Association Between Red-Cell Transfusion in Childbirth and Long-Term Risk of Lymphoma and Autoimmune Disease: A Swedish Nationwide Cohort Study

Abstract

Red-cell transfusions during childbirth are essential for managing significant blood loss, but may have long-term immunological implications. While immediate risks like transfusion-transmitted infections are well-documented, less understood are the potential associated risks of future lymphoma and autoimmune disease [1]. We performed a study that aimed to assess the long-term risk of developing non-Hodgkin lymphoma (NHL), rheumatoid arthritis (RA), systemic lupus erythematosus (SLE), and systemic sclerosis, among women who received red-cell transfusions during childbirth.

A nationwide cohort study was conducted using data from SCANDAT3-S, the Swedish portion of the Scandinavian Donations and Transfusions database, covering blood donations and transfusions from 1968 to 2017. The database is linked to national health registers, including the National Patient Registers, National Medical Birth Register, National Cancer Register, and the National Prescribed Drug Register, providing comprehensive delivery information and health outcomes [2].

The study included women aged 18–50 years who had their first childbirth between January 1, 1987, and December 31, 2017. Women with prior diagnoses of autoimmune diseases, lymphoma, or those who had received blood transfusions before their first childbirth were excluded. The cohort comprised 1 043 713 women with 1 999 013 registered deliveries. Of these women, 42 333 (4.1%) received one or more red-cell transfusions during one or several deliveries.

The primary exposure was the receipt of at least one red-cell transfusion in the period from 3 days before delivery until 7 days postpartum. Exposure status was updated at each delivery; women who received transfusions during any delivery were categorized as “transfused” from that point onwards. The main outcomes were the first recorded diagnosis of NHL, SLE, systemic sclerosis, or RA, identified through the Swedish National Patient and Cancer Registers using ICD codes, which are available in Appendix A.

Cox proportional hazards models were used to estimate hazard ratios (HRs) for each outcome, comparing red-cell transfused women around delivery to those who were never transfused. We used “time since the most recent delivery” as the time scale, and follow-up began 6 months postpartum to minimize reverse causation. This choice allowed us to align the follow-up period with the postpartum interval for each childbirth. If a woman had more than one delivery during the study period, the time scale was reset at each delivery, and exposure status was updated accordingly.

Models were adjusted for potential confounders, including maternal age (years), body mass index (BMI), parity, mode of delivery, preeclampsia, socioeconomic factors (income, education), smoking status, ABO blood group, and birth origin. Follow-up continued until an outcome occurred, at which point a woman diagnosed with the outcome no longer contributed further time to the risk set. Censoring occurred at date of emigration, HIV diagnosis, pregnancy-related endocrine disorders, study end, death, or receipt of other blood products or blood product transfusion outside of obstetric care. Follow-up was also censored at the time of IVF initiation. IVF was censored due to the possible confounding by hormonal and immunologic changes associated with assisted reproduction, and because women pregnant by IVF may have a different risk of being transfused.

Sensitivity analyses included stratifying by the number of transfused units (1–2 units vs. ≥ 3 units), calendar periods (1987–1996, 1997–2006, and 2007–2017), varying the start of follow-up postpartum (no delay, 1 year, and 2 years), mode of delivery, and presence of preeclampsia.

At baseline, transfused women were slightly older and had a higher BMI compared to non-transfused women. The median age at first delivery was higher in the transfused group, and a greater proportion had a BMI of 25 or higher. Preeclampsia and cesarean sections were more common among transfused women. Cohort characteristics are summarized in Table S1.

During follow-up, transfused women had a higher incidence of SLE (10.2 vs. 7.0 per 100 000 person-years) and systemic sclerosis (3.6 vs. 1.8 per 100 000 person-years) compared to non-transfused women. The adjusted hazard ratios (aHR) for SLE was 1.38 (95% CI, 1.01–1.87), indicating increased risk among transfused women also after adjusting for confounders.

For systemic sclerosis, the corresponding aHR was 1.89 (95% CI, 1.21–3.21). For NHL, incidence rates were similar between transfused and non-transfused women (4.0 vs. 4.6 per 100 000 person-years), with an aHR of 0.86 (95% CI, 0.53–1.40). Rheumatoid arthritis incidence rates were also similar between the groups (39.5 vs. 44.0 per 100 000 person-years), with an aHR of 1.07 (95% CI, 0.92–1.24), suggesting no significant increased risk associated with transfusion for neither condition. Results are summarized in Table 1.

TABLE 1. Incidence rates, events per person-years, unadjusted and adjusted cause-specific hazard ratios for diagnosis of NHL, SLE, Systemic sclerosis or RA in relation to red-cell exposure in delivery using a 6 months' latency in follow-up postpartum.

	Incidence rate /100 000 person-years	Events/person-years	Unadjusted HR (95% CI)	Adjusted HRa (95% CI)
Diagnosis groups
Non-Hodgkin lymphoma
Transfusion in delivery	4	17/420460	0.93 (0.57–1.50)	0.86 (0.53–1.40)
No transfusion in delivery	4.6	540/11805084	1.00 (ref)	1.00 (ref)
Systemic lupus erythematosus
Transfusion in delivery	10.2	43/420460	1.46 (1.07–1.98)	1.38 (1.01–1.87)
No transfusion in delivery	7	824/11805084	1.00 (ref)	1.00 (ref)
Systemic sclerosis
Transfusion in delivery	3.6	15/420460	2.05 (1.21–3.45)	1.89 (1.12–3.21)
No transfusion in delivery	1.8	208/11805084	1.00 (ref)	1.00 (ref)
Rheumatoid arthritis
Transfusion in delivery	44	185/420460	1.13 (0.98–1.31)	1.07 (0.92–1.24)
No transfusion in delivery	39.5	4666/11805084	1.00 (ref)	1.00 (ref)

• ^a Adjusted for maternal age at delivery, year of delivery, parity, blood group, birth origin, early pregnancy BMI, smoking, smoking, snuff use, income, level of education, hospital level, mode of delivery and preeclampsia in any delivery.

Women who received a higher number of transfused units had a non-significant trend toward increased risk for SLE and systemic sclerosis. For those who received 1–2 units, the aHR for SLE was 1.25 (95% CI, 0.85–1.84), and for those who received ≥ 3 units, the aHR was 1.66 (95% CI, 1.01–2.73). Similarly, for systemic sclerosis, the aHRs were 1.83 (95% CI, 0.97–3.47) for 1–2 units and 2.03 (95% CI, 0.84–4.92) for ≥ 3 units. For NHL or RA, we did not observe such a pattern. Stratifying by calendar periods did not show significant changes over time, indicating that the association between transfusion and increased risk of autoimmune diseases remained consistent despite changes in transfusion practices, such as the gradual implementation of leukodepletion from the mid-1990s. Varying the delay in the start of follow-up postpartum did not significantly affect the results for NHL, SLE, and RA. However, for systemic sclerosis, the association was stronger when follow-up began immediately after childbirth (aHR 1.98; 95% CI, 1.19–3.30) compared to starting at 1 or 2 years postpartum. Stratifying by caesarian section or preeclampsia did not significantly alter the findings. Results of sensitivity analyses are presented in Figure S1.

No association was found between transfusion during childbirth and all-cause mortality. The aHR for death was 0.99 (95% CI, 0.80–1.21), indicating that transfusion did not affect overall survival during the follow-up period.

In this large, nationwide cohort study with long-term follow-up, red-cell transfusions during childbirth was associated with a modestly higher risk of developing SLE and systemic sclerosis, but not of NHL and RA. These associations persisted after adjusting for various potential confounders and were consistent across sensitivity analyses.

Pregnancy itself induces a chimeric state due to the exchange of cells between the mother and fetus. The additional exposure to foreign cells through transfusion may potentially trigger autoimmune responses in genetically or immunologically susceptible individuals, due to the immunomodulatory effects of transfusions and the introduction of donor cells leading to microchimerism.

The association with SLE aligns with a smaller Swedish case–control study reported a non-significant odds ratio of 1.8 for the association between red-cell transfusion and SLE [3]. We observed a similar association in our study, with a statistically significant hazard ratio, which reinforces the potential link.

The connection with systemic sclerosis is also supported by evidence implicating microchimerism in its pathogenesis. Studies have detected fetal DNA in skin lesions and peripheral blood of women with systemic sclerosis [4], and it is hypothesized that persistent foreign cells may contribute to disease development. The stronger association observed when follow-up began immediately after childbirth may indicate an element of reverse causation or surveillance bias, but one may also consider that transfusion might accelerate the onset of systemic sclerosis in susceptible individuals.

The lack of association with NHL is persistent in our analyses, and although there are few cases which limits the power of our study, it has the strength of not being as vulnerable to detection bias as other studies in the field. For RA, previous research has resulted in mixed findings [5, 6]. Our results suggest that transfusion during childbirth does not significantly impact RA risk in this population.

Despite the study's strengths, including its large size, long-term follow-up, and comprehensive data sources, there are limitations. Residual confounding may still exist due to unmeasured factors such as genetic predisposition or environmental exposures. However, as such factors would need to be associated with both transfusion at delivery and with the risk of the outcomes considered here to confound the results, we believe it is unlikely that they have contributed significantly to our findings. Still, although we adjusted for numerous potential confounders, we cannot exclude the possibility that early, undiagnosed symptoms of autoimmune diseases, relating for example to coagulopathy or propensity for tissue injury and hemorrhage, could have led to increased transfusion rates due to complications during childbirth. To mitigate this problem, we starting follow-up 6 months postpartum in the main analyses and performed sensitivity analyses with progressively longer delay of start of follow-up with largely unchanged results. Still, reverse causation (i.e., protopathic bias) may have influenced the findings. Similarly, outcome misclassification is possible, as some autoimmune diseases may have been underdiagnosed or misclassified, especially in the earlier years of the study. However, such misclassification is likely non-differential with regards to blood transfusion at delivery and would—if anything—tend to bias results toward the null rather than produce spurious associations.

Lastly, our findings may not be generalizable to populations outside Sweden or to men, as the study focused exclusively on women who gave birth in Sweden. Differences in transfusion practices and environmental exposures could affect the applicability of the results to other settings.

Red-cell transfusions during childbirth may be associated with a modestly increased long-term risk of developing SLE and systemic sclerosis. Although the observed risk increase was small, these findings further highlight the importance of cautious use of transfusions and consideration of alternative treatments when appropriate. Clinicians should weigh the immediate benefits of transfusion against potential possible long-term risks, especially in cases where transfusion may not be strictly necessary. Further research is needed to understand the mechanisms underlying these associations.