Effect of the Soluble TNF-Antagonist Etanercept on Tumor Necrosis Factor Bioactivity and Stability

Abstract

Background: Targeted anti-tumor necrosis factor (TNF) strategies in patients with rheumatoid arthritis have resulted in new and/or worsening heart failure in individuals who were free of cardiovascular disease.

Methods and Results: To determine the mechanism of new and/or worsening heart failure in patients who were receiving the soluble TNF-antagonist etanercept, we analyzed frozen plasma samples from a previous clinical trial with etanercept in heart failure patients, and conducted complimentary mechanistic in vitro studies. Analysis of the clinical trial data showed that use of etanercept resulted in a significant 70-fold increase in the level of immunoreactive TNF. Complimentary in vitro studies using an L929 bioassay showed that at low concentrations of etanercept relative to TNF there was an unexpected 1.5- to 1.75-fold increase in the absolute level of TNF bioactivity. We also examined the effect of etanercept on TNF stability and the results showed that there was a two-fold increase in the mass of bioactive homotrimeric TNF when the molar ratio of TNF to etanercept was approximately 200:1.

Conclusion: Etanercept increases the immunoreactive mass of TNF in heart failure patients, as well as augments TNF cytotoxicity in certain settings, thus suggesting one potential mechanism for the worsening heart failure in some patients who were receiving this agent.

Introduction

Tumor necrosis factor (TNF) inhibitors have emerged as a new treatment option for rheumatoid arthritis, as well as psoriatic arthritis and Crohn's disease (reviewed in Ref. 1). However, the long-term experience with TNF inhibition has been limited. Moreover, there have been increasing reports of unforeseen cardiovascular side effects with these agents. Indeed, a recent analysis of the US Food and Drug Administration's MedWatch program described patients who developed new and/or worsening heart failure during TNF-antagonist therapy with either infiximab or etanercept.² The recently reported ATTACH trial (anti-TNF-α in congestive heart failure) showed that there was a significant worsening of heart failure for patients with New York Heart Association (NYHA) class III or IV heart failure who received the highest doses of infiximab.³ Finally, in the RENAISSANCE trial (randomized etanercept North American strategy to study antagonism of cytokines), in which patients with NYHA class II–IV heart failure were treated with etanercept, the risk of worsening heart failure was increased for the patients who were treated with the highest dose of etanercept for the longest period of time.⁴ To address the potential mechanisms for worsening heart failure that have been reported with etanercept, we have analyzed frozen plasma from a previously reported clinical trial with etanercept in heart failure patients,⁵ as well as conducted complimentary mechanistic in vitro studies.

Methods

Study cohort

Plasma samples were obtained from frozen samples from a previously reported study of 47 patients with stable NYHA class III–IV heart failure, who were randomized to receive placebo (n= 16), 5 mg/m² of etanercept (n= 12), or 12 mg/m² etanercept (n= 15).⁵ TNF levels were measured using an ELISA (R&D systems, Minneapolis, MN, USA) that measures “total” TNF (i.e., free [unbound] cytokine and cytokine bound to receptors).

Effects of etanercept on TNF bioactivity

Recombinant human TNF-α (R&D systems, Minneapolis, MN, USA) was used throughout these studies. TNF bioactivity was assessed in vitro using an L929 bioassay as described in Ref. 6. First we employed a fixed concentration of TNF (200 U/mL) that was incubated with increasing concentrations of monomeric p55 and 75 TNF receptors (0.2 × 10⁻¹² M to 5 × 10⁻⁹ M), as well as increasing concentrations of etanercept (0.2 × 10⁻¹² M to 3.9 × 10⁻¹¹ M). Next, we used three dieferent concentrations of TNF (50–200 U/mL) with increasing concentrations of etanercept (5.2 × 10⁻¹² M to 8.8 × 10⁻¹² M). The mixture of TNF and TNF receptors was added to the L929 cells, and cytotoxicity was determined at 24 hours.⁶ All values were reported as the mean of duplicate measurements.

Effects of etanercept on TNF stability

TNF exists in the circulation as biologically inactive TNF monomers and dimers, as well as biologically active trimers. In order to determine whether etanercept disrupted the equilibrium between TNF monomers, dimers, and trimers, we incubated a fixed concentration of ¹²⁵I-TNF (2 ng/mL; 200 U/mL) with increasing concentrations of etanercept, followed by cross-linking with bis(sulfosuccinimidyl)-suberate (BS³; Pierce), as described in Ref. 6. The gels were then dried, and the receptor-ligand complexes visualized by autoradiography. Final results were expressed as the ratio of trimeric to monomeric TNF, in order to account for any possible differences in loading of the gel.

Statistical analysis

Data are expressed as mean ± SD. A one-way analysis of variance (ANOVA) was used to examine differences in immunoreactive TNF levels between patient groups and to analyze differences in TNF bioactivity with different concentrations of etanercept. Two-way ANOVA was used to analyze differences in TNF bioactivity with different concentrations of etanercept. Two.-way ANOVA was used to analyze differences in TNF bioactivity between p55, p75, and etanercept treatment groups. Post hoc analysis of variance testing (Tukey's test) was performed where appropriate. A significant difference was said to exist at p < 0.05.

Results

Immunoreactive TNF levels in heart failure patients

Figure 1 shows the peripheral circulating levels of immunoreactive TNF were not different from baseline for the patients who received placebo, whereas there was an approximate 70-fold increase in immunoreactive TNF levels for the patients who received 5 and 12 mg/m² of etanercept. Analysis of variance indicated that there were significant (p < 0.0001) overall differences between groups; post hoc analysis indicated that there was a significant increase in immunoreactive plasma levels between etanercept and placebo (p < 0.001), but not between etanercept groups (p= 0.98) .

Figure 1

Effect of etanercept on TNF immunoreactivity. Circulating levels of TNF were measured at baseline, and after 3 months of therapy with placebo (A) 5 mg/m² of etanercept (B) and 12 mg/m² of etanercept in patients with NYHA class III–IV heart failure (*=p < 0.001 compared to placebo).

Effects of etanercept on TNF bioactivity

Figure 2A (n= 5 experiments) shows the results of the studies wherein we treated L929 cells with a fixed concentration of TNF (200 U/mL) in the presence of increasing concentrations of monomeric p55 TNF receptor, monomeric p75 TNF receptor, and dimeric p75 TNF receptor (etanercept). As expected, low concentrations of monomeric p55 and p75 receptors and etanercept had no effect on TNF bioactivity, whereas high “neutralizing concentrations” of the monomeric p55 and p75 receptors and etanercept completely abrogated TNF-induced killing of the L929 cells. As shown, on a molar basis, etanercept was more efficient than either the p55 or p75 monomeric receptors in terms of neutralizing the cytotoxic effects of TNF. However, the surprising finding shown by Figure 2A was that there was a significant (p < 0.05) 1.6-fold increase in the absolute level of TNF bioactivity, as the concentration of etanercept was increased from 2 × 10⁻¹³ M to 6 × 10⁻¹³ M.

Figure 2

Effect of soluble TNF receptors on TNF immunoreactivity. (A) A fixed concentration of TNF (200 U/mL) was incubated with increasing concentrations of monomeric p55 and 75 TNF receptors (0.2 × 10⁻¹² M to 5 × 10⁻⁹ M) and increasing concentrations of etanercept (0.2 × 10⁻¹² M to 3.9 × 10⁻¹¹ M). (B) Increasing concentrations of TNF (50–200 U/mL) were incubated with increasing concentrations of etanercept (5.2 × 10⁻¹² M to 8.8 × 10⁻¹² M). TNF bioactivity was determined at 24 hours (see text for details). All values are expressed as a fold-change relative to the level of TNF bioactivity in the absence of soluble TNF receptors. Values above the dotted line indicate an absolute increase in TNF bioactivity, whereas values below the dotted line indicate an absolute decrease in TNF bioactivity (*=p < 0.05 compared to placebo).

To further explore the increase in TNF bioactivity with etanercept, we used three different concentrations of TNF in the presence of increasing concentrations of etanercept. Figure 2B (n= 5 experiments) shows two important fndings. First, the molar ratio at which etanercept resulted in an increase in TNF bioactivity varied in relation to the concentration of TNF employed in the bioassay That is, etanercept led to an increase in TNF bioactivity at a molar ratio of 200:1 (TNF:etanercept) when 200 U/mL of TNF was employed, whereas etanercept led to an increase in TNF bioactivity at a molar ratio of 35:1 (TNF:etanercept) when 50 U/mL of TNF was employed. Second, the absolute magnitude of the increase in TNF bioactivity with etanercept was directly proportional to the level of TNF.

Effects of etanercept on TNF stability

To determine whether the concentrations of etanercept that increased absolute TNF bioactivity also affected TNF stability, we performed cross-linking experiments with a fixed concentration of ¹²⁵I-TNF and increasing concentrations of etanercept. Figure 3 shows that as the concentration of etanercept increased to 0.6 × 10⁻¹² M, the ratio of trimeric to monomeric TNF increased approximately two-fold relative to values that were observed in the absence of etanercept, suggesting that etanercept increased the stability of TNF as a biologically active homotrimer. This autoradiogram is representative of three similar experiments.

Figure 3

Effects of etanercept on TNF stability. (A) A fixed concentration of ¹²⁵I-TNF was incubated with increasing concentrations of etanercept (8 × 10⁻¹⁴ M to 6.25 × 10⁻¹⁰ M) or diluent, followed by cross-linking and autoradiography (see Methods for details). This figure is representative of three separate experiments. (B) Final results were expressed as the ratio of trimeric to monomeric TNF, and the results displayed as the fold-change in this ratio relative to control values (i.e., diluent).

Discussion

In an effort to delineate the potential mechanism(s) for worsening heart failure that have been reported with TNF inhibitors,^2,4,⁵ we analyzed frozen plasma samples from a previous clinical trial with etanercept in patients with heart failure,⁵ as well as conducted complimentary in vitro studies. The results of this study show that etanercept resulted in a significant (p < 0.0001) 70-fold increase in TNF immunoreactivity (Figure 1), consistent with prior reports in humans.^7,8 There was no significant change in immunoreactive TNF levels in the placebo-treated patients. Insofar as there was insufficient plasma to perform TNF bioassays, which would have determined whether the increase in immunoreactivity represented an absolute increase in free and/or bound TNF, we performed in vitro bioassays that modeled the effect of changing the concentration of TNF relative to the concentration etanercept. As expected, we found that at high “neutralizing” concentrations, etanercept completely abrogated TNF bioactivity, as reported previously⁶ However, to our surprise there was a significant increase in TNF cytotoxicity from 1.5-to 1.75-fold (p < 0.05), when TNF was present in molar excess of etanercept. Importantly, this increase in TNF bioactivity was not observed with either of the monomeric p55 and p75 receptors, suggesting that the unique dimeric configuration of etanercept was responsible for increasing TNF bioactivity. The studies wherein we examined the effect of etanercept on TNF stability showed that etanercept shifed the equilibrium between TNF monomers, dimers, and trimers in favor of more biologically active TNF homotrimers (Figure 3). Importantly, the concentration of etanercept that led to an increase in the ratio of homotrimeric TNF overlapped the concentration of etanercept that resulted in a 1.75-fold increase in TNF bioactivity (Figure 2B). These in vitro findings confirm and expand previous studies that have suggested that etanercept stabilizes TNF bioactivity,⁹ and that etanercept can act as a “stimulating antagonist” in certain setting. There are two limitations to this in vitro study that should be discussed. First, the concentrations of etanercept used in our in vitro studies are less than those reported in patients with heart failure.¹⁰ Even though etanercept increased the circulating concentration of TNF by approximately 70-fold in heart failure patients (Figure 1), we cannot exclude the possibility that etanercept may be in molar excess to TNF in patients with heart failure. Second, it is important to recognize that it is not possible to precisely know what the actual concentration of any drug or bioactive substance is at the tissue/receptor level in vivo, because of protein binding, enzymatic breakdown, and/or the redox state of the cellular mileau, any or all of which may affect TNF signaling. Thus, our in vitro studies may not predict the behavior of etanercept in vivo. Tat said, the important point raised by our in vitro bioactivity experiments suggest that a ratio of TNF/etanercept can be reached where enhancement of TNF bioactivity occurs over a wide range of different experimental conditions. Moreover, etanercept has been show to increase TNF bioactivity in vivo.¹¹

Conclusion

Taken together, the results of the present study suggest, but do not prove, that etanercept may contribute to worsening heart failure in susceptible individuals by virtue of the unique aspects of this dimeric TNF-receptor fusion protein to stabilize and increase the cytotoxic activity of TNF. Given that TNF is not irreversibly bound to etanercept, but rather has an “on-rate” approximately equal to 620 milliseconds,¹² the increase in the absolute mass of biologically active TNF might lead to adverse consequences if the TNF that is released from etanercept binds preferentially to cognate receptors on endothelial cells, cardiac fibroblasts, and cardiac myocytes, as opposed to preferentially rebinding to etanercept. Our in vitro studies further suggest that when TNF is present in molar excess of etanercept, etanercept stabilizes TNF in its homotrimeric form and increases TNF bioactivity. Our in vitro results suggest that the agonistic properties of etanercept are unlikely to be deleterious when circulating concentrations of TNF are relatively low, insofar as the absolute increase in TNF bioactivity is modest (i.e., from 50 to 75 U/mL). This experimental finding may offer one potential explanation for why etanercept is well-tolerated in rheumatoid arthritis, wherein circulating concentrations of TNF are low. However, when circulating concentrations of TNF are higher, such as in sepsis or heart failure, the agonist properties of etanercept may be harmful in that the absolute increase in TNF bioactivity is more clinically significant (i.e., from 200 to 350 U/mL). Thus, the cardiovascular effects of etanercept in a given patient may be unpredictable, and will depend on the levels of circulating TNF in the periphery and/or changes in the dosing of etanercept that would permit untoward imbalances in the stoichiometric ratio of TNF to etanercept. We would like to emphasize that the results of this study are provisional, and should not be used as a reason to withhold treatment with etanercept from patients with rheumatoid arthritis or psoriatic arthritis, who have clearly been shown to benefit from treatment with this biological response modifier.¹ Nonetheless, our findings are consistent with the clinical trial results wherein treatment with etanercept led to a significant increase in morbidity and mortality.^4,13 Taken together these studies raise a cautionary note, and argue for more physician awareness and/or cardiovascular monitoring for patients who may require this drug to improve their quality of life.

Acknowledgments

We gratefully acknowledge the technical assistance of Lifang Zhao and Dorellyn Blanchette Lee and the secretarial assistance of Ms. Mary Helen Soliz. This research was supported by research funds from the N.I.H. (P50 HL-54313 and RO1 HL58081, RO1 HL61543, RO1 HL73017).