Simulating dialyzability across the blood–dialysate countercurrent circuit — molecular weight, protein binding, and volume of distribution as the physicochemical gatekeepers of extracorporeal drug clearance
Hemodialysis moves blood outside the body through a semipermeable hollow-fiber membrane bathed in dialysate flowing in the opposite direction. This countercurrent geometry is the single design decision that makes modern dialysis efficient: it keeps the diffusion gradient favorable across the entire length of every fiber, rather than letting blood and dialysate equilibrate partway through the exchange, as would happen with co-current (parallel) flow.
Vascular access is the practical bottleneck of dialysis delivery:
• Arteriovenous (AV) fistula: surgical connection of artery to vein, matured over 6–8 weeks; lowest infection and thrombosis risk; preferred long-term access, supports Qb of 400–500 mL/min. • AV graft: synthetic conduit between artery and vein when native vessels are inadequate; usable sooner than a fistula (2–3 weeks) but higher thrombosis/infection rates. • Central venous catheter (tunneled or non-tunneled): immediate access, no maturation time, but highest infection risk and lower achievable blood flow (~250–350 mL/min); used when fistula/graft is not yet ready or not feasible.
Blood path: access → arterial line → blood pump (peristaltic, sets Qb) → dialyzer inlet → thousands of parallel hollow fibers (inner diameter ~200 μm, wall thickness ~30–50 μm) → dialyzer outlet → venous line → return to patient. Heparin (or citrate for regional anticoagulation) is infused pre-filter to prevent clotting within the extracorporeal circuit.
Fick's first law states solute flux across the membrane is proportional to the local concentration gradient: J = -D·A·(dC/dx). In co-current flow, blood and dialysate move the same direction, so the gradient is large at the inlet but collapses toward zero as the two streams approach equilibrium near the outlet — clearance drops off sharply along the fiber length.
In countercurrent flow, fresh (solute-free) dialysate meets blood that has already been substantially cleared near the dialyzer outlet, while fresh, solute-rich blood meets dialysate that is already partially saturated near the inlet. The net effect is that a favorable concentration gradient is sustained across nearly the entire fiber length, roughly 10–20% higher clearance than an equivalent co-current design for the same membrane area and flow rates. This single architectural choice is why every clinical dialyzer built since the 1960s uses countercurrent flow.
Because clearance is flow-limited for small solutes, increasing blood flow (Qb) or dialysate flow (Qd) increases clearance only up to the point where diffusion across the membrane itself becomes rate-limiting (KoA, the mass transfer-area coefficient, plateaus). Beyond that point, a larger or higher-flux membrane — not a faster pump — is what improves clearance further.
Every dialysis membrane has a characteristic pore-size distribution that determines its sieving coefficient — the fraction of a solute at a given molecular weight that can pass through per unit time. Small solutes diffuse almost freely; middle molecules are removed slowly and incompletely; molecules above the membrane's cutoff, including nearly all plasma proteins, are essentially retained in blood regardless of the concentration gradient.
Sieving coefficient (SC) is defined as the ratio of solute concentration in the dialysate ultrafiltrate to that in plasma water; SC=1 means the solute crosses as freely as water, SC=0 means complete rejection. SC falls off steeply as molecular radius approaches the effective pore radius of the membrane.
Low-flux membranes (regenerated cellulose, early synthetic polymers): pore radius rejects most solutes above ~5,000 Da; adequate for small solutes (urea 60 Da, creatinine 113 Da, lithium 7 Da, most aminoglycosides 400–600 Da) but poor for middle molecules.
High-flux membranes (polysulfone, polyethersulfone, polyamide): larger, more uniform pores allow passage of solutes up to ~20,000–30,000 Da, enabling clinically useful removal of middle molecules such as vancomycin (1,449 Da, ~30–40% removed per high-flux session) and β2-microglobulin (11,800 Da, a marker of middle-molecule clearance adequacy).
Above roughly 40,000–50,000 Da — essentially all plasma proteins including albumin (66,500 Da) and protein-drug complexes — standard hemodialysis provides negligible clearance. This is precisely why protein binding (Stage 3) matters even for small drugs: a small molecule bound to albumin behaves, for filtration purposes, like a 66,500 Da protein.
A drug circulating in plasma exists in two pools: free (unbound) drug dissolved in plasma water, and drug reversibly bound to plasma proteins — chiefly albumin for acidic drugs, alpha-1 acid glycoprotein for basic drugs. Only the free fraction is small enough and mobile enough to diffuse across the dialysis membrane at a clinically meaningful rate; the protein-bound fraction is retained in blood as if it were the protein itself.
The free fraction (fu) is the pharmacologically and dialytically relevant quantity: fu = C_free / C_total. For a drug that is 99% protein bound, fu = 0.01 — only 1% of total plasma drug is available to cross the membrane at any instant. Even though bound and free drug exist in rapid equilibrium (binding kon/koff typically on the order of milliseconds), the equilibrium re-establishes fast enough that dialysis clearance is governed by the free fraction at steady state, not by some slow bottleneck of dissociation.
Dialyzability correlates inversely with % protein binding almost independent of molecular weight, once weight is below the membrane cutoff:
• fu near 1 (low binding, e.g. lithium, aminoglycosides, most beta-lactams <40% bound): high dialyzability, 50–90% removed per session. • fu 0.1–0.5 (moderate binding, e.g. vancomycin ~30–55% bound): partial dialyzability, compounded further by its middle molecular weight. • fu <0.05 (high binding >95%, e.g. warfarin >99%, most NSAIDs >95%, diazepam ~98%): negligible dialyzability regardless of the drug's small size — the effective "molecule" crossing the membrane is a rare, fleeting free monomer.
A clinically important caveat: in uremia, protein binding of many acidic drugs decreases (competitive displacement by accumulated uremic toxins, hypoalbuminemia), which can transiently raise fu and therefore dialyzability compared with the same drug in a patient with normal renal function.
Warfarin is the textbook example of a small, protein-bound drug that is clinically NOT dialyzable: at 308 Da it easily fits through any dialysis membrane pore, but at >99% protein binding, dialysis removes under 5% of total body warfarin per session — no dose adjustment around hemodialysis timing is needed for warfarin itself, even though the patient may be on dialysis for unrelated renal failure.
Volume of distribution (Vd) describes how a drug partitions between the blood/extracellular compartment sampled by the dialyzer and the much larger tissue compartment it cannot reach. A drug with a small Vd stays largely confined to plasma and extracellular fluid, so most of the body's drug burden is presented to the dialyzer over the course of a session. A drug with a large Vd is mostly sequestered in fat, muscle, or intracellular space — the dialyzer can only clear what briefly re-equilibrates back into blood.
Vd relates total body drug amount to plasma concentration: Vd = Dose / C_plasma. A large apparent Vd does not mean the drug is physically diluted in a huge volume of fluid — it means most of the drug has left plasma and bound to tissue components (lipid, intracellular protein, receptors), so a small plasma concentration corresponds to a large total-body amount.
Digoxin illustrates the principle cleanly: molecular weight 781 Da sits comfortably in the middle-molecule range, and protein binding is only modest (~20–30%) — by size and binding alone it looks dialyzable. But Vd ≈ 500 L (roughly 7 L/kg, far exceeding total body water) reflects extensive tissue binding, particularly to Na+/K+-ATPase in skeletal and cardiac muscle. At any moment, well under 1% of total body digoxin is in the plasma pool the dialyzer samples.
The rebound phenomenon follows directly from Vd/compartment kinetics: even when a session measurably lowers plasma concentration of a high-Vd drug, tissue stores re-equilibrate into blood over the following hours, and plasma concentration partially rebounds toward pre-dialysis levels. This is why apparent post-dialysis clearance can overstate true whole-body drug removal for drugs like digoxin, phenytoin (Vd ~0.6–0.8 L/kg, borderline), and especially lipophilic tricyclics — a second, delayed post-dialysis level is often checked clinically to confirm the true nadir.
A drug can be small and only lightly protein-bound and still be clinically non-dialyzable if Vd is large — which is exactly the case for digoxin toxicity: hemodialysis and hemoperfusion are considered ineffective, and treatment of severe digoxin toxicity instead relies on digoxin-specific antibody fragments (Fab), not extracorporeal removal.
Once dialyzability is characterized, dosing regimens for patients on maintenance hemodialysis must account for the drug removed during each session. For dialyzable drugs, standard practice is to administer a supplemental dose after dialysis to replace what was filtered out; for non-dialyzable drugs, dosing is unaffected by the dialysis schedule and is based purely on residual renal function and interdialytic accumulation.
The dialyzability estimate — driven by molecular weight, protein binding, and Vd together — translates directly into three dosing decisions:
1. Timing relative to dialysis: for a dialyzable drug, giving the dose before a session wastes a large fraction of it into the dialysate; the standard approach is to dose AFTER the session ends, so the full dose is available during the interdialytic interval. For a non-dialyzable drug (e.g. warfarin, digoxin), timing relative to dialysis is irrelevant — dosing follows the drug's own half-life and the patient's residual renal clearance.
2. Supplemental replacement dose: for antibiotics with narrow therapeutic windows and significant dialyzability (vancomycin, many aminoglycosides, several cephalosporins), a post-dialysis supplemental dose approximates the amount removed, often calculated as (pre-HD level) × (fraction removed) or via published dialysis dosing nomograms rather than empiric guessing — under-dosing risks therapeutic failure or resistance, over-dosing risks accumulation between sessions.
3. Membrane and modality matter: high-flux membranes and longer or more frequent sessions (e.g. nocturnal 6–8 hr dialysis) remove middle molecules substantially more than conventional low-flux, 3–4 hr thrice-weekly schedules — the same drug can require different supplemental dosing depending on which dialyzer and prescription the dialysis unit uses. Extended-release oral formulations add another layer: absorption may still be ongoing during a session, so the "amount available to be removed" is not simply the administered dose.
Vancomycin dosing on hemodialysis is the clinical workhorse example: a loading dose is given, then levels are checked and a maintenance dose is administered after each high-flux session (roughly every 48–72 hours) rather than on a fixed daily schedule — dosing interval is set by the dialysis calendar, not the clock, precisely because a meaningful fraction of circulating vancomycin is filtered out each time.