Developing mechanistic models of drug effects in disease states requires understanding of both disease progression and how the drug acts on intermediary factors. We examined CIA progression in the rat and subsequent effects of DEX on the relevant factors driving edema and bone turnover. Endogenous CST plays a role in mediating cytokine turnover in disease progression (Turnbull and Rivier, 1999
; Neeck et al., 2002
). Therefore, effects of DEX and other CS may be included with minimal modifications of the disease progression model. Only the drug kinetics and drug-receptor binding need definition. Our experimental design effectively integrates DEX pharmacodynamics with disease progression to better resolve parameters of both components.
Pharmacodynamic data for DEX are critical for understanding the interrelationships between cytokines, GR mRNA, and CST. As the disease progresses, immune factors such as chemo/cytokines turn over quickly. Corticosterone concentrations and GR mRNA are up-regulated slowly so that cytokines remain in balance and no abrupt inhibition or changes are observed in the natural disease progression. However, when DEX is administered and concentrations of DRN spike, rapid changes occur in all relevant profiles revealing rate-limiting steps, rates of loss and production, and edema and BMD responses to different concentrations of cytokines. This modeling effort not only extracts quantitative biological relationships concerning cytokine effects on paw edema and BMD, but also yields implications about the role of corticosteroids and other therapies that target cytokines in inflammatory arthritis.
The time course of TNF-α mRNA after acute 0.225 and 2.25 mg/kg SC DEX was unusual because of the increase in TNF-α expression both early and later in time. Interestingly, IL-6 was shown to inhibit the expression of TNF-α (Schindler et al., 1990
). With the rapid and near complete drop in IL-6 mRNA production post-dose, it is possible that if IL-6 consistently inhibited TNF-α production, then TNF-α mRNA would increase when IL-6 concentrations fell. Based on the in vitro
evidence for this suppression, the maximal suppression of TNF-α mRNA was fixed at 30% inhibition (Schindler et al., 1990
). The model then captured both this rise and fall and rise again in TNF-α mRNA response.
That the expression of IL-6 mRNA was highly sensitive to DEX indicates that the decline observed in IL-6 mRNA was related to CST. With increased CST and GR mRNA, DRN was increased sufficiently to cause a drop in IL-6 but not in the other measured cytokine mRNA. While this explains in part the decline of IL-6 mRNA, it could not be the entire reason. If DRN increased rapidly as in the case of CS dosing, CST would be inhibited almost completely and for a prolonged time, such that when DRN returned to normal and CST remained suppressed, the IL-6 mRNA production would overshoot the measured response. The ‘remission’ compartment presented in Part I helped capture this decline. It is possible that as the disease develops into a more chronic state there is a shift from the innate to a more humoral immune response, effectively altering concentrations of pro-inflammatory cytokines.
Turnover of GR mRNA after DEX behaved differently than the cytokines. There was an abrupt decline followed by a rebound to values higher than found in the natural disease progression with the low dose. If TNF-α had a larger contribution to GR mRNA up-regulation, then it is possible that this rebound would have been observed. This would have also been seen for the higher dose. A time delay was necessary to account for the slow rate of disease progression that did not plateau as quickly as cytokine mRNAs while a fast drop was observed in response to DEX. Transit compartments accounted for the delayed response to cytokines while allowing the equation describing GR to have a high rate of turnover to reflect rapid drug effects.
Binding of DEX to GR was modeled using literature reported KD
values to adjust the binding of CST in the presence of DEX in such a way that only ffgc
was fitted for both drugs. This parameter corrects for both the fraction of drug that equilibrates from plasma into tissue and the free fraction able to bind receptor. Distribution is assumed to be instantaneous. The release rate constants (kre_C
) from the nucleus were reasonably estimated compared with the kre
for methylprednisolone (MPL, kre_M
) since CST is less potent than MPL while DEX is more potent. The release constant for DEX-GR complex from the nucleus was also much lower than that of CST-GR complex and MPL-GR complex. If both molecules exert their effects to the same extent through GR in the nucleus then DEX would need to remain bound in the nucleus for a longer duration to yield the differences in observed inhibition of cytokine mRNA for CST during disease progression versus DEX concentrations following dosing. Interestingly, the profile for DRN
after chronic dosing of 0.045 mg/kg DEX () quickly approached the steady-state DRN
as exhibited for 0.225 mg/kg DEX. This meant that the single 2.25 mg/kg and 7-day 0.225 multiple-doses of DEX were likely exerting near maximal responses. This steady-state was limited by the concentrations of free GR receptor. The estimate for synthesis of GR in inflamed tissue was 0.1054 hr−1
compared to 0.54 hr−1
in liver tissue reported previously (Schindler et al., 1990
; Ramakrishnan et al., 2002
; Hazra et al., 2007b
; Hazra et al., 2007c
As DEX affects cytokine mRNA expression and not edema directly, it is the inhibition of cytokines that governs the decline in paw edema. In general the model captured that data well with modest over-prediction for the 0.225 mg/kg acute and chronic dosing. This over-prediction may be due in part to the linear additive relationship between cytokine mRNA and paw edema. If the relationship between cytokines and edema is receptor driven, then a Hill-type effect relationship may be more appropriate. It is also possible that because there are other processes which contribute to edema not included in the model, their effects on up-regulating paw edema are missed. However, altering each signaling pathway alone and simultaneously in different combinations while measuring relevant paw edema is impractical, thus the true effect of cytokines on edema may be mis-specified. For the key inflammatory signaling pathways that were measured, this model captured the data well and linear stimulatory parameters reflect the relative contributions of each cytokine to paw edema.
The effect of DEX on bone was observed by two different responses. 1) An increase in BMD was observed almost immediately in arthritic animals as concentrations of pro-inflammatory cytokines fell. 2) Effects of DEX on osteoblast apoptosis/reduced-activity resulted in decreased BMD in all scanned regions of both healthy and arthritic animals approximately 100 hr (5 days) after the first dose. Higher doses of CS overpowered the anti-inflammatory effects yielding a reduced BMD. indicated that much lower doses of DEX were sufficient to reduce cytokines IL-1β and IL-6 mRNA, increasing BMD and mitigating adverse effects. The cytokine mRNA with the least effect on BMD was TNF-α, which had the highest IC50
value (550 nM). The cytokine with the lowest IC50
value (4.5 nM), IL-6, had the greatest contribution to BMD loss. The observation that the cytokine most sensitive to DRN
contributed the most to reducing BMD density provided a potential explanation for why recent studies involving low dose CS were able to halt radiographic damage to joints (Svensson et al., 2005
; Wassenberg et al., 2005
; Da Silva et al., 2006
The doses of 0.225 and 2.25 mg/kg were chosen to produce DEX concentrations in rat plasma similar to the lowest and highest exposures in humans. If these are relevant clinical doses based on PK, free fraction of drug, and drug-receptor binding constants, then would suggest that the minimally effective doses (0.225 mg/kg in the rat) are 20 times in excess of the optimal dose. Concentrations at higher doses of DEX will produce effects on bone loss which will dominate the protective effects. Owing to the lower binding affinity and potentially reduced kre
value for prednisolone, concentrations of DRN
may approach those of this lower ‘optimal’ dose and potentially explain why recent studies with low-dose prednisolone have appeared protective for BMD (Buchwald, 2007
). Additionally, the model suggests that since lower doses of CS should be given and these doses may be insufficient to suppress TNF-α, a second agent given to either reduce concentrations of TNF-α or inhibit TNF-α effects on inflammation would be beneficial to both edema and BMD responses.
Owing to the extensive clinical history of corticosteroid use, their pharmacokinetics are well established as are receptor binding constants making this modeling paradigm potentially useful for translation to the clinic. It is advantageous that the model processes are based on known physiology of RA and that the major factors understood to control inflammation can be related to each other to describe disease progression and drug effects on both the molecular and symptomatic aspects of chronic auto-immune arthritis.