Abstract
Monoclonal antibodies directed against tumor necrosis factor alpha (TNFα) are currently employed in the treatment of various immune-mediated diseases. These studies were designed to evaluate potential effects of anti-TNFαtreatment in mice during pregnancy and lactation on the development of the immune system in the F1 generation. Pregnant CD-1 mice were treated with vehicle or with 10 or 40 mg/kg of an anti-mouse TNFαmonoclonal antibody (mAb) (cV1q) on days 6, 12, and 18 of gestation and on days 3, 9, and 15 of lactation. Evaluation of immune system functionality was conducted in F1 generation mice at 11 weeks of age. Immune function was evaluated by splenocyte phenotyping, immunoglobulin M (IgM) antibody response to sheep red blood cells (SRBCs), spleen cell proliferative response to anti-CD3, and natural killer cell activity. Treatment of pregnant mice with cV1q produced no adverse effects in the dams and no adverse effects in the F1 generation. In general, the functioning of the immune system of the F1 generation did not appear to be adversely affected following exposure to cV1q in utero and during lactation. The only statistically significant change was a slight (~20%) reduction in the spleen cell expansion in response to SRBC immunization in the female F1 mice from the 40 mg/kg cV1q treatment group. In conclusion, administration of a monoclonal antibody against mouse TNFαduring pregnancy and lactation had little or no effect on selected immune parameters in mice, with only a possible minor attenuation of spleen cell response to immunization noted in the female F1 generation at 11 weeks of age.
Tumor necrosis factor alpha (TNFα) is a proinflammatory cytokine that plays a central role in the pathogenesis of autoimmunity. Inhibitors of TNFα, including infliximab (Remicade), etanercept (Enbrel), and adalimumab (Humira), have proven to be effective treatments for various immune-mediated diseases and the risk-benefit profile of these agents has been acceptable. However, treatment with anti-TNFα therapies, especially in combination with immunosuppressive agents, has been associated with an increased incidence in infections, including some serious infections (Hochberg et al. 2005).
Because anti-TNFα therapies are indicated for the treatment of immune-mediated diseases that affect women of child-bearing potential, an understanding of their effects on reproduction and development is necessary before they can be considered for use during pregnancy. Infliximab is a chimeric mouse-human monoclonal antibody that binds to both soluble and membrane-bound human TNFα. Infliximab binds only to human and chimpanzee TNFα and therefore reproductive toxicity testing could not be conducted with infliximab. In order to evaluate the effects of inhibition of TNFα during pregnancy, a rat anti-mouse TNFα surrogate monoclonal antibody (cV1q) was developed. Embryofetal development studies conducted in mice with cV1q showed no abortifacient or teratogenic effects following treatment of dams during pregnancy (Treacy 2000). Abortifacient and teratogenic effects were also not seen in rabbits treated with etanercept during pregnancy or in macaques treated with adalimumab during pregnancy. (The Food and Drug Administration [FDA] summary basis of approval information available at www.drugs@FDA.gov.) A review of data from pregnant women receiving infliximab for the treatment of Crohn’s disease or rheumatoid arthritis showed pregnancy outcomes that did not differ than those of the general disease population (Katz et al. 2004; Mahadevan et al. 2005). Overall, these studies show that maternal exposure to anti-TNFα therapies during pregnancy does not result in any gross developmental abnormalities in the offspring. However, the effects of anti-TNFα treatment during pregnancy and lactation on the development and maturation of the immune system have not yet been fully evaluated.
Mice that are genetically modified to lack TNFα are fertile and also show no gross abnormalities (Pasparakis et al. 1996). However, the genetically TNFα-deficient mice do show an increased susceptibility to infection and reduced delayed-type hypersensitivity (DTH) and humoral immune responses. These TNFα knockout mice lack splenic primary B-cell follicles and cannot form organized follicular dendritic cell networks and germinal centers (Korner et al. 1997). Based upon this information, it might be predicted that anti-TNFα therapies could cause impairment in the development and maturation of the immune system in the F1 generation.
The current studies were designed to evaluate the potential effects on the developing immune system following maternal exposure to an anti-TNFα monoclonal antibody during pregnancy and lactation.
MATERIALS AND METHODS
All in-life procedures were conducted in compliance with the Animal Welfare Act, the Guide for the Care and Use of Laboratory Animals, and the Office of Laboratory Animal Welfare. Study protocols were reviewed and approved by the test site Institutional Animal Care and Use Committee.
Anti-Mouse TNFαMonoclonal Antibody (cV1q)
A chimeric anti-mouse TNFα monoclonal antibody (cV1q; Scallon et al. 2004) was used for these studies. cV1q was engineered to have mouse immunoglobulin (Ig) G2a and kappa constant regions.
Animals and In-Life Evaluations
Crl:CD-1 (ICR) BR male and female mice, obtained from Charles River Laboratories (Wilmington, MA) were used in the studies.
Female mice (10 to 25 per treatment group) were cohabited with male mice for a maximum of 5 days. Female mice with a copulatory plug, observed in situ, or sperm in a vaginal lavage were considered to be at gestation day 0 and were assigned to treatment groups. cV1q (10 or 40 mg/kg), vehicle control, or antibody control was administered by intravenous injection on days 6, 12, and 18 of presumed gestation (GDs 6, 12, and 18) and on days 3, 9, and 15 of lactation (LDs 3, 9, and 15). The day of birth was defined as day 1 of lactation (LD 1).
In a subset of mice, maternal blood was collected on LD 15 and pup blood was collected on LDs 2 and 15 for measurement of cV1q concentrations.
Immunological Evaluations
Immunological evaluations were conducted in selected F1 generation mice (9 to 10 per gender per treatment group per assay) at 11 weeks of age. The details of these assays have been described elsewhere and are described briefly below (Luster et al. 1992; Guo et al. 2001).
Splenocyte Phenotyping
Spleen cell suspensions from 9 to 10 11-week-old F1 generation mice per treatment group were prepared and evaluated for lymphocyte subsets by flow cytometry analysis. B cells, T cells, and natural killer (NK) cells were labeled using commercially available anti-mouse antibodies (BD PharMingen, San Diego, CA). Total B cells were enumerated using antibodies directed against Ig-positive cells (Ig+) and for total T cells an antibody to the CD3 epitope (CD3+) was used. T subsets were evaluated using antibodies to CD4 (CD4+CD8−) for T helper cells and CD8 (CD4−CD8+) for cytotoxic T cells. NK cells were enumerated as cells that were NK1.1 positive CD3 negative (NK1.1+CD3−). A subset of the vehicle-control F1 generation mice were treated with cyclophosphamide (50 mg/kg, intraperitoneal [i.p.] for 4 consecutive days prior to sacrifice) as a positive control for immunotoxicity.
Spleen Cell Proliferative Response to Anti-CD3
Untreated splenocytes or anti-CD3 treated splenocytes from 9 to 10 11-week-old F1 generation mice per treatment group were incubated at 36°C to 38°C at 5% to 7% CO2 for 3 days. After 18 to 24 h of incubation, cells were pulsed with 3H-thymidine and thymidine incorporation was measured as a measure of cellular proliferation. A subset of the vehicle-control mice was treated with cyclophosphamide as a positive control.
Natural Killer Cell Activity
Spleen cells from 9 to 10 11-week-old F1 generation mice in the pretreatment group were adjusted to six concentrations to obtain effector-to-target ratios of 200:1, 100:1, 50:1, 25:1, 12.5:1, and 6.25:1. The splenic NK cell was the effector cell and a mouse lymphoma cell (YAC-1; American Type Culture Collection, ATCC TIB 160) was the target cell. YAC-1 cells were adjusted to a concentration of 107 cells/ml and were incubated with 500 μCi of 51Cr for approximately 90 min at 36°C to 38°C. Cytotoxicity of YAC-1 cells was measured as release of 51Cr. For this assay a subset of control F1 mice was treated with anti-asialo GM1 (AAGM1) (1:10 dilution, 0.2 ml, intravenous [i.v.], 24 h prior to sacrifice) as a positive control for NK-cell inhibitory activity.
Spleen IgM Antibody Response to Sheep Red Blood Cells
Eleven-week-old F1 generation mice (9 to 10 per treatment group) were sensitized with 1 × 108sheep red blood cells (SR-BCs) intravenously 4 days prior to termination. Spleens were removed and splenocytes isolated. Spleen cells were mixed with guinea pig complement, SRBCs, and agar and plated onto Petri dishes. Following 3 h of incubation at 36°C to 38°C, the number of plaques, representing a single IgM antibody-producing B cell, were counted.
A subset of the 11-week-old F1 generation mice from the vehicle-control treatment group was treated with cyclophosphamide as a positive control.
Because of a potential treatment-related effect observed in the SRBC IgM response in the 11-week-old F1 generation female mice, a second confirmatory study was conducted. In the second study, pregnant mice were administered vehicle control, an irrelevant antibody control 40 mg/kg, or cV1q 40 mg/kg intravenously during pregnancy on gestation days 6, 12, and 18 and on lactation days 3, 9, and 15. F1 generation mice were evaluated at 11 or 22 weeks of age for IgM response to SRBCs.
Serum and Milk cV1q Concentrations
Serum and milk levels of cV1q were measured by a quantitative sandwich enzyme immunoassay (ELISA) using two non-competing anti-idiotypic monoclonal antibodies (one for capture and one for detection) that recognize the variable portion of cV1q. The capture antibody was immobilized on a microplate. Standards and serum samples are added to the microplate and the immobilized capture antibody bound any cV1q present in the serum samples. Bound cV1q was measured using the biotinylated detection antibody followed by streptavidin–horseradish peroxidase. The addition of tetramethylbenzidine and sulfuric acid resulted in color development in proportion to the amount of cV1q bound in the initial step.
Statistics
Data were first tested for homogeneity of variances using the Bartlett’s chi-square test. Homogeneous data were evaluated by a parametric one-way analysis of variance. When significant differences occurred, treatment groups were compared to the vehicle-control group using Dunnett’s t test. Nonhomogenous data were evaluated using a nonparametric analysis of variance. When significant differences occurred, treatment groups were compared to vehicle control using the Gehan-Wilcoxon test.
For the evaluation of the effects of cV1q treatment on SRBC immunization parameters, data from the two independent studies were analyzed individually as described above and were also analyzed as a pooled data set using a linear mixed-effects model that accounted for any between study variability.
RESULTS
Splenocyte Phenotyping
The results from the splenocyte phenotyping are summarized in Table 1, where absolute cell numbers are presented. Treatment of 11-week-old control F1 generation mice with cylophosphamide produced a significant reduction in the number of spleen cells, primarily due to reductions in Ig+ cells (B cells) and NK cells. A trend towards a reduction in T, B, or NK cells was evident in males and to a lesser extent in females. However, these apparent changes did not attain statistical significance and individual values were within normal ranges for CD-1 mice.
Spleen Cell Proliferative Response to Anti-CD3
Treatment of splenocytes in vitro from 11-week-old F1 generation mice with anti-CD3 antibody produced a cellular proliferation, as measured by an increase in 3H-thymidine incorporation, in both vehicle-control F1 generation mice and the cV1q-exposed mice (Table 2). There was no significant difference in the magnitude of the proliferative response in the cV1q-exposed F1 mice versus the vehicle-control mice.
Natural Killer Cell Activity
NK cell activity was not inhibited in cV1q-exposed 11-week-old F1 generation mice relative to control mice at any of the target:effector ratios evaluated. Results for the 100:1 ratio are summarized in Table 2. In contrast, treatment of a subset of control 11-week-old F1 mice with AAGM1 produced a profound (80% to 100%) reduction in NK cell activity (data not shown).
Spleen IgM Antibody Response to SRBCs
In the first of the two studies designed to evaluate the effects of cV1q exposure on the splenic IgM antibody response, immunization of 11-week-old mice with SRBCs produced an increase in spleen weight and in the spleen to body weight ratio as compared to nonimmunized mice (Figure 1). The magnitude of the increase in spleen weight was 24% in the males and 27% in the females. In the 11-week-old female F1 mice that were exposed to cV1q and immunized with SRBCs, the increase in spleen weights was significantly lower by 17% and 22% in the 10 and 40 mg/kg treatment groups, respectively, as compared to the vehicle-control group. The magnitude of the decrease in the males (12% to 14%) did not attain statistical significance. Treatment-related differences in spleen weights were not observed in female or male mice that were not immunized with SRBCs. Cyclophosphamide produced a significant reduction in absolute spleen weights in both nonimmunized and immunized mice when compared to control animals and immunization with SRBCs did not induce an increase in spleen weight. Body weights of the F1 generation mice were unaffected by cV1q exposure or by cyclophosphamide treatment.
Because spleen weights and spleen cell numbers were significantly lower in the SRBC-immunized, cV1q-exposed F1 female mice versus the control mice; the IgM antibody response was also significantly reduced when expressed on a per spleen basis (Figure 2). However, when the IgM antibody response was normalized to the number of spleen cells, no significant differences were observed between cV1q and vehicle treatment groups. No significant differences were observed in male mice (Figure 2).
In the second confirmatory study (study 2), no significant cV1q (40 mg/kg) treatment-related effects were seen in spleen weights or spleen cell number (data not shown). IgM responses in male or in female SRBC-immunized mice at either 11 or 22 weeks of age were not affected by cV1q treatment (Figure 2). In this study, an irrelevant antibody–control group was also included. The purpose of this group was to determine whether any potential cV1q treatment effects were a specific effect of anti-TNFα antibody treatment or was a nonspecific effect due to administration of an antibody. There was no statistically significant difference between the IgM antibody responses to SRBC immunization in the irrelevant antibody–control group when compared to the vehicle-control group (data not shown).
When the IgM data from the 11-week-old F1 generation mice were pooled across the two studies the statistical analysis of the pooled set data indicated a significant reduction in the 40 mg/kg females versus control for the end points spleen weight (p = .002), number of spleen cells (p = .004), and IgM per spleen (p = .008). The reductions were 19%, 22%, and 37%, respectively. A significant reduction in the number of spleen cells relative to the vehicle control was also found for the females treated with 10 mg/kg cV1q (p = .041). There were no significant differences in body weight.
Serum and Milk cV1q Concentrations
The results from the analysis of pup and dam sera and breast milk for cV1q concentration are summarized in Table 3. In the pups from dams that were treated with cV1q during pregnancy and lactation, high serum concentrations were detected during the lactation period. In pups from dams that treated only during gestation, serum cV1q concentrations decreased from LDs 2 to 15. In a separate study, blood was collected from the pooled fetuses from two dams that were treated during gestation. These fetuses had high serum concentration of cV1q. Low concentrations of cV1q were detected in breast milk.
DICUSSION
These studies demonstrate that treatment with an anti-TNFα monoclonal antibody throughout pregnancy and lactation had little or no effect on selected immune parameters in F1 generation through 22 weeks of age.
The dose range of cV1q selected for these studies (10 to 40 mg/kg) has previously been shown to be efficacious at reducing disease severity in a murine model of Crohn’s disease (Marini et al. 2003). Therefore, in these studies the dams were administered doses of cV1q that would be sufficient to neutralize mouse TNFα during active inflammation.
Exposure of F1 generation mice to cV1q during gestation and during the postnatal period produced a slight but significant reduction in the SRBC-induced spleen cell expansion in the 11-week-old females when compared to the vehicle control–treated females. A statistical difference in spleen weights was not detected in male F1 generation mice. Because the number of spleen cells was lower in the cV1q-treated female mice, the IgM antibody response per spleen was significantly reduced. However, when the IgM response was normalized to the number of spleen cells, no significant differences were evident. Therefore, potential treatment related effect of cV1q appears to be a reduction in splenic cellular expansion in response to SRBC immunization rather than to a reduced ability of the cells to generate IgM. Although a slight reduction in the splenic response to SRBC immunization was detected in 11-week-old female mice in these studies, the small magnitude of the effect and the fact that it was detected in only one of two studies and was not observed in 22-week-old mice suggests that, although statistically significant, this finding may not be clinically significant. Exposure to cV1q had no significant effect on lymphocyte subset populations and did not inhibit NK cell activity or T-cell proliferation in vitro in cells from nonimmunized mice.
Although serum concentrations of cV1q were not measured in 11-week-old pups, it is possible that the pups that were exposed to high concentrations of cV1q during gestation and lactation may still have had sufficient cV1q in their circulation at 11 weeks of age to produce a pharmacological effect. Antibodies are known to be cleared slowly from the serum and in pups that were exposed only during gestation, the serum concentrations of cV1q decreased by approximately 50% over the 2-week postpartum period, suggesting a half-life of cV1q in the order of 2 weeks. Therefore, the effects seen at 11 weeks of age could be a direct biological effect of TNFα inhibition rather than to a long-term adverse effect on immune system development.
The mouse data with cV1q provide supportive evidence for a lack of a long-term clinically significant detrimental effect on the developing immune system following maternal exposure to an anti-TNFα monoclonal antibody. However, the data cannot be considered conclusive evidence for human safety because there are human versus mouse differences in the development of the immune system and in the way in which IgG antibodies are transferred between mother and fetus. With regard to the immune system, mice are developmentally delayed relative to humans such that a mouse at birth is similar to a human fetus at the end of the second trimester (Holsapple, West, and Landreth 2003). In rodents, transfer of IgG antibodies from the mothers to the pups occurs via the fetal yolk sac during gestation and via the milk during lactation, whereas in humans IgG antibodies are transferred primarily across the placenta (during the latter part of gestation), with IgA antibodies being the predominant subtype secreted in the milk (Van de Perre 2003). However, irrespective of the route of exposure to monoclonal antibodies, in both mice and primates, exposure occurs during gestation and during the postnatal period and covers a critical period in the development of the immune system.
In rodents, the demarcation of the splenic architecture occurs at about 6 days after birth, B-cell follicles appear at about 2 weeks of age, and the ability to form germinal centers occurs at about 2 to 3 weeks of age (Dijkstra and Dopp 1983). Because these splenic developmental events are altered in the TNFα knockout mice, then this is the period that would be expected to be the most susceptible to anti-TNFα monoclonal antibody treatments. In the current study, pups received substantial exposure to cV1q during gestation and during the postnatal period covering these critical periods in immune system development. However, although a transient inhibitory effect on immune function during cV1q exposure cannot be ruled out from this study, there appeared to be no notable immune suppression in 11- or 22-week-old mice.
In humans, the demarcation of the splenic architecture occurs at about week 26 of gestation and germinal centers form by 6 to 8 weeks after birth (Holsapple, West, and Landreth 2003). IgG antibodies cross the human placenta to the greatest extent during the third trimester such that at the time of birth, the neonatal antibody concentrations can exceed the maternal concentrations (Malek, Sager, and Schneider 1994; Malek et al. 1996). Therefore, the greatest exposure to an anti-TNFα monoclonal antibody in humans would be expected during the third trimester when demarcation of the splenic architecture occurs and for a number of months postnatal when germinal centers are developing. Therefore, both the mouse and the human would be exposed to antibodies during similar critical periods in the development of the immune system.
In summary, these studies demonstrate that treatment with an anti-TNFα monoclonal antibody throughout pregnancy and lactation had little or no effect on selected immune parameters in mice, with only a possible minor attenuation of spleen cell response to immunization noted in the female F1 generation at 11 weeks of age.
Footnotes
Figures and Tables
Acknowledgements
The authors would like to thank George Dearlove (CRL) for study direction; Charles Pendley, Bruce Miller, and Joyce Ford (Centocor) for their contributions to the cV1q assay development and conduct; Bill Pikounis and Charles Miller (Centocor) for statistical analysis; and Hugh Davis (Centocor) for critical review of the manuscript. These studies were funded by Centocor R&D, Inc.