Convalescent plasma use against epidemic respiratory viruses dates back to the influenza pandemic of 1918. This intervention has persisted for over a century based on high face-validity, rather than any high-quality evidence.
The COVID epidemic reignited enthusiasm for this therapy, leading to its use in tens of thousands of patients (e.g., an expanded-use program organized by the Mayo Clinic administered plasma to >20,000 patients). Unfortunately, evidence supporting the practice has remained unimpressive. Available studies indicate that plasma isn’t beneficial among patients hospitalized due to COVID (although one study suggested a potential benefit if administered to outpatients, within 72 hours of the initial diagnosis).
The RECOVERY trial is currently reporting its results involving 11,558 patients hospitalized with COVID. This is an open-label, pragmatic, multicenter RCT. Patients randomized to the plasma group were intended to receive two units of high-titer convalescent plasma.
Patients were randomized a median of 9 days after initial COVID symptoms, which is often when respiratory deterioration emerges. At randomization, 87% of patients were receiving only oxygen therapy with or without non-invasive mechanical ventilation (with 5% of patients intubated and 8% receiving no oxygen). So patients were treated relatively early in their hospitalization, at a timepoint when convalescent plasma might be expected to be beneficial.
81% of patients randomized to convalescent plasma received two infusions (table below). Other interventions were similar between the two groups.
There were no differences in any of the endpoints, including the primary endpoint (mortality). In most cases, the endpoints were nearly identical. There was a nonsignificant trend towards reduced extubation rates among patients treated with plasma (29% vs. 36%, p = 0.07).
Subgroup analysis also detected no benefits from convalescent plasma. Specifically, among the subgroup of patients who hadn’t yet generated their own antibodies, convalescent plasma remained ineffective. Likewise, no benefit was seen among patients included within <7 days after symptoms. This suggests that high-titer convalescent plasma simply doesn’t work among hospitalized patients (regardless of whether it were given a bit earlier, or to a slightly different subgroup of patients).
Severe allergic reactions were reported in 16 patients in the convalescent plasma group compared to two patients in the usual care group. This includes nine patients with pulmonary reactions (including three deaths possibly related to transfusion). Since the study was open-label, it is conceivable that biased reporting could explain some of these differences. However, the occurrence of transfusion reactions has been noted in several studies previously, so this remains a real concern.
These findings are consistent with previous RCTs investigating the use of convalescent plasma among hospitalized patients: there is no benefit from plasma, with some signs of potential harm due to transfusion reactions. Transfusion reactions do seem to occur, but they are rare enough that they don’t swing the aggregated results of the trial towards harm. A meta-analysis of this trial combined with prior trials argues strongly against any mortality benefit:
- High-titer convalescent plasma is ineffective in COVID-19 among hospitalized patients (irrespective of timing, or the presence of antibodies against COVID).
- A small number of patients treated with convalescent plasma appear to be severely harmed, due to transfusion reactions.
- Convalescent plasma use should be restricted to the context of randomized controlled trials.