Supplemental Materials

Modeling of the “dosing effect”

Apparent IL-6/CRP increases in response to IV dosing were noted in our study. As shown in sFig 1A, following an IV bolus dose of PBS (Groups 3 and 4, prior to siltuximab dosing), a pronounced IL-6 elevation was observed. Siltuximab was able to reduce this initial IL-6 elevation in a dose-dependent manner. In addition, the Cmax of IL-6 following the 4 hour IV infusion of IL-6 was shown to be at 4.5 hr (sFig 1B), suggesting the increase in IL-6 was a mixed effect of both the “dosing effect” and the amount of IL-6 being infused. The “dosing effect” was most likely due to the body’s natural stress response to dosing.

To model the “dosing effect” in response to IV bolus administration, the additional production of IL-6 was approximated using a first-order depot input. Equation (4) was modified to the following form:

dRtot/dt = Rin1 · exp(–ka · t) + ksyn – (kel_cplx – kel_IL-6) · (Ctot – C) – kel_IL-6 · Rtot (s1)

where Rin1 represents the first-order input rate of free IL-6 from the depot at time 0 and ka represents the first-order rate constant of the depot input.

To model the “dosing effect” in response to the 4 hour IV infusion, the additional production of IL-6 was approximated using a zero-order infusion input of IL-6 followed by a first-order input of IL-6 from a depot. Equation (4) was modified to the following form:

dRtot/dt = Rin2 · exp(–ka · t) + OCC1 · Fin · Rin2 + OCC2 · Fin · Rin2 · exp(–ka ·(t – 1.167)) + ksyn – (kel_cplx – kel_IL-6) · (Ctot – C) – kel_IL-6 · Rtot (s2)

where Rin2 represents the first-order depot input rate of free IL-6 at time 0 due to the IV bolus injection of siltuximab, ka represents the first-order rate constant of the depot input, OCC1 and OCC2 are two dummy indicators with OCC1=1 between 0 to 4 hours after the initiation of IV infusion and 0 otherwise, and OCC2=1 after 4 hours past the initiation of IV infusion and 0 otherwise, Fin represents the fraction change of input rate of free IL-6 between the IV bolus injection and the following IV infusion, and t represents time in days.

For Groups 1, 3 and 4, animals only received an IV bolus dose of siltuximab on Day 0, so only an IV bolus “dosing effect” (Eq. s1) was incorporated. For Group 2, since the animals received an IV bolus dose of siltuximab (Day 42) followed by an IV infusion of IL-6 (Day 43), both IV bolus and IV infusion (Eq. s2) dosing effects were incorporated. Additionally, the initial value for Rtot was defined by a new parameter, Base, instead of ksyn/kel_IL-6, because IL-6 elevation was already apparent at time zero and the IL-6 levels at time 0 do not truly reflect the actual baseline levels.

The estimated “dosing effect” related parameters for the PK/TE Model was presented in Supplemental Table I (sTable I). The normal baseline production rate constant of IL-6 is 0.00712 nM/day, i.e. ~200 pg/mL/day. The transient production of IL-6 in response to dosing event (Rin) was shown to be significantly higher than the normal IL-6 production rates, indicating that the production of IL-6 has the potential to increase dramatically in response to stimulus. Interestingly, the transient production rate constant of IL-6 for the high dose groups was ~10-folder lower than those for the low dose groups, presumably because animals in the high dose groups (Groups 3 and 4) were relatively less stressed at the time of siltuximab dosing because they had been through the IV bolus dosing process once (i.e. PBS dosing).

Model Diagnostics

For the PK/TE model from low siltuximab dose groups, the Goodness-of-fit plots showed that the fitting is good for both total siltuximab and total IL-6 with no bias (sFigs. 2 and 3). The fitting for free IL-6 is also reasonably good with some over-predictions and under-predictions (sFig. 4). The individual fit plots for PK/TE model from low siltuximab dose group was presented in sFigs.5 - 7. The black circles represent measured data, the red lines represent individual fitting, and the blue dotted lines represent population fitting. The LLOQ of free IL-6 assay is shown as a green dotted line (1.7x10-5 nM, or 0.48 pg/mL), and free IL-6 levels <LLOQ was plotted at ½ of LLOQ (sFig.7). As shown in sFig.7, nearly all the over-predictions of free IL-6 occurred at the <LLOQ data, which may be related to the fact that the predicted free IL-6 values were very close to LLOQ. As to the under-predictions of free IL-6, they all happened at very early time points. One possible explanation is that the total/free IL-6 ratio was relatively high at these very early time points, which can lead to bias in free IL-6 measurement.

For the PK/TE model from low siltuximab dose groups, the fitting is good for both total siltuximab and total IL-6 with no apparent bias (data not shown). For the TE/PD model, the Goodness-of-fit plots showed that the fitting is good with no apparent bias (sFig 8).

Free IL-6 concentration from Day 0-42 of Group 2 animals

Free IL-6 concentrations from Day 0-42 of Group 2 animals, i.e. following the first two IL-6 infusions and prior to the dosing of siltuximab are shown in sFig 9.