Reply to “Hypokalaemia probably associated with ceftolozane/tazobactam treatment: Three case reports”

Dear Editor,

We read with great interest the article by Toro Blanch et al.,¹ describing three patients with hypokalaemia probably secondary to treatment with ceftolozane/tazobactam which, according to its summary of product characteristics,² is a common adverse effect (≥1/100 to <1/10). Alterations in potassium metabolism are among the most common metabolic disorders in clinical practice. Both hyperkalaemia and hypokalaemia can give rise to different clinical manifestations, which can even lead to death. As such, a systematic approach to its diagnostic procedure is needed.³

Beta-lactam antibiotics can cause hypokalaemia owing to the presence of nonresorbable anions, which leads to increased K⁺ and H⁺ excretion, giving rise to metabolic alkalosis. The presence of a β-lactam ring, a feature shared by beta-lactam antibiotics, has implicated other antibiotics from this group in the genesis of hypokalaemia. Brunner and Frick reported the development of hypokalaemia in nine patients who were being treated with 100 mega-units of penicillin G sodium for subacute bacterial endocarditis.⁴ However, it has also been reported with the use of ampicillin⁵ and carbapenems, such as meropenem.^6–8 As Bharti et al.⁸ acknowledge in their communication, these cases, and those reported by Toro Blanch et al., require urine K⁺ levels to be determined in order to distinguish between renal and non-renal causes for hypokalaemia. In hypokalaemia secondary to beta-lactam antibiotics, the underlying mechanism is an increase in the renal elimination of potassium, whereby urinary potassium in an isolated sample would be greater than 15 mmol/l and the trans-tubular potassium gradient would be lower than 4 (absence of mineralocorticoid activity in the distal convoluted tubule).

We agree with the authors on the degree of assignability, according to the Naranjo algorithm, of probable in the first two cases and certain in the final case. Nonetheless, it should be stressed that without reintroducing the drug (ethically unacceptable in the majority of cases) a proven or certain causal link cannot be established. That is why in cases of probable causality, the presence of high urine potassium with a trans-tubular potassium gradient under 4 would be more supportive of the diagnosis.

This being so, and without challenging the diagnosis, it would have been important to ascertain the level of potassium in an isolated urine sample and, in the event that this were greater than 15 mmol/l, to calculate the trans-tubular potassium gradient. Only by being aware of the aforesaid data would it have been possible, in the first two cases, to rule out other possibilities, such as extra-renal losses due to profuse sweating in the context of septic shock, gastrointestinal causes (fistulas, drainage, etc.) or pseudohypokalaemia secondary to extreme leukocytosis.

Notwithstanding the foregoing, we would like to congratulate the authors for the interesting nature of the cases and for the contributions of their work.