Why Dialysis Removal Needs Its Own Exposure Calculation
For a drug cleared partly by dialysis, a low concentration immediately after a session does not by itself show how much drug the dialyzer removed, how much was cleared by other routes, or what concentration will be measured after redistribution. Those distinctions matter when estimating exposure during dialysis and when considering a supplemental dose. The calculation should begin with the relevant clearance and session conditions, then account for changing concentration and rebound.
The model developed here is a deliberately explicit starting point. It describes a well-mixed, one-compartment body with constant volume of distribution and constant clearances during a dialysis session. Real drugs can depart from these assumptions, especially when distribution between compartments is slow or protein binding limits transfer to the dialyzer. The calculations are therefore estimates conditional on the model, not universal dose rules.
Removal During a Session: Model and Derivation
Let \(C(t)\) be the plasma concentration during dialysis, \(V\) the apparent volume of distribution, \(CL_d\) the dialysis clearance, and \(CL_o\) the combined non-dialysis clearance. Assume \(V\), \(CL_d\), and \(CL_o\) are constant over a session of duration \(T\); drug follows linear, one-compartment kinetics; and the body is well mixed so the concentration presented to the dialyzer is represented by \(C(t)\). Under these assumptions, the amount in the body is \(A(t)=VC(t)\), and the rate of loss is the sum of the two clearance pathways:
Define \(k=(CL_d+CL_o)/V\), with units of inverse time. Solving the differential equation gives \(C(t)=C_0e^{-kt}\), where \(C_0\) is the concentration at session start. The amount removed by the dialyzer is the integral of its instantaneous removal rate, \(CL_dC(t)\). Integrating the exponential concentration profile yields:
The same derivation with \(CL_o\) gives the amount lost through non-dialysis clearance during the session. A useful check is mass balance: the sum of those two amounts must equal the decrease in body amount, \(V[C_0-C(T)]\). This check can reveal arithmetic mistakes or inconsistent assumptions.
Record \(C_0\), \(V\), session duration \(T\), and the clearances that apply during the session. Keep units consistent.
Calculate \(k=(CL_d+CL_o)/V\), then \(C(T)=C_0e^{-kT}\).
Calculate \(A_d(T)\) and the non-dialysis amount separately; verify their sum equals \(V[C_0-C(T)]\).
Worked Example: Partitioning Removal During a Four-Hour Session
Consider a hypothetical drug with \(C_0=8.00\) mg/L, \(V=40.0\) L, \(CL_d=18.0\) L/h, and \(CL_o=5.00\) L/h during a four-hour session. Assume the one-compartment, constant-clearance model applies.
The total clearance is \(23.0\) L/h, so \(k=23.0/40.0=0.575\) h\(^{-1}\). The predicted end-session concentration is:
Dialyzer removal is calculated using the integrated concentration, not by multiplying the starting concentration by the whole session duration:
Non-dialysis removal during the same session is:
The modeled decrease in body amount is \(40.0\ \text{L}\times(8.00-0.8021)\ \text{mg/L}=287.9\) mg. The pathway amounts sum to \(225.3+62.59=287.9\) mg, subject to rounding. Thus, about 225.3 mg is attributed to dialysis and 62.59 mg to other clearance in this model.
Protein Binding and Interpreting Dialysis Clearance
Only drug that reaches the dialyzer and can cross its membrane is available for extracorporeal removal. Molecular size, membrane properties, blood or plasma flow, and binding to plasma proteins can all affect that transfer. For a highly protein-bound drug, the unbound fraction may be small, limiting immediate availability for membrane passage. However, protein binding alone does not determine dialysis clearance: bound drug can dissociate as unbound drug is removed, and the rate of that exchange matters.
Consequently, multiplying total clearance by the unbound fraction is not a general substitute for a measured dialysis clearance. If inlet and outlet concentrations are measured appropriately, \(Q_pE\) estimates clearance on the concentration basis used in that calculation. The result should be interpreted with attention to whether concentrations are total or unbound and whether the flow is plasma or blood flow. A clearance measured on one basis should not be combined with concentrations on another without a justified conversion.
Worked Example: Estimating Clearance From Dialyzer Extraction
In a hypothetical session, plasma flow through a dialyzer is \(0.750\) L/min. Matched plasma samples show \(C_{in}=2.00\) mg/L and \(C_{out}=0.800\) mg/L. The drug's measured unbound fraction is \(0.050\). Assume the inlet and outlet measurements represent the same flow conditions and use plasma concentrations with plasma flow.
The measured extraction ratio is:
Convert plasma flow to hourly units: \(0.750\ \text{L/min}\times60=45.0\) L/h. Therefore:
This is an estimate based on the observed total-plasma extraction and stated flow. The unbound fraction of \(0.050\) signals that binding deserves attention, but it does not imply that the clearance must equal \(0.050\times27.0\) L/h. Nor does this single clearance measurement establish how clearance changes as concentration falls or binding equilibrates. Those questions require suitable measurements or a more detailed model.
Post-Dialysis Rebound Is Not Additional Drug Input
A concentration measured immediately after dialysis may be lower than a later post-session concentration. One explanation is redistribution from tissues or other compartments into plasma after the dialyzer has stopped removing drug. The plasma concentration rises, but that rise does not mean the body has gained drug: it reflects movement within the body. Rebound can therefore make an immediate post-session sample a poor proxy for a later concentration or for total body amount.
A descriptive rebound percentage can be defined relative to the immediate post-session concentration, but the denominator must be stated. For example, if \(C_{imm}\) is the immediate concentration and \(C_{later}\) is a later value, the percentage increase relative to \(C_{imm}\) is \(100(C_{later}-C_{imm})/C_{imm}\). This is not a percentage of the predialysis concentration, and neither percentage directly measures the fraction of the removed amount that returned to plasma.
Worked Example: Quantifying a Delayed Concentration Rise
Suppose hypothetical samples show \(C_{imm}=0.800\) mg/L immediately after a session, \(C_{later}=1.25\) mg/L two hours later, and a predialysis concentration of \(8.00\) mg/L. Assume the samples are comparable and the later rise occurs after dialysis has stopped.
The absolute increase is \(1.25-0.800=0.450\) mg/L. Relative to the immediate value, the increase is:
Relative to the predialysis concentration, the increase is instead \(100(0.450/8.00)=5.625\%\). These are different summaries with different denominators. The observed rise is consistent with redistribution, but two concentration measurements alone do not identify compartment volumes, transfer rates, or the amount redistributed.
Supplemental Dosing: Estimate the Exposure Change, Not Just the Removed Amount
The amount removed by dialysis is not automatically the amount that should be given back. A supplemental dose decision depends on the therapeutic target, dose route and bioavailability, residual drug, post-dialysis redistribution, clearance after the session, and the timing of the next scheduled dose. A dose intended to restore concentration immediately may be inappropriate when absorption is delayed or rebound is expected. Conversely, an amount removed during dialysis may have had limited impact on the clinically relevant exposure interval.
Under linear pharmacokinetics, if an oral supplemental dose \(D_s\) has bioavailability \(F\) and post-session clearance \(CL_o\) remains constant, its eventual incremental area under the concentration-time curve is \(F D_s/CL_o\). This calculation assumes the dose is absorbed and eliminated under those conditions, and does not describe its concentration-time peak or the timing of that exposure. It is one way to quantify the potential size of an exposure change, not a dosing recommendation.
Worked Example: Evaluating a Hypothetical Partial Replacement Dose
Use the dialysis-removal estimate of 225.3 mg from the first example. For an illustrative decision analysis, consider replacing 25% of that dialysis-attributed amount. Suppose the drug is given orally with \(F=0.800\), and post-session clearance is \(5.00\) L/h. A regular oral dose is 400 mg. Assume linear pharmacokinetics and constant post-session clearance.
The chosen systemic replacement amount is \(0.25\times225.3=56.33\) mg. The oral amount needed to deliver this systemic amount under the assumed bioavailability is:
The corresponding incremental AUC is \(F D_s/CL_o\):
For comparison, the modeled AUC contribution from the 400 mg regular oral dose under the same assumptions is \((0.800)(400\ \text{mg})/(5.00\ \text{L/h})=64.0\ \text{mg}\cdot\text{h/L}\). The hypothetical supplement therefore adds \(11.27/64.0=17.6\%\) of that reference dose's AUC contribution. If a prespecified decision rule in this hypothetical exercise regarded an increment above 10% as material, this increment would meet that criterion. That conclusion is conditional on the assumed replacement fraction, bioavailability, clearance, and linearity; it does not establish that 25% replacement is appropriate clinically.
Diagnostics, Alternatives, and Limitations
A one-compartment calculation is useful when the goal is a transparent first estimate and the data support an approximately exponential decline. Check units, ensure clearances apply to the same concentration basis, and verify mass balance. If the measured concentration profile does not resemble the predicted exponential, or if estimated dialyzer extraction changes substantially during a session, constant clearance may be inadequate. Sampling time also matters: an immediate post-session sample and a delayed sample can differ because of rebound.
The main alternatives answer different questions. Direct inlet-outlet extraction measurements can estimate instantaneous dialyzer clearance without assuming the whole-body concentration trajectory, but they may not describe clearance over the entire session. A one-compartment model can integrate changing concentrations and partition removal pathways, but assumes rapid mixing and fixed parameters. A multi-compartment model can represent delayed distribution and rebound, but requires enough informative concentration-time data to estimate its additional parameters. More mechanistic dialysis models may also represent flow, membrane transfer, and binding explicitly; greater detail is useful only when supported by data.
Common failure modes include treating \(CL_d C_0 T\) as session removal despite a declining concentration, attributing the entire body-amount decrease to dialysis, inferring negligible removal from high protein binding alone, and interpreting rebound as drug newly entering the body. Another pitfall is to replace every milligram removed with a supplemental dose. That ignores continuing non-dialysis clearance, dose bioavailability, the exposure interval, and concentration rebound. A meaningful dose assessment states its target and assumptions, calculates the expected exposure change, and considers uncertainty in clearance and distribution.
Check Your Understanding
Attempt each question, then check the solutions below.
- A hypothetical one-compartment drug has \(C_0=5.00\) mg/L, \(V=20.0\) L, \(CL_d=4.00\) L/h, \(CL_o=2.00\) L/h, and a two-hour session. What is \(k\), and what is the predicted end-session concentration?
- For the same session, write the expression for dialysis-attributed removal. Why is \(CL_dC_0T\) not the correct integrated amount under the stated model?
- A concentration is 0.600 mg/L immediately after dialysis and 0.900 mg/L one hour later. Calculate the percentage increase relative to the immediate concentration. What does this calculation not establish?
- In one sentence, explain why a high protein-binding fraction cannot by itself determine dialysis clearance.
Solutions. (1) \(k=(4.00+2.00)/20.0=0.300\) h\(^{-1}\), and \(C(2)=5.00e^{-0.600}\approx2.744\) mg/L. (2) \(A_d(2)=(4.00\times5.00/0.300)(1-e^{-0.600})\approx30.08\) mg. The concentration declines during the session, so clearance must be integrated against \(C(t)\), rather than multiplied by the starting concentration for the full duration. (3) \(100(0.900-0.600)/0.600=50.0\%\). This does not identify the amount redistributed, the compartment volumes, or a total-body drug amount. (4) Binding affects the availability and transfer of drug, but the observed clearance also depends on factors such as binding dissociation, flow, and membrane transfer, so binding alone is insufficient to calculate it.