Reply to Sanfilippo et al. and to Caviedes et al.

From the Authors:

We thank Sanfilippo and colleagues for their letter regarding the potential for convalescent plasma (CP) to promote macro- and microvascular thromboses in patients with coronavirus disease (COVID-19). CP is obtained through apheresis, at least 14 days following full recovery, from COVID-19 survivors who mount a satisfactory response with a high level of IgG antibodies against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) (1). Implicit in this is that, at the time of collection, 1) the inflammatory response has subsided in potential donors and 2) levels of procoagulant (e.g., factor VIII, fibrinogen, von Willebrand factor, and inflammatory cytokines) and anticoagulant (including protein C, protein S, and antithrombin) proteins have normalized. In this way, the composition of CP simulates that of standard fresh frozen plasma (FFP), except for the presence of neutralizing antibodies against SARS-CoV-2. Both standard FFP and CP contain procoagulant and anticoagulant proteins, together with fibrinolytic activators such as plasminogen. One unit of CP comprises a volume of 200–250 ml with a standard dose being a total of 2 units given 12 hours apart. A standard adult dose of FFP (namely, 10–15 ml/kg and equivalent to 4 U of FFP) should increase coagulation factor levels by around 20–30% (2). Therefore, by infusing 1 unit of CP, we expect coagulation factors to rise only by 5.0–7.5% from baseline levels, and as most of the coagulation factors have a shorter half-life—approximately 8–12 hours for factor VIII and around 16 hours for Von Willebrand factor—hypercoagulability induced by increasing the levels of these coagulation factors is unlikely to be significant. That said, clearly, this needs to be evaluated systematically through longer follow-up. At present, studies in the UK are assessing the thrombotic events up to 90 days following the infusion of CP (3). Hypothetically, CP contains neutralizing antibodies to SARS-CoV-2, which would be expected to reduce the inflammatory and prothrombotic effect of the virus on endothelium and, in so doing, might potentially downregulate any endothelial injury, thereby minimizing immunothrombosis. Although inflammation undoubtedly promotes both micro- and macrothrombosis, patients with COVID-19 seem to develop more microangiopathy as opposed to thrombotic embolism per se as described in our original work (4). Any potential impact of CP should be considered in the context of the postulated mechanisms that lead to the immune-mediated thrombosis of COVID-19 and acknowledge that the mechanisms for thrombosis in the microcirculation may differ from those seen in systemic macrothrombosis and, notably, pulmonary embolism.

We also thank Caviedes and colleagues for their letter discussing the calculations of dead space ventilation. We agree that ventilatory ratio (VR) does not consider CO2 production and that the Enghoff equation [(PaCO – PeCO)/PaCO] (where PeCO represents partial pressure of expired carbon dioxide) probably reflects the dynamics more accurately. Indeed, although all these formulae have been validated in a number of settings to relate to dead space, it is important to note that shunting increases not only the alveolar–arterial O2 gradient but also the arterial–alveolar CO2 difference and, because the Enghoff dead space assumes PaCO as a surrogate for PaCO, this may increase calculated physiological dead space. Needless to say, calculation of dead space cannot be performed if an artificial membrane lung is used during extracorporeal support. For clarity, we performed computed tomography (CT) scans during the pandemic for patients transferred on extracorporeal membrane oxygenation (ECMO) on admission, and calculations of VR were taken using the last ventilation settings and matched blood gases prior to the patient being placed on ECMO. In non-ECMO patients, ventilator settings used were immediately before CT scanning. However, it is important to note that our study was not a prospective physiological study and, accordingly, we did not capture PeCO. Instead, we retrospectively obtained temporally matched end-tidal capnography from 20 patients within our cohort in whom there was a significant correlation between VR and Enghoff Vdphys (physiologic dead space, which is the sum of the anatomic and alveolar dead space)/Vt (Figure 1: P = 0.0208; Spearman r = 0.513). Of note, these PeCO readings were captured without analysis of capnograph waveforms and, so, the actual reading may not accurately reflect the proportion of dead space. Nonetheless, a correlation between Enghoff and VR suggests a relationship between these dead space measurements between individuals. We also note the authors’ own data indicating the strong correlation with e/co2 in moderate and severe acute respiratory distress syndrome. We did not have the opportunity to test this index, as we did not measure co2. However, if it were possible to accurately measure volumetric capnography, we would have advocated that Enghoff Vdphys/Vt be calculated using PaCO and PeCO rather than any of its surrogate indices. Indeed, in our center, arterial blood gases are frequently measured and are not an operational overreach. Our reporting of physiology was a pragmatic approach to benchmark our observations to other reports for COVID-19 (5) and to facilitate discussion of the vascular abnormalities seen on imaging in patients with severe COVID-19.

Figure 1.

Ventilatory ratio and dead space as calculated with the Enghoff equation shows a good correlation in 20 patients with severe coronavirus disease (COVID-19) (Spearman r = 0.513, P = 0.0208). Enghoff Vdphys/Vt = physiologic dead space, which is the sum of the anatomic and alveolar dead space/Vt.

In summary, COVID-19 has a significant vascular inflammatory component, which, in addition to the dynamic immune response (6), activates the coagulation cascade. The early onset of angiogenesis in COVID-19 pathogenesis has yet to be confirmed. However, our observations from dual-energy CT (DECT) imaging and those of others have revealed perfusion defects even in patients with mild (non-ICU) COVID-19 (7). Hence, CT findings (such as vascular tree-in-bud or DECT perfusion abnormalities) might provide a vital noninvasive “window” to better highlight disease progression and/or ascertain the beneficial or deleterious treatment responses to therapeutic strategies such as CP. In this regard, we have recently shown that perfusion defects on DECT imaging tend to resolve over time in COVID-19 (8), an observation that might influence the nature and direction of ongoing research efforts. Finally, careful monitoring (including radiological evaluation) of patients for both short- and long-term complications of CP in well-controlled randomized clinical studies is of utmost importance. The understanding of disease pathogenesis is key to the personalized application of therapies, and prospective validation of accurate bedside measures of dead space ventilation could present opportunities to target angiopathy through physiological enrichment in clinical trials.