Low-Dose Radiation Therapy for Severe COVID-19 Pneumonia: A Randomized Double-Blind Study.

PURPOSE: The morbidity and mortality of patients requiring mechanical ventilation for coronavirus disease 2019 (COVID-19) pneumonia is considerable. We studied the use of whole-lung low-dose radiation therapy (LDRT) in this patient cohort. METHODS AND MATERIALS: Patients admitted to the intensive care unit and requiring mechanical ventilation for COVID-19 pneumonia were included in this randomized double-blind study. Patients were randomized to 1 Gy whole-lung LDRT or sham irradiation (sham-RT). Treatment group allocation was concealed from patients and intensive care unit clinicians, who treated patients according to the current standard of care. Patients were followed for the primary endpoint of ventilator-free days at day 15 postintervention. Secondary endpoints included overall survival, as well as changes in oxygenation and inflammatory markers.
RESULTS: Twenty-two patients were randomized to either whole-lung LDRT or sham-RT between November and December 2020. Patients were generally elderly and comorbid, with a median age of 75 years in both arms. No difference in 15-day ventilator-free days was observed between groups (P = 1.00), with a median of 0 days (range, 0-9) in the LDRT arm and 0 days (range, 0-13) in the sham-RT arm. Overall survival at 28 days was identical at 63.6% (95% confidence interval, 40.7%-99.5%) in both arms (P = .69). Apart from a more pronounced reduction in lymphocyte counts after LDRT (P < .01), analyses of secondary endpoints revealed no significant differences between the groups.
CONCLUSIONS: Whole-lung LDRT failed to improve clinical outcomes in critically ill patients requiring mechanical ventilation for COVID-19 pneumonia.

RCT Entities:   Population Interventions Outcomes

Introduction

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the viral cause of the coronavirus disease 2019 (COVID-19) pandemic. Although symptoms of COVID-19 are mild to moderate in the vast majority of cases, some patients present with severe illness, which may quickly deteriorate to acute respiratory distress syndrome (ARDS) or end-organ failure. , The mortality of critically ill patients requiring intensive care unit (ICU) admission remains considerable, with global ICU mortality rates of 30% to 40% in most geographic regions.

Respiratory failure from ARDS is the leading cause of mortality in patients with COVID-19. Viral pneumonia can induce a hyperinflammatory syndrome characterized by a cascade of cytokine activation, overwhelming systemic inflammation, and multiorgan failure. In addition to direct viral damage, this excessive host immune response is thought to play a key role in the pathophysiology of lung injury in COVID-19, characterized by diffuse alveolar damage, inflammatory infiltrates, and microvascular thrombosis. , Evidence of hyperinflammation has led to the use of glucocorticoids in patients with severe COVID-19, an approach supported by findings of the randomized RECOVERY trial, which demonstrated that dexamethasone reduces mortality in patients requiring supplemental oxygen or invasive mechanical ventilation. Remdesivir, an antiviral agent, has also been shown to shorten the time to recovery in hospitalized patients with COVID-19 pneumonia. However, because remdesivir and other repurposed antiviral drugs appear to have little or no effect on overall mortality, the World Health Organization has recommended against their use based on results of the multinational Solidarity trial. Therefore, despite improvements in survival seen in ICU patients, the clinical management of COVID-19 remains largely supportive, and additional improvements remain desirable for critically ill patients requiring intensive care.

One novel approach that has been suggested is the use of whole-lung low-dose radiation therapy (LDRT) to treat COVID-19 pneumonia. This unconventional use of ionizing radiation is based on its anti-inflammatory and immunomodulatory effects, which have been well established in preclinical models. These effects are the likely mechanism by which low doses of x-rays were historically effective in treating various forms of pneumonia in the first half of the 20th century. Owing to the limited treatment options for COVID-19 pneumonia, several clinical trials of LDRT have been initiated, using radiation doses in the range of 0.3 to 1.5 Gy. Initial experiences have suggested that LDRT is a feasible and well-tolerated treatment that appeared to be associated with a reduction of inflammation and possibly signs of clinical improvement in these small single-arm studies.13, 14, 15, 16 However, owing to the lack of prospective controls, the efficacy of LDRT in the treatment of COVID-19 remains unknown. We therefore performed a randomized double-blind study of whole-lung LDRT in patients with severe COVID-19 pneumonia requiring mechanical ventilation.

Methods and Materials

Study design and participants

COVID-RT-01 (NCT04598581) was a randomized double-blind phase 2 trial conducted at the University Hospital of Basel in Basel, Switzerland. Patients with COVID-19–related pneumonia requiring mechanical ventilation were included. The lower age limit for male and female patients was 40 and 50 years, respectively, with exclusion of pregnancy in women of childbearing potential. No other exclusion criteria were applied. The trial was approved by the Ethics Committee of Northwestern and Central Switzerland.

Patients with a confirmed SARS-CoV-2 infection requiring intensive care were screened for eligibility by 3 subinvestigators present in the ICU. The presence of COVID-19–related pneumonia was identified based on clinical and radiological findings, including ground glass opacities and other typical characteristics observed in thoracic computed tomography (CT) imaging. , All patients were dependent on mechanical ventilation, applied using an endotracheal tube after endotracheal intubation (ETI) or using face masks for continuous or intermittent noninvasive ventilation (NIV). Informed consent was granted by the legal representatives of the patients before trial inclusion, and deferred consent was later obtained from recovering patients. The aim was to carry out the study intervention within 72 hours after onset of mechanical ventilation.

Randomization and masking

Patients were randomly assigned with a 1:1 ratio to either whole-lung LDRT or sham irradiation (sham-RT). No stratification criteria were applied. Randomization was performed on the Castor EDC platform (Castor, Hoboken, NJ), using variable block sizes of 4, 6, and 8 patients. The 2 principal investigators (PIs) in the radiation oncology department, who carried out the randomization, and 1 medical physicist who was responsible for quality assurance were aware of the group allocation. All other investigators remained blinded for the duration of the study. This included the entire ICU treatment team, who continued to treat patients in accordance with local standards, without any involvement by the 2 study PIs.

Procedures

The study procedure was performed in compliance with hospital-wide measures for infection prevention, including use of personal protective equipment, dedicated access ways, and decontamination measures. Patients were transported from the ICU to the radiation oncology department under continuous surveillance by 1 ICU physician and 1 ICU nurse. Treatments were performed after the regular daily program, using 2 beam-matched Elekta Synergy (Elekta, Stockholm, Sweden) linear accelerators (Linacs).

Two radiation oncologists, including one of the PIs, carried out the patient positioning inside the Linac vault with assistance of a radiation therapy technologist. Patients remained in their hospital beds and were positioned under the Linac gantry, with the treatment couch rotated away from the isocenter. The gantry angle was set at 0° (perpendicular to the patient), and all beam blocking leaves of the multileaf collimator were left open, corresponding to a maximum treatment field size of 40 cm × 40 cm at the isocenter. Using the field light as guidance, the position of the patients’ beds was adjusted so that the projected field covered the entire thorax at a source-to-skin distance (SSD) of 110 cm (Fig. 1 ). Treatment was performed in the supine position for patients on NIV and in the supine or prone position (depending on ICU preference) for intubated patients. After patient setup, one of the in-room cameras was directed at the ICU monitor for remote patient surveillance, and all personnel exited the treatment vault. The medical physicist, aware of the group allocation, authorized treatment with multileaf collimator leaves fully open (whole-lung LDRT) or closed (sham-RT). The radiation therapy technologist then initiated the approximately 60 seconds of beam-on time. Treatment group allocation remained invisible to the blinded personnel, for whom both interventions were completely indistinguishable.

Fig. 1

Whole-lung low-dose radiation therapy was delivered using a linear accelerator, with patients remaining in their hospital bed under continued intensive care surveillance (left image). Using the field light (green) as guidance, the position of the patients’ beds was adjusted so that the projected field covered the entire thorax up to the supraclavicular fossae (middle and right image; re-enactment using a clinician). Patients were treated at a central source-to-skin distance of 110 cm, which was verified using an optical distance indicator (red). (A color version of this figure is available at https://doi.org/10.1016/j.ijrobp.2021.02.054.)

Treatment was delivered using a single 10-MV photon beam with a tissue phantom ratio (TPR20,10) of 0.728 and a dose maximum depth at 2.1 cm. The prescribed dose was 1 Gy at the reference point, which was set to the depth of dose maximum. Prestudy treatment planning was performed using a collapsed cone algorithm in Monaco (v.5.51.02; Elekta), taking into account the estimated anteroposterior electron density configuration of an average thorax. The mean lung dose was estimated at 0.8 Gy, with a predicted lung dose range of 0.5 to 1.0 Gy based on phantom simulations. For illustration, an estimated dose distribution (calculated retrospectively using a diagnostic CT scan of an average-sized study patient) is shown in Figure E1. A set of SSD-specific monitor unit calculations was prepared, although all patients were ultimately treated using the prespecified SSD of 110 cm, which appeared adequate in all cases. We performed no patient-specific simulation, dosimetry, or image guidance, thereby limiting the unattended in-vault time to 1 to 2 minutes. After treatment, patients were transported back to the ICU, completing the study procedure within a total out-of-ICU time of less than 20 minutes.

Outcomes and statistical methods

The primary endpoint was the number of ventilator-free days (VFDs) at day 15, calculated from the day of intervention (day 0). VFDs were defined as the number of days a patient was alive and free of mechanical ventilation. Patients who died before day 28 were assigned 0 VFDs. VFDs at 15 days were assumed at 3.93 days with standard of care, based on unpublished in-house data of COVID-19 patients admitted to the ICU in spring 2020. We hypothesized that LDRT would increase VFDs to 10 days, which required randomization of 22 patients to detect superiority of LDRT with a power of 90%, a significance level of 5%, and a standard deviation of 4.36 days (based on in-house data). Exploratory secondary endpoints, based on routinely conducted measurements, included changes in Pao 2/Fio 2 ratio (Horowitz index), measured from baseline (day 0) compared with the lowest observations within 24 hours (day 1) and on subsequent days; overall survival at days 15 and 28; and levels of inflammatory markers up to day 15. Baseline comorbidity assessment was performed using the Charlson comorbidity index (CCI), based on pre-existing conditions before SARS-CoV-2 infection. The Simplified Acute Physiology Score (SAPS II), a predictor of hospital mortality, was calculated within 24 hours after ICU admission. ARDS severity was defined according to Berlin criteria based on the lowest Pao 2/Fio 2 ratio measured on treatment day (before the intervention). Continuous and categorical data were compared using 2-tailed Wilcoxon rank sum test and Fisher's exact test, respectively. A P value <.05 was considered statistically significant. Overall survival was calculated using the Kaplan-Meier method, calculated from the day of intervention. All statistical analyses were performed using RStudio (v.1.3.1093; Boston, MA).

Results

Recruitment and patient characteristics

A total of 58 SARS-CoV-2–positive patients admitted to the ICU were screened for trial inclusion in November and December 2020 (Fig. 2 ). Reasons for noninclusion were lack of consent by a legal representative (n = 14), respiratory improvement with expected weaning within 24 to 48 hours (n = 8), prolonged ventilation (>7 days) before screening (n = 7), and hemodynamic instability or death (n = 7).

Fig. 2

Enrollment, randomization, and inclusion in the primary analysis. Twenty-two patients were enrolled and randomized to either whole-lung low-dose radiation therapy (LDRT) or sham irradiation (sham-RT).

Twenty-two patients were enrolled, randomized, and treated per protocol. The characteristics of these patients are summarized in Table 1 . Patients were a median of 75 years old (range, 54-84) and had a median comorbidity index of 5 (CCI; range, 1-11) before SARS-CoV-2 infection. Most patients were male (77%), and two-thirds (64%) had a body mass index (BMI) of greater than 25 kg/m2, with a median of 26.9 kg/m2 for the entire cohort. Smoking history was common (59%), although only 1 patient was a current smoker. Patients were diagnosed with SARS-CoV-2 a median of 7 days (range, 0-16) before ICU admission, which occurred after a median hospitalization duration of 1 day (range, 0-10).

Table 1

Characteristics of 22 patients randomized to either whole-lung LDRT or sham irradiation

	Whole-lung LDRT (n = 11)	Sham-RT (n = 11)
Age, median (range), y	75 (54-84)	75 (69-82)
Sex
Male	10 (91%)	7 (64%)
Female	1 (9%)	4 (36%)
BMI (kg/m2), median (range)	24.8 (19.1-37.0)	28.0 (23.5-31.1)
Medical history (before COVID-19)
Smoking history	6 (55%)	7 (64%)
Cardiovascular disease	6 (55%)	3 (27%)
Diabetes	6 (55%)	6 (55%)
Pulmonary disease	4 (36%)	1 (9%)
Malignancy	3 (27%)	1 (9%)
CCI, median (range)	5 (1-11)	4 (3-7)
SAPS II, median (range)	52 (29-67)	43 (29-64)
ARDS severity (Berlin definition)
Mild	1 (9%)	0 (0%)
Moderate	5 (45%)	7 (64%)
Severe	5 (45%)	4 (36%)
Type of ventilation
Endotracheal intubation	8 (73%)	5 (45%)
Noninvasive ventilation	3 (27%)	6 (55%)
Ventilator days before intervention, median (range)	2 (0-7)	2 (1-5)
COVID-19 drug treatment
Dexamethasone	11 (100%)	11 (100%)
Remdesivir	4 (36%)	7 (64%)
Experimental drug	1 (9%)	2 (18%)

Abbreviations: ARDS = acute respiratory distress syndrome; BMI = body mass index; CCI = Charlson comorbidity index; COVID-19 = coronavirus disease 2019; LDRT = low-dose radiation therapy; SAPS II = Simplified Acute Physiology Score II; sham-RT = sham irradiation.

Data are presented as number (percentage) of patients unless indicated otherwise.

The patterns of COVID-19 treatment were overall similar between groups (Table 1). All patients (100%) received standard of care with dexamethasone, which was initiated a median of 3.5 days (range, 1-12) before the study intervention and given for a median of 10 days (range, 5-11) total. Remdesivir was given to 50% of patients for a median of 5 days (range, 5-6), and 3 patients (14%) received experimental drugs (canakinumab, conestat alfa) as part of ongoing clinical trials. At the time of study intervention, patients had been mechanically ventilated for a median of 2 days (range, 0-7) by way of ETI (59%) or NIV (41%). The proportion of patients managed with ETI was 73% in the LDRT group, and 45% in the sham-RT group. SAPS II scores were a median of 52 and 43, respectively, in the LDRT and sham-RT arms.

Outcomes

The study intervention occurred after a median of 0 days (range, 0-3) after patient enrollment, with most patients being treated in supine position (86%). The procedure was executed without incident in all cases, and no noticeable adverse events were observed in the hours and days after the intervention.

Individual outcomes for the primary endpoint are visualized in Figure 3 . Median VFDs at 15 days were 0 days (range, 0-9) in the LDRT group and 0 days (range, 0-13) in the sham-RT group. No significant difference in VFDs at 15 days was observed between the groups (LDRT vs sham-RT: difference 0 days, 95% CI [–4.0 to +2.0]; P = 1.00). Tracheostomy was performed in 6 patients with prolonged intubation. During the first 15 days, 9 (41%) patients were weaned from mechanical ventilation for ≥1 day, and 8 (36%) of these patients subsequently remained ventilator-free. In contrast, 0 VFDs were observed in 13 patients (59%) during the first 28 days, and only 3 (14%) of these patients were ultimately discharged to rehabilitation after prolonged hospitalization periods.

Fig. 3

Visualization of individual patient outcomes after either whole-lung low-dose radiation therapy (LDRT) or sham irradiation (sham-RT). Patients receiving mechanical ventilation by way of either endotracheal intubation (ETI; red) or noninvasive ventilation (NIV; orange) were eligible for inclusion. No difference in ventilator-free days (VFD; green) at day 15 was observed. Eleven deaths (indicated by “x”) were observed during follow-up, most of which occurred within the first 2 weeks. (A color version of this figure is available at https://doi.org/10.1016/j.ijrobp.2021.02.054.)

Kaplan-Meier plots of overall survival are shown in Figure 4 . Median follow-up was 45 days (range, 2-91). The estimate of survival at 15 days was 72.7% (95% CI, 50.6%-100%) in the LDRT group and 63.6% (95% CI, 40.7%-99.5%) in the sham-RT group. Estimates of 28-day survival were identical at 63.6% (95% CI, 40.7%-99.5%) in both arms. No difference in survival was observed between the groups (P = .69). At the time of writing, a total of 11 patients (6 in the LDRT group and 5 in the sham-RT group) had died, with most deaths occurring within the first 2 weeks after study inclusion. The primary cause of death was ARDS due to COVID-19 in all but 1 patient, who died of infectious complications secondary to malignant disease.

Fig. 4

Kaplan-Meier plot of overall survival after the study intervention (day 0). No difference in survival was observed between patients who underwent whole-lung low-dose radiation therapy (LDRT) or sham irradiation (sham-RT).

Measurements of secondary outcome parameters are summarized in Table 2 , with postintervention changes in inflammatory markers visualized in Figures E2 to E4. Measurements of secondary endpoints revealed overall similar results in both groups. Relative reductions in lymphocyte counts (baseline vs lowest measurement up to 15 days or death) were more pronounced after LDRT compared with sham-RT (P < .01), although most patients had a largely stable lymphopenia (Fig. E4). The statistical difference remained when 2 patients in the LDRT group, who had marked lymphocytosis at baseline due to known chronic lymphocytic leukemia, were excluded from the analysis (P < .01). Otherwise, no significant difference in the reduction of inflammatory markers was observed between groups. Similarly, no significant differences were seen in oxygenation changes within 24 hours (LDRT vs sham-RT: median Pao 2/Fio 2 change + 5 vs + 9, P = .49), nor in serial longitudinal measurements, which were limited owing to the low number of (alive) patients still on ventilator (data not shown).

Table 2

Summary of changes in inflammatory markers and pulmonary function (Pao2/Fio2 ratio) after LDRT and sham irradiation (sham-RT)

	Whole-lung low-dose radiation therapy						Sham irradiation						Δ
	Day 0	Day 5		Day 10		Nadir (15 d)	Day 0	Day 5		Day 10		Nadir (15 d)
CRP (mg/L), median (range)	103.6 (13.0-329.3)	73.9 (27.2-249.4)		132.8 (12.2-225.8)		36.4 (12.2-111.2)	108.9 (4.1-184.6)	74.6 (16.9-337.6)		162.0 (64.4-292.6)		32.8 (2.4-112.6)
CRP reduction (baseline to nadir)	Median –60% (range, –90% to –6%)						Median –61% (range, –96% to –11%)						P = .85
Ferritin (μg/L), median (range)	1733 (489-3846)	1557 (653-3980)		2048 (673-4739)		1045 (602-2864)	1067 (431-7251)	926 (496-3712)		1052 (541-1449)		854 (419-3712)
Ferritin reduction (baseline to nadir)	Median –24% (range, –66% to +29%)						Median –17% (range to –49%, 0%)						P = .59
Lymphocytes (109/L), median (range)	0.6 (0.4-23.0)	0.6 (0.2-11.3)		0.6 (0.2-1.4)		0.4 (1.0-2.5)	0.5 (0.3-1.3)	0.6 (0.3-1.5)		0.7 (0.3-1.7)		0.3 (0.2-1.0)
Lymphocyte reduction (baseline to nadir)	Median –66% (range, –97% to –30%)						Median –10% (range, –59% to 0%)						P < .01
	Day 0		Day 1 (24 h)		Day 5		Day 0		Day 1 (24 h)		Day 5
Pao2/Fio2 ratio, median (range)	101 (69-238)		113 (102-182)		107 (80-123)		108 (56-173)		92 (78-182)		133 (56-217)
Pao2/Fio2 ratio, change within first 24 h	Median +5 (range, –68 to +102)						Median +9 (range, –57 to +34)						P = .49

Abbreviation: CRP = C-reactive protein; LDRT = whole-lung low-dose radiation therapy; RT = radiation therapy.

Median values are based on living patients for whom observations were available. Reductions in inflammatory markers were measured as relative reduction from baseline to the lowest value observed until 15 days or death. With the exception of a more pronounced relative reduction in lymphocyte counts after LDRT, analyses of secondary outcome parameters did not reveal significant differences between the groups.

Discussion

Despite global efforts to improve outcomes in patients hospitalized with COVID-19, the mortality rates remain high in patients requiring mechanical ventilation. We now report on, to the best of our knowledge, the first randomized investigation of whole-lung LDRT in this patient population. In our study, whole-lung LDRT failed to improve VFDs compared with sham-RT, suggesting a lack of clinical benefit in critically ill patients requiring mechanical ventilation for COVID-19 pneumonia.

We used a primary endpoint of VFDs, a composite outcome measure commonly reported in ARDS trials, and we based our statistical power calculation on ICU data gathered during the early phase of the COVID-19 pandemic. The observed outcomes were worse than predicted, which is likely a consequence of patient selection for our study, which enrolled elderly and comorbid patients with poor or uncertain prognosis. In addition, changes in ICU admission practice likely increased the average case severity because noncritical patients are now routinely treated in the dedicated COVID-19 ward. We acknowledge that our initial hypothesis was optimistic in regard to the magnitude of improvement with whole-lung LDRT. However, we were looking for a clear clinical benefit that would justify potential risks for patients and staff related to the procedure and reflect the rapid symptom reversal described in historical series of x-ray therapy for pneumonia and in some early experiences of LDRT use for COVID-19 pneumonia.13, 14, 15, 16 Our study was not powered to detect small differences in outcomes, which would require a larger sample size. However, because we failed to detect any meaningful signal in primary and secondary endpoints, it appears questionable whether larger studies in similar cohorts are warranted unless more robust (preclinical) data become available.

We chose a simple approach to deliver whole-lung LDRT due to the clinical priorities in these critically ill patients. In particular, we did not perform patient-specific (CT-based) treatment planning, and we decided against using more sophisticated radiation therapy techniques, which could be used to achieve a more favorable dose distribution. Rather, we optimized our workflow to minimize the risk of unexpected events while outside the ICU. , This included treating patients in their hospital beds to eliminate the need for patient transfer and using simple setup and delivery techniques to reduce treatment times. Because we used only 1 photon beam, the prescribed dose of 1 Gy was not delivered homogeneously to the lungs. Rather, the lungs were irradiated with a spectrum of anti-inflammatory doses in the range of approximately 0.5 to 1.0 Gy, which is comparable to other ongoing studies. , Because we did not perform serial CT imaging, we were unable to quantify radiologic responses in different areas of the lung and correlate changes to the dose distribution. Furthermore, owing to the simple technique used, LDRT was also delivered to parts of the liver and spleen, as well as axillary, supraclavicular, and upper abdominal lymph nodes. The impact of irradiating these regions in the context of systemic hyperinflammation remains unknown.

The doses used in clinical trials of LDRT are highly unlikely to cause any deterministic side effects, such as radiation pneumonitis or fibrosis. However, there is a risk of radiation-induced cardiac disease and a stochastic risk of cancer induction,26, 27, 28 which led us to introduce a lower age limit in our study. The long-term risks appeared to be of little concern in our patients, considering their age and risk of early mortality from COVID-19 pneumonia. However, the ratio of risk to potential benefit appears less favorable in younger patients and in less critical patients in earlier phases of SARS-CoV-2 infection. Based on historical data, LDRT may be most effective in treating interstitial pneumonia when delivered early, and similar assumptions could be made for COVID-19 pneumonia based on the pathogenesis of ARDS. Although we enrolled patients as soon as possible after onset of mechanical ventilation, the lack of effect could be attributed to the advanced stage of COVID-19 pneumonia in our cohort. In addition, the use of dexamethasone as a standard of care could have masked anti-inflammatory effects of LDRT in our patients, most of whom already had lymphopenia at study entry, reflecting a severe clinical course. , The latter can be further complicated by an array of extrapulmonary manifestations of COVID-19, which are unlikely to be affected by LDRT and which add another dimension of complexity to these patients. Despite these complicating factors, the possible lack of a clinical effect of LDRT for any stage of SARS-CoV-2 infection has to be considered in the absence of randomized evidence.

Our results are in contrast to the initial experiences reported from small single-arm studies, which observed signs of clinical and radiographic improvement after LDRT for COVID-19 pneumonia.13, 14, 15, 16 These studies used LDRT doses in the range of 0.5 to 1.5 Gy and treated nonintubated patients receiving supplemental oxygen via nasal cannula or face mask. Due to the lack of prospective control groups and the small sample size of 2 to 10 patients per report, no conclusions can be drawn regarding the efficacy of LDRT based on these studies. Furthermore, because these studies were conducted during an earlier phase of the COVID-19 pandemic, patients generally did not receive either remdesivir or dexamethasone, the latter of which remains the only agent shown to reduce mortality in patients hospitalized with COVID-19. In our study, in which all patients received dexamethasone as a standard of care, we were unable to reproduce anecdotal evidence of rapid clinical improvement after LDRT. This included secondary outcome measures, such as improvements in Pao 2/Fio 2 ratio after 24 hours, which were not different between the groups. Whether a different outcome would have been observed in patients not requiring mechanical ventilation remains unclear because results of ongoing clinical trials are pending.

The main strengths of our study are its randomized and double-blind design, the swift accrual period of 2 months, and the patient-centered clinical endpoint. The main weaknesses are the small sample size and possibly the inclusion of only patients requiring mechanical ventilation. The latter was a consequence of our decision to focus on critically ill patients, for whom the ratio of risk to potential benefit appeared most favorable, which is relevant considering the experimental nature of the intervention. Although the baseline characteristics were overall similar in both groups, the impact of random differences has to be considered when interpreting our results. This includes a higher proportion of patients managed with ETI, and a numerically higher rate of comorbidities, in the LDRT group. The influence of these and other factors, such as the number of ventilator days before the intervention, could not be studied owing to the small sample size. However, because these small imbalances would not have changed the overall outcome of our study, we believe that future efforts would need to explore a different approach than reported here. This could, at least in theory, involve the use of a different LDRT regimen and technique or application in earlier clinical stages of COVID-19 pneumonia.

Conclusions

Whole-lung LDRT failed to improve clinical outcomes in critically ill patients requiring mechanical ventilation for COVID-19 pneumonia. Although results of ongoing studies are awaited, there is currently no role for the routine use of LDRT in this setting.