For the autoantibody aficionados out there, here’s a problem I’ve been thinking about a lot recently: Typically, any antibody that binds a self-antigen is considered an “autoantibody.” But when you examine reactivity patterns, an important distinction emerges between antibodies that bind predominantly to a single antigen (e.g., anti-TSHR antibodies in Graves’ disease) and promiscuous antibodies with highly polyreactive features. I spend most of my time thinking about the former. These are functional antibodies with specific biological effects, and I consider them a lens for discovering causal biology in humans. These are the types of responses we hoped to find when we developed REAP. I also suspect this is what most people intuitively mean when they use the term “autoantibody.” It’s unclear to me that the latter should be considered autoantibodies at all. They aren’t exclusively autoreactive: these antibodies often bind multiple unrelated self-antigens as well as microbial or viral molecular patterns. Their promiscuity is likely an evolved functional feature, rather than an immunological accident. They frequently become prominent during inflammation and can arise through extrafollicular B cell responses with reduced dependence on cognate T cell help. IGHV4-34–encoded responses exemplify this biology. These antibodies are associated with autoimmune disease and can clearly drive pathology, as in SLE, but their effects are often difficult to attribute to a single antigen–antibody interaction. I believe we should NOT call responses like these “autoantibodies” at all. We should call them what they are: polyreactive antibodies. The problem with calling both groups “autoantibodies” is that it conflates two entirely different phenomena. These antibody classes can have different origins, physiological effects, and therapeutic implications. Understanding the nature of the antibody responses likely will have a major impact whether we should pursue total immune reset vs precision antigen-specific immunomodulation vs IGHV4-34-directed targeting. Distinguishing antigen-specific autoantibodies from polyreactive responses is also critical because investigators sometimes infer specificity for an individual antigen when they are actually studying a broader polyreactive response. They may then attribute the effects of polyreactive antibodies to antigen-specific biology, which is a serious mechanistic error. This becomes particularly problematic when human antibodies are transferred into nonhuman preclinical models. For example, an increasingly popular approach involves injecting human IgG into mice. Polyreactive antibodies are known to be strongly xenoreactive, but the sets of human and mouse antigens they recognize might not overlap. That can make downstream effects extremely difficult to interpret, even before considering differences in Fc biology, pharmacokinetics, and the immunogenicity of human IgG in mice. I clearly have a dim view of this approach. Distinguishing monospecific from polyreactive responses in seromic data from patient samples can be very difficult, especially when both types of reactivity are present. However, we are having increasing success with new metrics (tentatively including a “promiscuity index”) that identify patterns of inter-antigen reactivity correlation that recur across patients in a dataset. These metrics do not prove polyreactivity, but they can flag signals that may be better characterized as polyreactive rather than specifically autoreactive and (de)prioritize them for experimental validation. I’m also hopeful that it may be possible to selectively deplete polyreactive antibodies from samples before specificity analysis, allowing seromic signals to be attributed more confidently to specific autoreactivity rather than polyreactivity. Anyway, if you got this far, thank you for reading and please add your thoughts below.