbs (CP-675,206; Pfizer, and ipilimumab; Medarex) that target cytotoxic T lymphocyte-associated antigen 4 (CTLA4) are undergoing Phase III studies in melanoma patients. Instead of acting directly on tumour cells, these mAbs target molecules that interfere with the body's response to tumours39. CTLA4 engagement on T cells can suppress the immune response to cancer, so blocking this molecule may allow a more effective antitumour response40.
