Why Verify Plasma and Urine Together?
A renal PBPK model can reproduce a plasma concentration profile while misrepresenting the processes that produced it. For example, excessive secretion paired with excessive reabsorption may yield a plausible net renal loss but incorrect transporter fluxes. Conversely, a model can predict a plausible cumulative amount in urine while placing the drug in plasma at the wrong times. Verification therefore asks whether the model accounts for when drug is in plasma, how much appears in urine, and where the administered amount goes.
The checks in this tutorial concern predictions against data, not parameter estimation. Compare predictions and measurements under matched conditions: dose, route, sampling times, renal function, urine collection interval, and any relevant transporter or co-medication conditions. The earlier material on renal clearance, filtration, secretion, and reabsorption provides the mechanistic quantities used below. The key task here is to test whether those quantities work together coherently.
Four Complementary Checks
First compare the plasma profile over time. Plot observed and predicted concentrations at the same times, then inspect both the size and direction of residuals. A consistent overprediction late in the profile may suggest a problem with the modeled rate of drug loss or distribution; an early mismatch may instead point to the initial concentration or distribution assumptions. A single summary such as AUC can conceal a timing error, so retain the profile comparison.
For a descriptive check, define the residual at time \(t_i\) as \(e_i=C_{\mathrm{obs},i}-C_{\mathrm{pred},i}\). If measurement uncertainty is available, a standardized residual can be calculated as \(e_i/s_i\), where \(s_i\) is the standard deviation assigned to that observation. Large or systematically signed residuals warrant investigation. A percentage error, \(100(C_{\mathrm{pred},i}-C_{\mathrm{obs},i})/C_{\mathrm{obs},i}\), can be useful when observed concentrations are not near zero; it becomes unstable when the denominator is small. These summaries describe prediction discrepancies, not a test that establishes whether the model is true.
Second compare urinary recovery. Cumulative urinary amount is the sum of the measured amount excreted across collection intervals. Check the interval amounts as well as the final cumulative total: agreement at the end can hide errors in the timing of excretion. Verify collection duration and units, and distinguish unchanged parent drug in urine from metabolites if the model predicts them separately.
Third calculate renal clearance over a defined interval. With \(A_{e,\mathrm{urine}}\) denoting the amount of unchanged drug excreted in urine over that interval and \(AUC_p\) the plasma concentration-time area over the same interval, the conventional interval estimate is:
When \(A_{e,\mathrm{urine}}\) is in mg and \(AUC_p\) is in mg·h/L, \(CL_r\) is in L/h. Use total plasma concentration for this conventional clearance ratio, and use the same plasma basis when comparing model and data. For constant unbound fraction \(f_u\) and glomerular filtration rate \(GFR\), the filtered amount over the interval is \(f_u\,GFR\,AUC_p\), so the filtration contribution to clearance referenced to total plasma concentration is \(f_u\,GFR\). If binding or renal function changes over time, this simple constant-parameter calculation needs the corresponding time-varying quantities.
The difference between observed renal clearance and \(f_u\,GFR\) is a net contribution beyond filtration. It can reflect secretion, reabsorption, or both. It is not, by itself, a direct measurement of transporter activity: opposing secretory and reabsorptive fluxes can produce the same net clearance.
Fourth check amount balance. For an intravenous dose of unchanged parent drug, a parent-drug balance at time \(t\) can be written as:
All terms must use compatible amount units. The terms for metabolism and other loss represent cumulative parent-drug disappearance, expressed in parent-equivalent amount; do not count a formed metabolite again as an additional loss in this parent balance. In a PBPK model, the body amount and cumulative pathway amounts should be consistent with integrated compartment and elimination fluxes. For an extravascular dose, account for the amount not absorbed; do not treat the administered dose as though it all entered the systemic model.
Why the Clearance Relation Works
The clearance calculation follows from the definition of a concentration-dependent elimination rate. If renal excretion rate is \(R_{\mathrm{urine}}(t)\) and, over the interval, it is represented as \(R_{\mathrm{urine}}(t)=CL_r C_p(t)\), integrating both sides gives:
Rearranging yields \(CL_r=\Delta A_{e,\mathrm{urine}}/AUC_p\). This is a useful interval summary, not a claim that clearance is always constant. If transport is saturable, inhibition changes during the interval, or other conditions make renal handling time-dependent, the ratio is an interval-average clearance. Compare intervals or use the model’s time-varying fluxes rather than interpreting one ratio as a universal parameter.
Worked Example: Comparing a Plasma Profile
Suppose a hypothetical intravenous study reports plasma concentrations in mg/L at 0, 0.5, 2, 6, and 12 hours. The observed values are 10, 8, 5, 2, and 0.5; the renal PBPK model predicts 10, 7.6, 5.4, 2.2, and 0.4. Compare the profiles and calculate AUC from 0 to 12 hours using the linear trapezoidal rule.
At 0.5 hours, the prediction is \(100(7.6-8)/8=-5.00\%\) relative to the observation. At 2, 6, and 12 hours, the corresponding percentages are \(100(5.4-5)/5=8.00\%\), \(100(2.2-2)/2=10.0\%\), and \(100(0.4-0.5)/0.5=-20.0\%\). The last percentage is based on a small concentration, so it should not outweigh the absolute difference of 0.1 mg/L.
For the observed profile, the trapezoidal AUC is \(0.5(10+8)/2+1.5(8+5)/2+4(5+2)/2+6(2+0.5)/2\) \(=4.50+9.75+14.0+7.50=35.75\) mg·h/L.
For the predicted profile, it is \(0.5(10+7.6)/2+1.5(7.6+5.4)/2+4(5.4+2.2)/2+6(2.2+0.4)/2\) \(=4.40+9.75+15.2+7.80=37.15\) mg·h/L. The predicted AUC is \(100(37.15-35.75)/35.75=3.92\%\) higher. This is fairly close in total exposure, but the pointwise residuals show that profile agreement is not uniform. The AUC comparison alone would miss that pattern.
Worked Example: Interpreting a Renal Clearance Difference
In a hypothetical 10-hour interval, the observed unchanged urinary amount is 6.00 mg and plasma AUC is 30.0 mg·h/L. The model predicts 3.60 mg in urine over the same interval and matches that AUC. Assume \(f_u=0.100\) and \(GFR=0.120\) L/h, both constant over the interval. Calculate observed and predicted interval renal clearance, the filtration contribution, and the net contribution beyond filtration.
The observed clearance is \(6.00/30.0=0.200\) L/h. The predicted clearance is \(3.60/30.0=0.120\) L/h. The filtration clearance referenced to total plasma concentration is \(f_u\,GFR=0.100(0.120)=0.0120\) L/h. Equivalently, the filtered amount is \(0.0120(30.0)=0.360\) mg.
The observed net contribution beyond filtration is \(0.200-0.0120=0.188\) L/h, or \(6.00-0.360=5.64\) mg over this interval. The corresponding model quantity is \(0.120-0.0120=0.108\) L/h, or \(3.60-0.360=3.24\) mg. Thus, with the stated assumptions and matched AUC, filtration alone cannot explain the difference; the model has a smaller net non-filtration contribution than the data imply. This points toward checking secretion and reabsorption inputs, but it does not identify which one is wrong without further evidence.
Worked Example: Checking Parent-Drug Mass Balance
A hypothetical 100 mg intravenous dose is evaluated at 24 hours. The model reports 12.0 mg of parent drug remaining in the body, 55.0 mg of unchanged parent recovered in urine, 29.0 mg of parent-equivalent drug metabolized, and 0 mg assigned to other loss. Calculate closure and identify the discrepancy.
The accounted amount is \(12.0+55.0+29.0+0=96.0\) mg. The unaccounted amount is \(100-96.0=4.00\) mg, or \(100(4.00/100)=4.00\%\) of the dose. The balance therefore closes to 96.0% under the stated accounting. This discrepancy is not automatically a renal error: it could reflect an omitted loss pathway, a bookkeeping or unit error, or an incorrectly reported amount in any balance term. The urinary amount must also be checked against the observed collection, rather than judged from closure alone.
Diagnosing Discrepant Patterns
The combination of checks helps narrow a diagnosis, but no single pattern uniquely identifies a faulty parameter. Use independent information—such as measured GFR, plasma protein binding, urine flow, or transporter evidence—to decide whether a mechanistic adjustment is defensible.
| Observed pattern | Inputs or checks to investigate | Interpretive caution |
|---|---|---|
| Plasma exposure is too high and urinary recovery is too low | Check renal clearance, GFR and \(f_u\), secretory transporter activity, and reabsorption assumptions; also inspect nonrenal clearance. | More than one pathway can affect plasma exposure. This pattern does not isolate renal handling. |
| Plasma profile is close but urine recovery is too low | Check urinary collection and units, urinary transit or recovery assumptions, and whether secretion or reabsorption fluxes offset in the model. | A good plasma fit can result from compensating errors in renal and nonrenal pathways. |
| Urine amount is close but plasma timing is wrong | Inspect the concentration-time profile, distribution, and the timing of renal fluxes; compare interval urine amounts rather than only the cumulative endpoint. | Matching the final urinary total does not establish that the model predicts correct exposure over time. |
| Clearance appears inconsistent with filtration | Recheck units, plasma basis, interval alignment, \(f_u\), and GFR before interpreting net secretion or reabsorption. | The difference from filtration is net; it does not separately quantify secretion and reabsorption. |
| Mass balance does not close | Check dose route and absorption, amounts versus concentrations, all integrated elimination fluxes, and whether parent-equivalent terms are counted once. | Balance closure does not prove that the distribution or pathway mechanisms are correct. |
Assumptions, Alternatives, and Common Pitfalls
These checks assume that observations and predictions refer to the same drug species, population conditions, dose, and time window. Urine amounts must reflect the intended collection interval, and plasma AUC must be computed on a compatible time basis. Sparse sampling can make AUC and profile comparisons uncertain, while incomplete urine collection can imitate underprediction of renal excretion. If concentrations are below quantification limits or measurement uncertainty is substantial, interpret small differences cautiously rather than treating every residual as mechanistic evidence.
The checks are complementary to alternatives, not replacements for them. A plasma-only comparison is useful for overall systemic exposure and timing but cannot validate urinary fluxes. A urine-only comparison evaluates renal output but may conceal an incorrect plasma driving concentration. A single cumulative endpoint is simpler than interval-by-interval comparison but loses timing information. Mass balance is a bookkeeping and conservation check; it cannot tell whether a correct total was produced by the correct compartments or pathways. Jointly examining profiles, interval and cumulative urine, clearance, and balance gives a more discriminating assessment.
A common mistake is to calculate \(CL_r\) from urinary amount and AUC, subtract \(f_u\,GFR\), and label the remainder “secretion clearance.” That reasoning assumes away reabsorption. The correction is to call the remainder the net contribution beyond filtration; secretion and reabsorption can both contribute, and their separate roles require additional evidence or model checks. Another frequent error is to compare an interval urinary amount with an AUC from a different interval. The clearance ratio only has the stated interpretation when the amount and AUC cover the same period.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions below.
- Over an interval, 4.00 mg of unchanged drug is excreted and plasma AUC is 20.0 mg·h/L. What is the interval renal clearance?
- If \(f_u=0.200\) and \(GFR=0.100\) L/h, what filtration clearance is referenced to total plasma concentration? Does a larger measured renal clearance prove secretion?
- A model matches observed cumulative urine at 24 hours but predicts too little urine in the first 8 hours. What does the endpoint comparison fail to show?
- A 50 mg intravenous dose has 5 mg of parent remaining, 30 mg recovered unchanged in urine, and 12 mg accounted as parent-equivalent metabolism. What percentage of the dose is unaccounted?
Solutions. (1) \(CL_r=4.00/20.0=0.200\) L/h. (2) Filtration clearance is \(0.200(0.100)=0.0200\) L/h. A larger measured clearance shows a net contribution beyond filtration under the assumptions, but does not prove secretion alone because reabsorption and other assumptions also matter. (3) The final total hides a timing discrepancy; compare interval urinary amounts and plasma concentrations across the interval. (4) Accounted amount is \(5+30+12=47\) mg, so \(50-47=3\) mg is unaccounted, or \(100(3/50)=6.00\%\) of the dose.