1. A brief history of clinical immunotoxicology
 The area of immunotoxicology encompasses four categories of immunotoxic effects, namely immunosuppression, immunostimulation, hypersensitivity and autoimmunity, the clinical consequences of which are fairly well established nowadays thanks to major immunotoxic events that occurred during the last 3 decades.
 Immunosuppression can result in more frequent and often more severe infections as well as virus-induced neoplasia, such as skin cancers and lymphomas. This was illustrated by infectious complications reported shortly after the introduction of immunosuppressive drugs in kidney transplant patients (1966), an outbreak of polychlorobiphenyl (PCB)-induced Yusho disease due to accidental rice oil contamination in Japanese inhabitants (1968), polybromobiphenyl (PBB)-exposure of the Michigan rural population via contamination of cattle feed (1973), US FDA warning on infliximab-associated tuberculosis in rheumatoid patients (2001), or the first case reports of progressive multifocal leukoencephalopathy in natalizumab-treated patients (2005).
 Adverse effects resulting from immunostimulation include cytokine release-associated clinical manifestations, more frequent autoimmune diseases and hypersensitivity reactions to environmental allergens, and inhibition of cytochrome P450-dependent pathways. Although these adverse effects had been identified as early as 1985, awareness only grew with reports of cytokine release syndromes in muromonab-treated patients (1990) and mainly the "cytokine storm" in human healthy volunteers following injection of TGN1412 (2006). Hypersensitivity-related events include immunoallergic reactions involving specific antibodies or T lymphocytes, and pseudo-allergic reactions due to nonimmune-mediated release of some mediators involved in "true allergy". Hypersensitivity has been a major cause of market withdrawals, e.g. zomepirac, a NSAID causing anaphylactic shock (1983); Althesin°, an iv general anesthetic causing acute pseudo-allergic reactions via complement activation (1984); nomifensine, an antidepressant causing immuno-allergic hemolytic anemias (1992). More recently, cetuximab was reported to induce anaphylactic shock (2008). Auto-immunity can manifest as organ-specific reactions mimicking spontaneous autoimmune diseases (e.g. myasthenia) or systemic reactions bearing many dissimilarities with spontaneous autoimmune diseases (e.g. lupus syndrome vs. SLE). Major immunotoxic events linked to autoimmunity include withdrawal of the β-blocker practolol (1976), the Spanish toxic oil syndrome (1981), fasciitis-eosinophilia due to L-tryptophan (1989), autoimmune thyroiditis in IL-2-treated cancer patients (1991) or the first case reports of red pure cell aplasia in patients treated with recombinant erythropoietin (2002).

2. Past strategies for the clinical evaluation of immunotoxicity
 Clinical aspects of immunotoxicity have long been overlooked: no review papers on immunotoxicity evaluation during clinical trials have been published prior to 1998. Two major publications deserve particular attention. One of these, "Immune Function Test Batteries for Use in Environmental Health Field Studies" by Straight et al (ATSDR, Atlanta, 1984) defined 3 levels of tests recommended to evaluate the effects of environmental immunotoxicants in humans. The first level (basic tests) included serum levels of antinuclear antibodies, C reactive protein, IgG, IgM and IgA, and total proteins, white blood cell count, total lymphocyte and eosinophil counts, CD4⁺ and CD8⁺ lymphocyte counts. The second level (focused tests) only concerned the followup of abnormal basic test results related to immune deficiency (antibody levels to a given antigen, CH50 assay, tetrazolium dye reduction assay, mitogen-induced lymphoproliferation and skin tests), hypersensitivity (total and specific IgE serum levels, leukocyte histamine release assay and skin tests) and autoimmunity (antithyroglobulin, antimitochondrial, anti-phospholipid, and antismooth muscle auto-antibodies and rheumatoid factor). Finally, the third level (research tests) covered all other tests not recommended for use in field studies. Two years before, the subcommittee on immunotoxicology of the US National Research Council Committee on Biologic Markers published a more detailed document on "Biologic Markers in Immunotoxicology" (1982). A 3-tier approach was recommended. Tier 1 included markers to be measured in all persons potentially exposed to an immunotoxicant: humoral immunity (serum IgG, IgM, IgA levels, and secondary antibody response to specific antigens, e.g. tetanus or influenza vaccine), cellular immunity (total and differential blood cell counts, lymphocyte surface markers, e.g. CD3, CD4, CD8, CD20, and skin testing with recall antigens, e.g. candida, diphtheria, tetanus), autoimmunity (same autoantibodies as above). Tier 2 was restricted to persons with abnormal results in tier 1 testing: primary antibody response to specific antigens (e.g. KLH), in vitro lymphoproliferation assay (ConA, PHA) and primary DTH response to KLH, extended panel of surface markers (e.g. NK cells, monocytes, and activation markers), and finally serum cytokine levels (e.g. IL-1, IL-2, IL-6). Tier 3 was restricted to persons with abnormal results in tier 2 testing and included, case by case, NK cell activity, extended lymphoproliferation assay (e.g. anti-CD3), in vitro antigen-specific T lymphocyte cytotoxicity assay, immunoglobulin subclass levels and/or antiviral antibody titers. Two major conclusions were presented, which are still largely valid today: "[tests] for humoral, cellular, and nonspecific immunity [] are not sensitive enough to meaningfully detect modest immunodeficiency in populations of individuals exposed to immunotoxic agents" and "because available tests can lack the sensitivity required to detect modest immunodeficiency, a major focus should be on devising more sensitive tests for markers of immune impairment".

3. Current strategies for the clinical evaluation of immunotoxicity
 Clinical immune monitoring can prove to be useful for immunopharmacology purposes (immunomodulation via on- and off-target effects) and immune safety assessment when a safety issue already exists or to evaluate a potential new issue not seen in preclinical studies. Currently, no regulatory opinion or guidance on the clinical monitoring for immune function assessment has yet been published so that each sponsor has to decide whether clinical immune monitoring is indicated based on the interpretation of available information including nonclinical data. One exception is immunogenicity related to therapeutic proteins. Available assays and endpoints to be included in the immune monitoring of a drug candidate are not markedly different from those recommended above. The most commonly used endpoints nowadays include blood cell counts (especially neutrophils, monocytes and lymphocytes), standard clinical chemistry (albumin and proteins, CRP and fibrinogen serum levels), and lymphocyte subset immunophenotyping. Major issues regarding the later assay are the quality and reproducibility of results especially if the analysis is not performed in a central laboratory. The significance of changes in complement (C3, C4, CH50) or immunoglobulin serum levels is debatable, unless very profound changes are observed. Among assays, which tend to be more frequently performed during clinical trials, immunization studies are the leading assays. The immunizing agent is most often a vaccine, either a killed/inactivated vaccine (e.g. tetanus, diphtheria, influenza, or hepatitis A or B - live vaccines should be avoided) or a T-independent vaccine (pneumococcal polysaccharide vaccine). Neoantigens can also be used such as KLH (with a risk of cross-reactivity in shellfish allergic patients) and PhiX174 bacteriophage. Several critical issues should be considered: is the vaccine approved for clinical use? Will the measured response be a primary vs. boost response? Will controls be healthy or diseased? What is the role of background medication (e.g. methotrexate)? Is a robust and validated antibody assay available? What is the validity of selected endpoints, e.g. protective titers, differences in geometric means or antibody kinetics? Other commonly used assays include delayed-type hypersensitivity (DTH) skin testing to KLH or recall antigens, and lymphocyte function assays, such as antigen specific or mitogen-induced lymphocyte proliferation and cytokine production (e.g. Elispot).
 Many questions remain unsolved as regards the clinical relevance of the data generated: is a statistically significant decrease in one given endpoint to be considered an immunotoxicologically relevant finding? Which level of decrease is to be considered relevant? How to deal with inconsistent results across measured endpoints? Obviously, these questions are also pending in the preclinical setting.

4. Conclusion
A number of improvements are urgently needed regarding clinical immune monitoring. There is a lack of adequate standardization for most immune monitoring tests and the aim should be that standardization of immunological endpoints matches that of clinical chemistry. Efforts have also to be paid to correlate changes in selected endpoints with the expected occurrence of clinically significant adverse effects such as infections (i.e. validation). Indeed, monitoring infections during clinical trials to assess the immunosuppressive potential of a drug candidate requires large groups of treated vs. untreated patients. Immunotoxicology can also make significant progresses by tailoring preclinical assays and evaluation strategies to better identify immunotoxicity warnings that can be further assessed during clinical trials (i.e. translational immunotoxicology). Finally, specific immunotoxicity biomarkers are clearly needed to improve the immune safety of drug candidates. It is important to keep in mind that clinical immunotoxicology is still in its infancy. Nevertheless, there is an obvious need to better address immune safety issues during clinical trials. There are many questions, but only few answers available today...