Mode of Delivery, Birth Characteristics, and Early-Onset Non-Hodgkin Lymphoma in a Population-Based Case–Control Study

Non-Hodgkin lymphoma (NHL) is one of the most commonly diagnosed cancers among children (0–14 years) and adolescents and young adults (AYA; 15–39 years) in the United States (1). NHL is an umbrella term comprising a large number of subtypes, many with distinct risk factors, at least in adults (2). At younger ages, patients tend to be diagnosed with NHL subtypes that are more aggressive and higher grade compared with those diagnosed at older ages including Burkitt lymphoma, diffuse large B-cell lymphoma (DLBCL), and T-cell lymphoma (3). Despite differences in the clinical presentation of NHL based on age of diagnosis, most etiologic studies have not included large numbers of pediatric and adolescent cases or addressed potential heterogeneity across varying ages at diagnosis. While some studies have focused exclusively on children, AYA have higher NHL incidence and lower rate of survival compared with children (3, 4). Noting that risk factors for early-onset NHL have been inadequately examined, we decided to focus on children and AYA.

Immunologic factors are associated with risk of NHL overall (5–8), and exposures at birth may influence the immune system in early life. Mode of delivery [vaginal versus cesarean section (c-section)] influences the microbiome composition at birth, and early-life perturbations in the microbiome have been linked to immuno-modulated diseases in later life (9–12). Moreover, whether c-section was elective or emergency has implications for microbial and hormonal exposures during delivery. Emergency c-section typically occurs after a period of labor during which the infant is exposed to stress hormones which promote cell differentiation (13) and microbes which are released during amniotic membrane rupture (14). Infants delivered via elective c-section may lack these exposures. Positive associations between prelabor c-section and risk of childhood acute lymphocytic leukemia have been reported in a pooled analysis of 13 case–control studies (15), a large population-based case–control linkage study (16), and a population-based cohort study (17). Other birth characteristics, such as birth order, may also influence the immune system. A higher birth order has been considered a marker of early exposure to pathogens, as older siblings may expose younger siblings to viruses and bacteria from outside the home through play and close contact (15, 16). The role of birth order has been explored (15–25), with most studies reporting a positive association between increasing birth order and adult NHL risk, especially for DLBCL (18, 19, 21, 22, 25, 26).

Leveraging a large population-based case–control linkage study in California, we evaluated whether mode of delivery and birth characteristics play a role in the etiology of pediatric and AYA NHL and the three most common histologic subtypes. The linkage design minimizes potential for selection bias and recall bias.

Collaborating with the California Department of Public Health, investigators from University of California, Berkeley and Yale University established the California Linkage Study of Early-onset Cancers, a population-based case–control study. Cases and controls were born in California during 1978 to 2015. A probabilistic record linkage was conducted between California birth records from 1978 to 2015 and cancer registry data from 1988 to 2015. Birth records are routinely collected for all California births by the Center for Health Statistics and Informatics, California Department of Public Health. California Cancer Registry routinely collects data on all cancer diagnoses statewide. The collection of both types of records is mandated by law and is expected to have complete or near complete coverage.

NHL cases (n = 3,101) were diagnosed at ages 0 to 37 years and identified based on International Classification of Childhood Cancer codes or SEER site recodes (27, 28). Cases were excluded if their mother resided outside California (n = 1) or if data on maternal age (n = 1), birthweight (n = 3), birth order (n = 26), maternal nativity (i.e., born in or outside the United States; n = 2), mode of delivery (n = 3), or congenital anomalies (n = 1) were missing. For each case, fifty controls were selected from statewide birth records based on the same eligibility criteria. Controls were frequency-matched to cases on year of birth and had no history of cancer based on California Cancer Registry records at the time of linkage. We chose a control–to–case ratio of 50:1 to improve statistical power but also to balance the need to preserve confidentiality of birth data.

The final dataset included 3,064 NHL cases and 153,200 controls. The study protocol was approved by the Institutional Review Boards at the California Health and Human Services Agency, University of California, Berkeley, and Yale University.

Mode of delivery and birth order were the two main variables of interest, though other birth characteristics including birthweight, gestational age, birth plurality, congenital abnormalities (yes/no), complications during pregnancy (yes/no), maternal history of miscarriage or stillbirth (yes/no), or previous maternal c-section (yes/no) were also obtained from birth records. We also retrieved demographic and parental characteristics including sex, race/ethnicity, maternal age at delivery, maternal nativity, maternal education, paternal age at delivery, and paternal education.

Additionally, we were interested in potential etiologic differences across ages of diagnosis (pediatric: 0–14 years, AYA: 15–37 years) and the most common subtypes of early-onset NHL (Burkitt lymphoma, DLBCL, and T-cell NHL), which were identified by the International Classification of Diseases for Oncology (3^rd Edition) histology codes (29). The codes were 9687 and 9826 for Burkitt lymphoma, 9680, 9688, 9737–9738, 9684 (grade 6 only), 9735 (grade 6 only), 9680, 9688, 9737–9738, 9712, 9678, 9679, 9687, and 9826 for DLBCL, and 9591, 9675, 9684, 9700, 9701, 9702, 9705, 9708, 9709,9714, 9716, 9717, 9718, 9719, 9726, 9727, 9729, 9735, 9827, 9831, 9832, 9834, 9835, 9837, and 9948 for T-cell NHL. We did not evaluate other NHL subtypes separately due to their lower frequencies.

The characteristics of cases and controls were described with frequencies and percentages for categorical variables, as well as means and SDs for continuous variables. Odds ratios (OR) estimates and 95% confidence intervals (CI) were derived from unconditional multivariable logistic regression models that included year of birth, sex, race/ethnicity (non-Hispanic White, Black, Hispanic, Asian/Pacific Islander, other), mode of delivery (vaginal vs. cesarean), birth order (first, second, third, or higher), plurality (single vs. multiple), congenital abnormalities (yes vs. no), pregnancy complications (yes vs. no), maternal history of miscarriage or stillbirth (yes vs. no), maternal history of cesarean delivery (yes vs. no), maternal age at delivery (≤24, 25–29, 30–34, ≥35 years), maternal education (less than high school, high school, beyond high school, unknown), maternal nativity (whether mother was born in the United States or other countries), birthweight (kg), and gestational age (weeks). The same variables were included in the multivariable logistic regression model for both the primary analysis and stratified analyses by age of diagnosis and NHL subtype.

We also evaluated potential interaction between mode of delivery and birth order. Additionally, several sensitivity analyses were also performed including an analysis excluding all multiple births and an analysis evaluating differences between emergency and elective c-section. Information on the type of c-section was not available for individuals born during 1989 to 2004, so fewer subjects were included in this analysis (1,766 cases and 88,300 controls).

All analyses were performed using SAS (version 9.4, SAS Institute) and all tests were two-sided with an α value of 0.05.

The data used in this analysis were obtained from the California Department of Public Health with specific approvals. We are prohibited by the California Department of Public Health from sharing the data with others due to confidentiality concerns.

Of 3,064 NHL cases (pediatric: 1,317; AYA: 1,747), 981 (32%) were DLBCL, 490 (16%) were Burkitt lymphoma, 340 (11%) were peripheral T-cell lymphoma, 257 (8.3%) were precursor T-cell NHL, and 152 (4.9%) were Mycosis fungoides/Sezary syndrome. Other NHL subtypes were less frequent and were not listed separately. Cases and controls were similar with regard to year of birth (i.e., the matching variable; Table 1).

In multivariable analysis, those born via c-section had a decreased risk of early-onset NHL compared with those born vaginally (OR = 0.88, 95% CI: 0.79–0.97; Table 2). Individuals with a birth order of third or higher also had a decreased risk of early-onset NHL compared with first-borns (OR = 0.89, 95% CI: 0.80–0.98). In addition, males had an increased risk compared with females (OR = 1.83, 95% CI: 1.69–1.98). NHL risk also varied by race/ethnicity, with Blacks and Asians/Pacific Islanders having a 19% and 22% increased risk compared with Whites, respectively. Children from multiple births had 1.46 times the odds (95% CI: 1.14–1.87) of early-onset NHL compared with those from singleton births. Compared with those born to mothers aged 25 to 29 years, those born to mothers aged ≤ 24 years had a decreased risk of early-onset NHL (OR = 0.81, 95% CI: 0.73–0.89). Furthermore, a lower maternal education level was associated with a lower risk of early-onset NHL – the OR for less than high school compared with high school was 0.75 (95% CI: 0.65–0.87). Additionally, birthweight was associated with NHL risk; the OR for each kilogram increase in birthweight was 1.14 (95% CI: 1.05–1.23; Table 2).

In stratified analyses by age of diagnosis and NHL subtype, we observed that males had a higher risk than females in all 6 subgroups, although the magnitude of association varied (Table 3). Other characteristics exhibited heterogeneity across the subgroups. For pediatric Burkitt lymphoma, Black and Hispanic children had a lower risk than White children, and delivery via c-section was associated with a reduced risk compared with vaginal delivery (OR = 0.71, 95% CI: 0.51–0.99). For pediatric DLBCL, children from multiple births had a higher risk than singleton births (OR = 2.61, 95% CI: 1.26–5.39). On the other hand, the risk of pediatric T-cell NHL appeared lower in children who were delivered via c-section (compared with vaginal delivery, OR = 0.73, 95% CI: 0.55–0.97), who had a birth order of second (OR = 0.73, 95% CI: 0.57–0.93), or third, or higher (OR = 0.76, 95% CI: 0.58–0.99) compared with first-borns, and who were born to younger mothers or mothers who had less education.

For AYA Burkitt lymphoma, a maternal age of ≤ 24 years was associated with a lower risk compared with maternal age of 25 to 29 years (OR = 0.76, 95% CI: 0.59–0.99). For AYA DLBCL, Black individuals had a lower risk than their White counterpart. Compared with individuals born to mothers who were aged 25 to 29 years, those born to younger mothers had lower risk (≤24 years: OR = 0.79, 95% CI: 0.65–0.96), and those born to older mothers had higher risk (30–34 years: OR = 1.24, 95% CI: 1.01–1.52; ≥35 years: OR = 1.33, 95% CI: 1.02–1.74). In addition, a lower maternal education level was associated with a lower risk of AYA DLBCL (less than high school compared to high school: OR = 0.69, 95% CI: 0.49–0.96). A heavier birthweight was also associated with an elevated risk of AYA DLBCL, with an OR of 1.16 (95% CI: 1.00–1.35) for each kg of increase in birthweight. AYA T-cell NHL was more common in Asians/Pacific Islanders than in Whites, those born as multiple births compared with singleton births (OR = 1.92, 95% CI: 1.03–3.58), and those born to older mothers (≥ 35 years compared with 25–29 years: OR = 1.55, 95% CI: 1.10–2.18). Furthermore, a lower maternal education level was associated with a lower risk of AYA T-cell NHL (less than high school compared with high school: OR = 0.56, 95% CI: 0.36–0.89; Table 3).

We did not find evidence of interaction between mode of delivery and birth order. The analyses excluding all multiple births yielded results that were nearly identical to the analyses including all subjects. Additionally, when evaluating c-section type, the ORs for emergency and elective c-section compared with vaginal birth were in the same direction and similar in magnitude for all analyses.

In this large population-based case–control study, individuals born via c-section had a decreased risk of early-onset NHL compared with individuals born vaginally, and those with a birth order of third or higher had a decreased risk of early-onset NHL compared with first-borns.

Mode of delivery may be associated with NHL through a biological mechanism in which the microbiome composition at birth affects immune system development. The microbiome of infants born vaginally differs from those born via c-section (9) and, in early life, the microbiome may influence immune system development and subsequent risk of immuno-modulated diseases (10–12). Recently, other studies have found positive associations between prelabor c-section and childhood acute lymphoblastic leukemia (30–32), suggesting that the microbiome may play a role in the etiology of these cancers. Other studies have found lower levels of cortisol and catecholamine in neonates delivered by c-section than in neonates who underwent vaginal delivery (33–36). There is some indication that cortisol may eliminate preleukemic clones arising in utero (37), but the relevance to NHL is unclear.

In our study, c-section was associated with a decreased risk of early-onset NHL overall, pediatric Burkitt lymphoma, and pediatric T-cell NHL. Mode of delivery has not been explored extensively in the context of NHL as we identified only two published studies. A population-based case–control study in Sweden reported that children born via c-section had 1.6 times the odds of NHL at the age of 0 to 14 years (95% CI = 1.0 to 2.6) compared with those born vaginally after adjusting for matching variables age and sex (38). A population-based follow-up study in Denmark, Sweden, and Finland reported no significant association between risk of childhood NHL (0–14 years) and mode of delivery (HR cesarean vs. vaginal = 1.21, 95% CI = 0.96 to 1.52) after adjusting for birth year, country, gestational age, maternal age, parity, and plurality (39). These results are inconsistent with our finding that c-section is associated with a decreased risk of early-onset NHL compared with vaginal birth, however, these discrepancies may be attributable to different covariate adjustment, smaller sample sizes [168 cases (38), 636 cases (39), vs. >3,000 cases in our study], and low proportions of c-section [9.5% among controls (38), 12.6% in the cohort (39), vs. 21.7% in our controls]. The less common use of c-section in those European countries may reflect different attitudes about this practice and possibly different clinical indications.

Birth order is a marker of early-life microbial exposure (15, 16), which likely impacts immune system development and function (11, 40). Lack of infections in early life increases the risk of allergic and autoimmune diseases (40) and may also play a role in the etiology of certain cancers. For example, Greave's ‘delayed infection’ hypothesis for pediatric acute lymphocytic leukemia proposes that following a lack of infections in early life, an inappropriate response to a ‘delayed’ infection along with other predispositions promote the development of the disease (41). Birth order may be associated with risk of NHL and other cancers of the hematopoietic system through a similar mechanism.

A birth order of third or higher was associated with a lower risk of early-onset NHL overall and pediatric T-cell NHL in our study. Other studies investigating birth order and risk of NHL have reported varying results. Our results are consistent with those from a case–control study investigating risk of NHL in adults in Europe (15) and a Swedish National Cohort Study investigating risk of NHL in individuals aged 0 to 37 years old (20). These studies found an elevated risk of NHL among first-borns and only children (15) and an increased risk of NHL with low birth order (20). Other studies are suggestive of lower odds of NHL with higher birth order (16, 24), however the CIs include one, perhaps due to smaller sample sizes (592 cases and 164 cases, respectively). Other studies have reported no association between birth order and NHL, with no apparent trend (17, 23, 42), or findings contrary to ours (18, 19, 22, 25, 26). A meta-analysis of 18 InterLymph studies also found a nonsignificant trend (P = 0.082) towards increasing birth order and increased risk of NHL and noted that this trend was statistically significant for some NHL subtypes (B-cell, follicular, precursor B-cell, T-cell, and mycosis fungoides; ref. 21). The authors also acknowledged high heterogeneity of results across included studies (21) and suggested that selection bias by socioeconomic background, discussed previously by Mensah and colleagues (23), might play a role. Specifically, individuals from a higher socioeconomic background tend to have fewer children and are more likely to participate in research studies, particularly among the control group (21, 23). Since our study was based on birth records and there was no potential for differential response/participation across socioeconomic status, our findings are not prone to selection bias. In addition, our study focused on early-onset NHL, while most of the previous studies such as the InterLymph studies included only adult cases, most of whom were older than 37 years at the time of NHL diagnosis. Therefore, it might not be meaningful to compare our results with those of the previous studies.

Studies from Israel and Sweden have found that hospitalization due to infections in infancy is associated with a higher risk of NHL among children and young adults (43, 44), which support the hypothesis that early exposure to microbes increases NHL risk and would be consistent with our c-section finding, as c-section often leads to delayed exposure to microbes. However, the same hypothesis would not explain our finding that higher birth order is associated with a lower risk of NHL. Existing studies include ours cannot yet conclude whether early exposure to microbes increases or decreases NHL risk. A major challenge lies in the fact that existing studies have only been able to use proxy measures of such an exposure, including hospitalization due to infections, mode of delivery, and birth order. More studies are needed to directly examine exposure to microbes in relation to NHL risk.

Our study had several limitations. Our dataset did not provide information regarding immunosuppression, prior infections, or family history of NHL, which are believed to play a role in the etiology of NHL (2, 5). Other possible risk factors that we were unable to evaluate include maternal smoking, alcohol intake, environmental and occupational exposure to toxins, type of anesthesia for delivery (if any), medication use during pregnancy, and history of infertility. While the overall sample size was large, the numbers of cases for some subgroup analyses were relatively small, reducing the statistical power. Information on the specific type of c-section was only available for 58% of subjects. Furthermore, we had no data on sibling's year of birth, which may be important to consider for the analysis of birth order. Additionally, there was potential for misclassification of controls who might have moved out of California and been diagnosed with early-onset NHL elsewhere. Based on age-specific incidence rates of NHL in California from 2000 to 2016, we estimate that fewer than 34 cases of NHL would have developed among the 153,200 controls by the age of 37 years if they had all moved out of California the day after birth (45, 46). Of course, this would have been the most extreme scenario. Given the large number of cases included in our study (n = 3,064), a possible misclassification of up to 34 controls would not have a substantial impact on our findings. Lastly, the linkage started with the birth year of 1978 (when California birth records first became computerized) and cancer diagnosis from 1988 (when the California Cancer Registry began statewide ascertainment of cancer incidence), and we could only retrieve birth data for individuals born in California. A comparison with records from the California Cancer Registry suggested that our study captured 56.4% of all NHL cases diagnosed at 0 to 37 years in California (75.5% in the age group of 0–14 years, 50.2% in the age group of 15–37 years). However, there is no reason to believe that cases born outside California have a systematically different etiology.

On the other hand, this study has multiple methodologic strengths. The linkage design minimized the potential for selection bias and recall bias. The study included the largest number of cases with early-onset NHL to date, allowing examination of potential etiologic differences across age of diagnosis and NHL subtype. We also adjusted for a number of covariates available from birth records, reducing possible confounding.

In summary, we found that cesarean delivery and a higher birth order were associated with a lower risk of early-onset NHL in this large population-based case–control study, although the associations exhibited variations across different demographic and histologic subgroups. Given the heterogeneity of NHL, future studies should stratify by age of diagnosis and NHL subtype. Mechanistic studies evaluating the role of early-life immune development in lymphomagenesis are also needed.

No disclosures were reported.

The ideas and opinions expressed herein are those of the author(s), and endorsement by the State of California, Department of Public Health, the NCI, and the Centers for Disease Control and Prevention or their Contractors and Subcontractors is not intended nor should be inferred.

K.E. Dwyer: Formal analysis, investigation, methodology, writing–original draft. R. Wang: Formal analysis, investigation, methodology, writing–review and editing. W. Cozen: Investigation, methodology, writing–review and editing. B. Cartmel: Investigation, writing–review and editing. J.L. Wiemels: Investigation, writing–review and editing. L.M. Morimoto: Investigation, writing–review and editing. C. Metayer: Data curation, investigation, writing–review and editing. X. Ma: Conceptualization, data curation, supervision, investigation, methodology, writing–review and editing.

The collection of cancer incidence data used in this study was supported by the California Department of Public Health as part of the statewide cancer reporting program mandated by California Health and Safety Code Section 103885; the National Cancer Institute's Surveillance, Epidemiology and End Results Program under contract HHSN261201000140C awarded to the Cancer Prevention Institute of California, contract HHSN261201000035C awarded to the University of Southern California, and contract HHSN261201000034C awarded to the Public Health Institute; and the Centers for Disease Control and Prevention's National Program of Cancer Registries, under agreement U58DP003862-01 awarded to the California Department of Public Health.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.