Infection in infancy and subsequent risk of developing lymphoma in children and young adults

Immunodeficiency and infections have consistently been associated with an increased risk of developing non-Hodgkin lymphoma (NHL),^1-4  although the degree to which these factors have contributed to the increase in population incidence rates of NHL in the second half of the 20th century is unknown. By using data from the Jerusalem Perinatal Study (JPS) cohort, among 24 554 nonmalformed singletons, Paltiel et al⁵  found that infants who were hospitalized for an infectious disease in the first year of life had a 3-fold greater risk of developing NHL in adolescence and young adulthood. No excess risk of Hodgkin lymphoma (HL) was found. To confirm and extend this finding, we conducted a large registry linkage study including 1125 NHL patients, 1388 HL patients, and 10 051 matched population-based control patients from Sweden.

Our samples and methods have been previously described^6,7  In brief, the Swedish Cancer Registry has operated since 1958 with near-complete coverage of the population. The Swedish Inpatient Registry started in 1965, so we included all lymphoma patients born in 1964 or later. A total of 1125 NHL and 1388 HL patients diagnosed from 1965 to 2004 were included. By using available ICD10 and Systematized Nomenclature of Medicine (ie, SNOMED) codes (available from 1993), we subclassified some NHL patients by using current World Health Organization criteria. We assessed infectious exposures in relation to risk of B- and T-cell NHL, respectively. Within the B-NHLs group, we classified diffuse large B-cell lymphoma and Burkitt lymphoma as “aggressive” and follicular, marginal zone, and mantle cell lymphoma, and hairy cell leukemia as “indolent.”

For each patient, 4 population-based control patients (matched by sex, year of birth, and county of residence) were chosen randomly from the Swedish Population database. All control patients were alive at the date of lymphoma diagnosis for the corresponding cases and had no cancer recorded previous to that date of the cases' diagnosis. Information on occurrence of infectious diseases in the first year of life was obtained through digital record linkage from the Inpatient Registry by use of the 7th, 8th, 9th, and 10th revisions of the International Classification of Diseases diagnoses codes for specific infections. To eliminate the bias resulting from infection as the first manifestation or trigger for a diagnostic work-up for lymphoma, we eliminated infections occurring less than 1 year before lymphoma registration, thus including cases with onset at age 2 or older. We used logistic regression to determine odds ratios (ORs) of infection history in cases compared with control patients controlling for sex, year of birth (4 categories), and year of diagnosis (4 categories). We computed 95% confidence intervals (95% CIs) and χ²P values.

Table 1 shows the demographic characteristics of the patients and control patients. The sex distributions are as expected on the basis of population incidence rates. The median age at diagnosis was 18 and 21 years, respectively for NHL and HL, reflecting the years of coverage of the hospital registry. Most patients were diagnosed after 1980, and 55% of NHL patients could be subtyped.

Table 2 shows the ORs (and CIs) for infection as a risk factor for subsequent development of lymphomas. For NHL as a whole, there was a nonsignificant excess risk associated with infection; the risk was similar magnitude for childhood and young adult cases. However, considering B-cell NHL only, the OR associated with infection was 1.71 and borderline significant. For aggressive B-cell NHL, the OR was 2.1-fold increased (95% CI 1.11-4.04; P = .02). Indolent lymphomas are a rare occurrence in this age group, and there were no cases of infection among these patients.

There were 14 patients with aggressive NHL subtypes with infection in infancy. One-half of these had Burkitt lymphoma and the remaining had anaplastic large B-cell and centroblastic lymphomas. The types of infections recorded were unremarkable because most were intestinal and respiratory infections.

For HL patients, there was a nonsignificant excess of infections in infancy compared with control patients, and risks were similar when we stratified the patients with childhood versus adolescent/young adult onset (Table 2). Risk was also the same among 651 patients with the nodular sclerosis subtype (not shown).

In this large, population-based case-control study of databases with high degrees of completeness⁸  and validity,⁹  we found that children and young adults with aggressive B-cell lymphoma were significantly more likely to have been hospitalized for infection in infancy than healthy age and sex-matched population-based control patients. This finding strengthens the original observation by Paltiel et al,⁵  who reported an increased risk of NHL among Israelis with similar exposures, and suggests that the association applies mainly to aggressive B-cell NHLs. The 2 studies cover a similar period in calendar time and therefore should reflect parallel types of exposures, as well as opportunities for these age cohorts to receive vaccinations or antibiotic treatment, where appropriate.

There are several potential mechanisms for this association. The infection itself may trigger continuous antigenic and immune stimulation. This model applies to Helicobacter pylori,¹⁰ Campylobacter jejuni,¹¹  or hepatitis C,¹²  in which an ongoing immune response triggers a clonal proliferation of B lymphocytes. Another possibility is that immune suppression caused by the infection facilitates the development of NHL, possibly through viral reactivation.¹³  The treatment for infections may itself cause immune modulation or suppression, which predisposes patients to NHL. There is increasing evidence that the presence of bacteria in the gut in infancy is required for development of the immune system.¹⁴  Infections in infancy may reflect innate but subtle immunodeficiency.¹⁵  Perhaps these immune defects have been masked by medical progress, and infants with these diseases would have previously died. In Western societies, they survive and are susceptible to the effects of further exposures, which may predispose them to NHL.

The strengths of our study include its size and population-based design. The Swedish cohort is unique in the richness of the data concerning childhood hospitalization and in the completeness of cancer ascertainment.⁸  In this study and in the JPS cohort, history of hospitalization was ascertained by use of the same databases both for cases and control patients, thus obviating recall bias. Limitations include incomplete information regarding NHL subtype and the limited power to study associations by NHL subtype given the rarity of the disease. Before 1993, most NHL patients were not subclassified in the cancer registry. The B-cell NHLs as well as aggressive and indolent subtypes were all diagnosed since 1993. If we analyze all NHL patients diagnosed since 1993, the results are similar to those of the entire group (Table 2). If we assume that all of the unclassified NHLs are B-cell type, the results are similar to those of known B-cell origin (data not shown). Finally, the Swedish hospital registry had regional but not complete nationwide coverage during the early years with possible less than-ascertainment of hospitalized infections. However, one matching factor for cases and control patients was county of residence, which should limit any less than-ascertainment bias. The consistency of these findings across study designs (cohort and case-control) in 2 geographic regions, and in populations with different ethnicities, strengthens the notion that this is not a chance association.

The findings in both the current and the JPS study⁵  are limited to children and young adults with NHL, as well as infections resulting in hospitalization. NHL incidence rates in children have been constant over time, whereas adolescents and young adults have shown secular increases.¹⁶  Our data are not informative for determining whether these associations are present in older adults, so we do not know whether this mechanism could have contributed to the large secular trends observed in NHL incidence rates in older adults. The findings of the current study, coupled with those of Paltiel et al⁵  strengthen the link between infection and B-cell malignancies. Future molecular studies are needed to elucidate the mechanisms underlying these findings.

This work was supported by the Intramural Research Program of the National Cancer Institute, National Institutes of Health and by grants from the Swedish Cancer Society, Stockholm County Council, and the Karolinska Institutet Foundations.

National Institutes of Health

Contribution: L.R.G., O.L., and O.P. designed the research; O.L., S.Y.K., and M.B. obtained data; L.R.G. analyzed the data; L.R.G. and O.P. drafted the manuscript; and all authors participated in the interpretation of the data and read, gave comments, and approved the final version of the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Lynn R. Goldin, PhD, Genetic Epidemiology Branch, DCEG, NCI, 6120 Executive Blvd, Rm 7008, MSC 7236, Bethesda, MD 20892-7236; e-mail:goldinl@mail.nih.gov.