LOVIT or leave it: The vitamin C debate continues

To the Editor,

There has been an increasing interest in the use of high doses of vitamin C (l-ascorbic acid) as a treatment for sepsis due to its potentially beneficial effects as an antioxidant, immune stimulant, anticoagulant and a cofactor for the synthesis of noradrenaline and vasopressin^1,2. Numerous clinical trials have examined the responses to doses of vitamin C between 25 and 200 mg/kg/day for 3–4 days with disparate results, ranging from mild signals of benefit,[3], [4], [5] to no effect.[6], [7], [8] The recently published LOVIT trial is the largest trial to investigate high-dose intravenous l-ascorbic acid in critically ill adults with sepsis.⁹ This phase III multi-centre, placebo-controlled trial, including 863 patients, reported an unadjusted increase in mortality and persistent organ dysfunction at day 28 risk ratio (RR) 1.21 [95% confidence interval (CI); 1.04–1.40] in the group treated with intravenous ascorbic acid (200 mg/kg/day for 4 days) compared with placebo.⁹ There was, however, no difference between the groups after adjusting for age, sex, severity of illness, steroid use, and time to randomisation RR 1.15 [95% CI; 0.90–1.47].⁹ A concurrently published meta-analysis by the LOVIT authors of 34 placebo-controlled clinical trials of vitamin C in sepsis in nearly 5000 patients reported point estimates favouring vitamin C in reducing in-hospital mortality RR 0.88 [95% CI; 0.73–1.06], 30-day mortality RR 0.83 [95%; CI 0.71–0.98] and early mortality RR 0.80 [95% CI; 0.68–0.93].¹⁰ Secondary analysis of the LOVIT trial⁹ reported that the increased mortality in the vitamin C group may not be a consequence of ongoing ascorbic acid infusion, but rather due to the abrupt termination of vitamin C therapy.¹¹ It was proposed that the increased metabolism of vitamin C in septic patients may cause a rebound reduction of plasma ascorbate to levels even lower than pre-treatment levels.¹¹ Similar findings were also observed in the CITRIS-ALI trial¹²

The biochemical composition of vitamin C used in the LOVIT trial may have also contributed to the putative signal for harm. A preclinical study in a large mammalian model of Gram-negative sepsis using intravenous sodium ascorbate (3.75 g/kg/7-h), at a substantially higher dose than that of l-ascorbic acid used in the LOVIT trial (0.2 g/kg/day), reported dramatic improvements in clinical state, withdrawal from vasopressors and reversal of organ injury.¹³ From a Stewart acid-base perspective,¹⁴ comparing intravenous administration of sodium ascorbate with ascorbic acid has parallels with comparing sodium acetate with acetic acid (vinegar). Administering ascorbate strong ions (ascorbic acid) would be expected to produce metabolic acidosis in vivo,¹⁵ with a wide anion gap and a strong ion difference. In contrast, sodium ascorbate is conceivably less acidifying in plasma due to the accompanying sodium cations. This has relevance in critically ill patients with septic shock in whom metabolic acidosis and diminished buffering capacity are ubiquitous and independently associated with mortality.¹⁶

Here, we report that l-ascorbic acid has a temperature-dependent acidic pH range (5.4–5.6) either as a stock solution or when diluted 1:1 in 5% dextrose, as was administered to septic patients in the LOVIT trial.⁹ In contrast, sodium ascorbate demonstrates a relatively acid-base balanced pH range (6.5–7.2) (Fig. 1A). We further demonstrate a stable arterial blood pH over time in five sheep with experimental Gram-negative sepsis who received intravenous sodium ascorbate (3 g/kg) compared with an overt reduction in pH in a septic sheep treated with a similar dosage of ascorbic acid (Fig. 1B). It has been long recognised in translational research that use of ascorbic acid is one way to induce metabolic acidosis and acidaemia¹⁷ and our results in ovine Gram-negative sepsis agree with these previous findings. These observations raise an important question of whether treating critically ill patients, who have compromised buffering capacities, with titratable acidic formulations is harmful. Future studies are warranted to investigate different compositions of vitamin C products, with appropriate consideration of acid-base balance and strong ion differences, that may explain the mixed findings in clinical trials that have universally, as far as we are aware, utilised l-ascorbic acid.

Fig. 1.

Comparison of the pH of l-ascorbic acid and sodium ascorbate in vitro and in vivo following intravenous infusion in ovine Gram-negative sepsis. (A) Temperature-dependent pH range of ascorbic acid versus sodium ascorbate as a stock solution or as a 1:1 dilution in 5% (w/v) dextrose. The pH values of the two vitamin C solutions were determined and compared using a combination glass pH electrode at the indicated temperatures. (B) Changes in arterial blood pH observed in septic sheep administered either sodium ascorbate (n = 5) or ascorbic acid (n = 1).