In regard to Hess et al.

To the Editor

We would like to comment on the recent publication by Hess et al. on the use of whole-lung low-dose radiation therapy (LDRT) in non-intubated oxygen-dependent patients with COVID-19-related pneumonia [1]. We congratulate the authors on the completion of their study, knowing first-hand the challenges of safely evaluating this approach during a global health crisis.

The study enrolled 20 patients that were treated with LDRT, and retroactively matched them to 20 controls that had received standard treatment. It is therefore a single-arm study of 20 patients with a retrospective comparative analysis, for which the presentation of a CONSORT flow diagram appears misplaced. More importantly, however, the authors fail to address a significant difference between the cohorts: 95% of patients in the LDRT group received dexamethasone, compared to only 50% of the controls. Dexamethasone reduces mortality, hospitalization duration, and rates of mechanical ventilation in oxygen-dependent patients, the cohort studied here [2]. Patients in the LDRT group started COVID-19 drugs on median hospital day 1, followed by LDRT on median hospital day 3. This overlaps with the time window during which a reduction in C-reactive protein (CRP) occurs as a sign of response to corticosteroid therapy in COVID-19 [3]. It is therefore difficult, if not impossible, to separate potential clinical and serological effects of LDRT from dexamethasone in this study.

The authors classified 16 of 20 irradiated patients as CRP responders, and compared them with their 16 matched controls. This comparison is invalid, since the exclusion of 4 CRP non-responders from the LDRT group introduces a significant bias. Indeed, 3 of 4 excluded patients required intubation by day 28 (refused by one patient), and all 4 patients had died by day 33. When comparing the entire cohorts, only a trend towards improved freedom from intubation was observed with LDRT. This analysis did unfortunately not account for the competing risk of death, meaning that the patient who refused intubation and died on day 11 was censored, favoring the LDRT group. Furthermore, it is unclear why additional patients were censored in the freedom from intubation (Figure 2) and intubation-free survival (Figure 3d) analyses. We assume that these were hospital discharges, in which case the patients should not have been censored, but followed-up until day 28 for these endpoints.

Hess et al. comment on our recent study, which randomized 22 patients receiving mechanical ventilation for COVID-19 pneumonia to either whole-lung LDRT or sham irradiation [4]. We agree that our study was not powered to detect small differences in outcomes, as acknowledged in the publication. However, the notion that our study had less than 1% of the power needed to detect a benefit equal to that of dexamethasone is incorrect, not least due to the fact that we used a different endpoint [2], [4]. We wish Hess et al. well in their endeavour to randomize 150 patients in a phase III trial. Vaccines and novel antiviral drugs have risen the bar for LDRT, and only meticulous clinical studies, devoid of any bias, will be able to truly define its value.