Infections and Acute Lymphoblastic Leukemia: Is the Sum Worth More than the Parts? Evidence from Birth Characteristics

The formation of the normal human immune system requires programmed mutations of antibody and T-cell receptor genes, creating an adaptive immune “genome” that is 4 orders of magnitude larger than the size of the human genome (1). This process proceeds in earnest upon environmental stimulation immediately after birth, and it is not surprising that the enzymes evolutionarily engineered to mutate-specific immune receptor genes can occasionally go awry resulting in off-target mutations contributing to the most common cancer in children, acute lymphoblastic leukemia (ALL). Given the tight interplay between the exposed antigenic repertoire and the development of a normal educated immune system (both innate and adaptive branches), a disordered or maladaptive response to infectious agents may impact the development of ALL which is essentially a cancer of the adaptive immune system. The effect of such exposures and their combinations is the subject of a Brazilian epidemiology study led by Pombo-de-Oliveira and colleagues presented in this commentary (2).

Direct measures of infections early in life are quite challenging to obtain for a disease as rare as childhood ALL, where causative exposures are typically studied retrospectively. Investigators are dependent on proxy measures including birth and childhood characteristics that typify patterns of early infectious exposures. Such birth characteristics include birth order and mode of delivery, while postnatal immune exposures include daycare attendance, vaccinations, and breastfeeding (Fig. 1; ref. 3). While individual epidemiologic case–control investigations examine these characteristics one at a time, rarely do they explore more than one such factor within the same investigation. Pombo-de-Oliveira and colleagues focus on two well-established “risk factors” for ALL—mode of birth [cesarean section (C-section) or vaginal] and birth order (first or later). Each of these conditions is associated with increased risk of ALL in prior studies. First, scheduled or elective C-section was clearly associated with ALL risk in a Greek study (4), followed by confirmation in an international meta-analysis (5). Second, birth order, or the identity as the first born in a family as opposed to the second or later born, was found to result in increased ALL risk (6) and confirmed in several subsequent investigations. Pombo-de-Oliveira and colleagues replicate these findings in a large Brazilian epidemiologic study, and furthermore include both risk factors in the same statistical model—showing a greater than multiplicative interaction, or synergy, when both are examined together. Interestingly, risk was further increased by the lack of breastfeeding, another postnatal characteristic facilitating immune isolation. In sum, the result is compatible with the longstanding “delayed infection” hypothesis originally explicated by Greaves (7), wherein a lack of antigenic stimulation early in life leaves the immune system vulnerable to damaging fulminant infections later in childhood.

A perspective on the predominant hypotheses regarding infection and ALL is appropriate. These hypotheses are (i) delayed exposure to common infections and microbial colonization in infancy may lead to aberrantly vigorous and damaging immune responses later in childhood, stimulating full-blown leukemic transformation in children harboring preleukemic clones and enhancing ALL risk [the “delayed infection hypothesis” from Greaves (3)], and (ii) a specific leukemia-causing virus may infect children who are immunologically naïve to such a virus in situations where enhanced population mixing occurs at community-wide scales (8). Where these related hypotheses differ is the specificity of infectious agent, and while Pombo-de-Oliveira and colleagues do not examine specific agents, the broad population-based investigation favors the role of common, ubiquitous, and endemic antigenic stimulation. A recently discovered specific infectious agent related to childhood ALL is cytomegalovirus (CMV) which is was found both at birth more commonly in children with ALL than controls (9), and in leukemic lymphoblasts at diagnosis (10). The CMV-ALL association does not conveniently fit into either of the two prevailing hypotheses concerning infection and ALL because neonatal CMV infection cannot be portrayed as “delayed,” nor has CMV has ever been proven to be an oncogenic virus. CMV's ubiquity in our population (>80% seropositivity worldwide, in both developed and developing countries) argues that CMV itself should not be the causative agent given the rarity of ALL in children, and is unlikely to be playing an instrumental role in the observations of Pombo-de-Oliveira and colleagues Instead, CMV may shape the “responder status” of the neonatal immune system to antigenic stimulation from other sources. CMV's complex interaction with the host immune system and ability to alter the response to subsequent antigenic exposures gives CMV the potential to act as an immune system accessory to the delayed infection paradigm (11, 12). Future investigations should include assessment of CMV and its impact or collaborative role with well-established ALL risk factors including birth order and delivery mode.

C-section, birth order, and breastfeeding all involve antigenic challenges, but each of them also impacts additional potential causal pathways that are not necessarily related to immune function. The risk imparted by C-section has been attributed to two separate mechanisms, described in Pombo-de-Oliveira and colleagues, these being the “adrenal hypothesis” related to stress-induced cortisol exposure proposed by Schmiegelow and colleagues (13), and the microbiome “trained immunity” inoculation provided by vaginal birth, described most recently by Hauer, Fischer, and Borkhardt (14). In the current study, the risk associated with birth order is attributed exclusively to infection exposures. It should be noted that birth order, or the ordinal position of a child within its family, is also associated with multiple conditions—including those as varied as mental health disorders, immunologic origin diseases (multiple sclerosis, allergies, type 1 diabetes), and a variety of cancer sites in children (15–18). While birth order is not necessarily itself a causal factor, it is inherently associated with prenatal developmental and environmental characteristics likely contributing to these health conditions, as well as postnatal immunologic stimuli (17) thought to convey ALL risk in the current analysis. The host of conditions upon which first-borns are at a higher risk include diabetes (19), high blood pressure (15), most childhood cancers (18), metabolic disorders (20), and immune-related disorders including ALL as mentioned above. Interestingly, first-born children are at lower risk for childhood acute myeloid leukemia (18) presumably via entirely different mechanisms than its sister cancer ALL. Apart from postnatal immune exposures, first born subjects have unique in utero characteristics including less sufficient placentation, higher estrogen levels, and lower insulin sensitivity, which could all contribute to disease risks (21–24). These in utero characteristics impact development in a systematic fashion that may be responsible for some ALL risk apart from birth order's impact on postnatal infectious exposures. Birth order indeed is related to greater height (compared with later-born siblings) which may be a consequence of greater bone growth in utero (15). Another well-recognized ALL risk factor is birthweight, which is thought to result in an increased stem cell pool and multiple growth factors (25); certainly birth order could be connected to such growth-related mechanisms. As an observational epidemiologic analysis, the Pombo-de-Oliveira and colleagues study is not capable of directly addressing such mechanisms, but the synergistic “collaboration” between C-section and birth order strongly favors the role an impact from infection and antigenic exposures rather than other mechanisms—given that a lack of infection and antigenic exposures are attributes shared “in common” among cesarean birth, first order birth, and lack of breastfeeding exposure after birth. Patterns of antigenic exposures are certainly principal modifiable risk factors for ALL thereby indicating potential for ALL prevention strategies.

So how do we proceed? While the results of Pombo-de-Oliveira and colleagues reveal tantalizing clues on ALL risk, they suffer from the dull instruments of proxy exposure measurements that plague retrospective case–control studies. Biological measures of infectious intermediates between these exposures and disease development are needed to fully understand the mechanism by which antigenic exposures contribute to leukemogenesis. Measurements from prediagnostic biological samples will be crucial to answer these questions. The use of state-maintained birth blood repositories has allowed investigation of some characteristic immune endpoints such as perinatal cytokine or viral exposures, and studies of postnatal medical records provide more definitive clues on infections, but pediatric cohort studies with actively stored biological specimens (blood, stool, etc.) from postnatal healthy children who later contract leukemia will be ideal to fully illuminate pathways between infectious disease history and development of ALL and potential avenues for prevention. Our children's health depends on it.

No disclosures were reported.

This work is funded by NCI (R01CA268494, J.L. Wiemels).