Amphotericin B

Limits on what this shows

• Null and contradicting findings: With a protocol of preemptive saline plus potassium, mean potassium stayed near baseline ("potassium values remained stable over time") although grade III events still occurred [11]; this shows replacement masks the loss, not that the loss is absent. In optimally monitored South African patients hypokalemia was not associated with death (p = 0.5) [9]. The 43% vs 71% gap in [9] reflects who was monitored and is a detection effect, not a formulation effect.
• At-risk subgroups: Higher cumulative or daily dose and longer duration (total liposomal dose, treatment dose and duration all tracked with needing potassium) [8]; concurrent corticosteroids or ACTH, digitalis, or neuromuscular blockers (label); hematologic malignancy and HIV cryptococcal meningitis cohorts; children (label rates 37% vs 55% for liposomal vs conventional); renal impairment (label lists both hypo- and hyperkalemia).

Regulatory status

Conventional label [2]: hypokalemia listed under renal adverse reactions; "hypo- and hyperkalemia" among laboratory findings; "Rapid intravenous infusion has been associated with hypotension, hypokalemia, arrhythmias, and shock"; monitor "serum electrolytes (particularly magnesium and potassium)." Liposomal label [1]: same monitoring statement; says AmBisome "had a lower incidence of ... hypokalemia" than conventional; hyperkalemia is in its less common adverse event list. Corticosteroids and ACTH "may potentiate hypokalemia"; hypokalemia potentiates digitalis toxicity and the curariform effect of muscle relaxants [1][2].

Population evidence

Label trials, all-grade hypokalemia (liposomal vs comparator) [1]: Study 94-0-002 (febrile neutropenia, n = 687 [3][5][6]; liposomal 3 mg/kg vs conventional 0.6 mg/kg): 42.9% vs 50.6%; severe hypokalemia (potassium at or below 2.5 mmol/L, per the label's definition) 6.7% vs 11.6%; pediatric 37% vs 55%. Study 97-0-034 (n = 244 [3]; liposomal 3 or 5 mg/kg vs lipid complex 5 mg/kg): 37.6% and 43.2% vs 39.7%. Study 94-0-013 (liposomal 3 or 6 mg/kg vs conventional 0.7 mg/kg; the doses and comparator match the Hamill 2010 cryptococcal meningitis trial, n = 87/86/94 [7], though the label does not name the paper): 31.4% and 51.1% vs 48.3%. Conventional, South Africa [9]: prospective surveillance of 524 patients on amphotericin B deoxycholate; potassium below 3.4 mEq/L in 43% (226/524) overall and 71% (167/235) among those with optimal monitoring, of whom 5% (8/167) were grade 4 (below 2.0 mEq/L). Liposomal, Japan [8]: retrospective single-center, 93 hematology patients; potassium below 3.0 mEq/L in 53.8% (50/93); mean potassium fell from 3.8 to a minimum of 2.8 mEq/L in those given potassium and from 3.9 to 3.2 in those not (both P<0.01), recovering to 3.7 and 3.4 after the course. Lipid complex, Lebanon [10]: 89 neutropenic cancer patients; hypokalemia in 72% (64/89), 10% severe; reversible in 63%, three discontinuations. Conventional, Kenya [12]: 70 consecutive patients; 93% hypokalemic. Conventional, trial cohorts [11]: 368 patients on a protocol with preemptive saline and potassium; baseline 3.9 mmol/L, mean change small, grade III in 5.6% (16/284), grade IV in 3 patients. Before and after replacement protocol, Uganda [13]: severe hypokalemia (below 2.5 mEq/L) 38% (8/21) in 2010 vs 8.5% (12/142) after universal potassium and magnesium replacement from 2011 (RR 4.5, 95% CI 2.1 to 9.7); mean day-14 potassium 3.0 (limited fluids) vs 3.6 mEq/L. Randomized trial, conventional amphotericin B, Iran [32]: 30 completers (16 on oral N-acetylcysteine, 14 on placebo), all on amphotericin B; hypokalemia in 70% and 75% across the two arms (P = 0.724; abstract only, definitions and denominators not stated). A 2000 review [15] and a clinical review [16] describe urinary potassium and magnesium wasting.

Mechanism

Named and present: amphotericin B interacts with cholesterol in human cell membranes, which is credited for its toxicity [27]; in isolated turtle bladder it caused a large increase in potassium permeability and a defect in acidification by increased passive permeability of the luminal membrane [18]. A review cites "urinary potassium wasting" [10][16]. The exact human nephron segment was not established from the sources opened. Direction: real renal loss (depletion), not dilution (Hard Rule 5).

Formulation split

Conventional deoxycholate: 43% to 93% across cohorts [9][12]. Liposomal 3 mg/kg: lower than conventional in both label comparisons (31.4% vs 48.3%; 42.9% vs 50.6%), but liposomal 6 mg/kg (51.1%) matched conventional and the lipid complex arm (39.7%) matched liposomal 3 mg/kg [1]. Hypokalemia also occurs with liposomal and lipid complex products in absolute terms (53.8% [8]; 72% [10]).

Class cohort test

Not a class extrapolation. Every cited source studied amphotericin B itself: conventional [2][7][9][11][12][13][32], liposomal [1][7][8], lipid complex [1][10].

Counter-evidence held on record

Counter-evidence and confounding Incidence is measured in treated patients without an untreated comparator; the populations (febrile neutropenia, advanced HIV, hematologic cancer) have other causes of potassium loss, and co-therapy with corticosteroids, diarrhea and poor intake add to it. The pre/post data support causation: mean potassium fell from 3.8-4.2 at screening to 3.0 by day 14 under limited replacement [13] and from 3.8 to 2.8 mEq/L [8], and the label's dose gradient (31.4% at 3 mg/kg to 51.1% at 6 mg/kg) points the same way [1]. Report-based rates vary with how often potassium was checked [9]. Study cohorts differ in site but were not verified to be fully non-overlapping. Framing caution This is a loss caused by the drug and managed inside treatment by clinicians using laboratory values; it is not a signal to take potassium on one's own. The label also lists hyperkalemia and the patients involved often have reduced kidney function, so self-supplementation is the hazard, not the remedy. No supplement dose is given here.

Limits on what this shows

• Null and contradicting findings: No Tier A study showing no change in serum magnesium on amphotericin B was found; the lowest incidence retrieved was 20% to 30% in a 30-completer trial [32]. Baseline hypomagnesemia is plausible in these populations (baseline 0.7 mmol/L is at the usual lower reference edge [11]), and saline loading used in several cohorts can itself increase renal magnesium loss; neither explanation was tested in the sources read. The Ugandan before/after replacement study could not measure magnesium on site [13].
• At-risk subgroups: Higher dose (6 mg/kg liposomal) and longer courses (the cited 96% figure was for more than 10 days [13]). Other subgroups, including children, were not separately reported in the sources read.

Regulatory status

Conventional label [2]: "Hypomagnesemia; hypo- and hyperkalemia; hypocalcemia" under laboratory findings and monitoring of "serum electrolytes (particularly magnesium and potassium)." Liposomal label [1] repeats the monitoring statement and tabulates hypomagnesemia (below).

Population evidence

Label trials, hypomagnesemia (liposomal vs comparator) [1]: Study 94-0-002: 20.4% (liposomal 3 mg/kg) vs 25.6% (conventional 0.6 mg/kg). Study 97-0-034: 15.3% (3 mg/kg) and 25.9% (5 mg/kg) vs 15.4% (lipid complex 5 mg/kg). Study 94-0-013 (same arms as [7]): 29.1% (3 mg/kg) and 48.9% (6 mg/kg) vs 40.2% (conventional 0.7 mg/kg). Trial cohorts, serum magnesium measured [11]: 368-patient cohort on conventional amphotericin B, magnesium measured only in the South African trials, where oral magnesium was part of the protocol; baseline 0.7 mmol/L (95% CI 0.6 to 0.8), change of 0.07 lower at day 7 (95% CI 0.12 to 0.02 lower) and 0.15 lower at day 14 (95% CI 0.1 to 0.2 lower); the text calls hypomagnesemia "much more common than hypokalemia." Supplement to [11], read in the repair pass [31]: serial magnesium in 117 patients; grade III (0.3 to 0.44 mmol/L) in 29 patients (24.7% as printed) and grade IV (below 0.3 mmol/L) in 7 patients (5.9% as printed), together 36 of 117 (30.8%); the printed percentages differ slightly from the counts (29/117 is 24.8%, 7/117 is 6.0%), and the supplement gives no comparison group. Direct measurement of renal handling, USA [30]: 10 adults aged 30 to 68 with systemic fungal infections on amphotericin B; renal magnesium wasting with mild to moderate hypomagnesemia by week 2 (mean cumulative dose 208 +/- 40 mg); serum magnesium lowest and fractional magnesium excretion highest by week 4 (mean cumulative dose 510 +/- 118 mg), then a plateau; urinary magnesium rose even though less magnesium was being filtered, which points to a tubular reabsorption defect (site not determined); in 3 patients re-studied about a year after stopping, serum magnesium and fractional excretion had returned to pre-treatment levels. Abstract level only (PubMed record); the abstract gives no numeric serum or urinary values and no comparison group. Randomized trial, conventional amphotericin B, Iran [32]: 30 completers (16 on oral N-acetylcysteine, 14 on placebo), all on amphotericin B; hypomagnesemia in 20% and 30% across the two arms (P = 0.468; the summarizing tool assigned the two percentages to opposite arms in two reads, so the arm assignment is not stated here; abstract only, definitions and denominators not stated). Kenya [12]: 70 consecutive patients on amphotericin B; 80% hypomagnesemic at trough magnesium (abstract level). A review of amiloride [15] states amphotericin B can lower potassium and magnesium "to levels requiring long-term supplementation" and that amiloride may lower magnesium need by limiting renal loss (abstract only; one controlled study on potassium reported). An Australian prescriber review lists amphotericin B tubular toxicity as a renal cause of hypomagnesaemia [17]. The Ugandan cohort paper [13] cites 96% hypomagnesemia after more than 10 days of conventional therapy (secondary citation, not verified) and reports no magnesium values of its own.

Mechanism

Named: renal magnesium wasting from tubular membrane damage; a 1992 clinical review states "urinary magnesium wasting may also be present" and cites Barton et al., Am J Med 1984 [16]; that study, opened at abstract level in the repair pass, documents the loss directly (rising fractional magnesium excretion with a falling filtered load, reversible after therapy, a tubular reabsorption defect of undetermined site) [30]. The membrane-permeability evidence is shared with the potassium entry [18][27]. Direction: depletion by renal loss, not dilution.

Formulation split

Liposomal 3 mg/kg 29.1% vs conventional 40.2% (same trial), but liposomal 6 mg/kg 48.9%; in the lipid complex comparison the liposomal 3 mg/kg arm equaled the lipid complex arm (15.3% vs 15.4%) and 5 mg/kg was higher (25.9%) [1]. Dose, more than product, tracks with the rate in these tables.

Class cohort test

Not a class extrapolation. Amphotericin B was the studied drug in [1][7][11][12][30][31][32]; [15][16][17] are reviews of amphotericin B and are used only for mechanism and context.

Counter-evidence held on record

Counter-evidence and confounding Every label rate comes from treated arms without an untreated control, and the populations are critically ill. Confounding by indication is presumed and is tempered by three points: the dose gradient inside one randomized trial (29.1% to 48.9%) [1]; a within-patient decline from a stated baseline under protocolized replacement [11]; and a consistent tubular-toxicity mechanism. The strongest single number (80%) is an abstract-level observational figure [12]. In the repair pass the original wasting study (Barton 1984) was read at abstract level [30] and the grade-by-grade frequencies in the supplement of [11] were read [31]. Barton is an independent second Tier A source that measured renal magnesium handling directly (10 patients, no untreated comparator, abstract without numeric values), and the supplement adds a 30.8% grade III or IV rate under a protocol that already replaced magnesium; the red grade is retained and now rests on direct renal and serum measurements rather than serum rates alone. Framing caution This is a drug-caused renal loss that clinicians monitor on a laboratory schedule during treatment. It does not mean magnesium should be taken without supervision, and impaired kidney function raises the risk of over-replacement. No supplement dose is given here.

Limits on what this shows

• Null findings: In the rat study the amphotericin B group's disequilibrium pH and urine pCO2 did not differ from control rats (the authors class it as a gradient-type model) [19]. No human null found.
• At-risk subgroups: Not established; presumed to follow cumulative nephrotoxic exposure (label: permanent impairment especially above 5 g) [2].

Regulatory status

Conventional [2]: "renal tubular acidosis" among renal adverse reactions. Liposomal [1]: acidosis in the less common event list (2% to 10%), no percentages; no bicarbonate statement. Bicarbonate is not a dietary nutrient; this entry concerns acid-base balance.

Population evidence

None quantified in humans. Tier B: isolated turtle bladder, amphotericin B reduced hydrogen ion secretion against a gradient through increased passive permeability and was proposed as a model of the human tubular defect [18]; rat papillary collecting duct, amphotericin B was one of four experimental models of a "gradient-type" distal RTA, with disequilibrium pH and urine pCO2 not different from control rats [19]. Wikipedia names distal RTA among kidney effects (Tier D) [28]. Rat in vivo microelectrode study [35]: amphotericin B 10 mg/kg intraperitoneally for 21 to 26 days (or 2 x 10^-5 M in the luminal perfusate) doubled apparent hydrogen ion permeability and hydrogen ion back-flux in distal tubules without lowering net hydrogen ion secretion (abstract level). Case reports (Tier B, individual patients, no rate or effect estimate): a 6-year-old with coccidioidomycosis after a total dose of about 1,000 mg developed glycosuria, a high urinary pH followed by metabolic acidosis, and hypokalemia that reversed with oral potassium bicarbonate and drug withdrawal, 1981, conventional product presumed, abstract only [33]; a 29-year-old with T-cell leukemia and febrile neutropenia developed non-anion-gap metabolic acidosis within 48 hours of liposomal amphotericin B (bicarbonate 14 mmol/L, chloride 112 mmol/L, potassium 3.1 mmol/L, dose not reported) that resolved after the drug was stopped, while she was also receiving foscarnet and amikacin, a non-peer-reviewed preprint [34]. A narrative review lists amphotericin B as a cause of distal RTA by reduced renal acid excretion, with no values [36]. The classic human series (McCurdy 1968, NEJM) appeared in searches but its page errored and it was not opened.

Mechanism

Present in tissue and animal models: back-diffusion of hydrogen ions through a leaky luminal membrane [18][35]; shares the permeability mechanism with the potassium entry. Direction: acid-base disturbance from renal tubular toxicity.

Class cohort test

No class extrapolation; amphotericin B was tested directly (tissue and animal).

Counter-evidence held on record

Counter-evidence and confounding No clinical series with bicarbonate, urine pH or urinary anion gap was retrieved, so incidence and size are unknown; the two case reports show reversal after the drug was stopped, but one involved other nephrotoxic drugs and is a preprint [34]. Acidosis in critically ill patients has many causes. Near-orange: a human cohort with serum bicarbonate, urine pH or urinary anion gap in amphotericin B recipients would move it up; case reports are Tier B and cannot by themselves reach orange. Framing caution A label-listed tubular effect, not a nutrient shortfall. No alkali or bicarbonate supplement recommendation follows from this entry.

Limits on what this shows

• Null findings: Rates were equal or higher with lipid complex vs liposomal 5 mg/kg (5.1% vs 4.9%) and did not rise with dose in a consistent direction; no clear drug effect over background can be read from the tables [1].
• At-risk subgroups: Not established.

Regulatory status

Conventional [2]: hypocalcemia in laboratory findings; nephrocalcinosis among renal reactions. Liposomal [1] hypocalcemia rates: Study 94-0-002, 18.4% (liposomal) vs 20.9% (conventional); 97-0-034, 10.6% (3 mg/kg) and 4.9% (5 mg/kg) vs 5.1% (lipid complex); 94-0-013, 12.8% (3 mg/kg) and 17.0% (6 mg/kg) vs 13.8% (conventional).

Population evidence

Label tables only (Tier A statements without an untreated comparator or albumin-corrected values). No separate calcium cohort or study was retrieved. A pharmacy-retailer page asserts that such drugs "often deplete calcium and potassium" without citing amphotericin B data (Tier D) [29].

Mechanism

Not established in the sources read. Hypocalcemia could follow low magnesium, but that link was not tested here, so it is not asserted.

Class cohort test

No class extrapolation; rates come from amphotericin B arms.

Counter-evidence held on record

Counter-evidence and confounding Populations (neutropenia, chemotherapy, HIV) have high background hypocalcemia from low albumin and other drugs; confounding by indication is presumed and no baseline comparison is available (Hard Rule 6). Nephrocalcinosis is a structural renal outcome, not a calcium nutrient finding. Framing caution A context flag only. No calcium supplement recommendation follows.

Limits on what this shows

• Null findings: None retrieved.
• At-risk subgroups: Not established.

Regulatory status

Liposomal [1]: hypophosphatemia in the less common event list (2% to 10% of treated patients in chemotherapy, transplant or HIV studies), no percentages or comparator; "False elevations of serum phosphate may occur" with the PHOSm assay (for example Beckman Coulter Synchron LX20). The conventional label read has no phosphate statement [2].

Population evidence

None beyond the label list. Direction (low phosphate) would be renal loss, but that mechanism is not documented in the sources read.

Mechanism

Absent from the sources opened; plausible only as part of generalized tubular toxicity, which is not asserted.

Class cohort test

No class extrapolation.

Counter-evidence held on record

Counter-evidence and confounding A label list of events in treated patients with no rate cannot separate drug effect from critical illness, refeeding or concurrent therapy. The assay warning is a measurement caveat that can mask or mimic changes. Framing caution A listed event and a lab-assay caveat. No phosphate supplement recommendation follows.