SPOTRIAS: Oxygen in Stroke Trial Stopped Prematurely

March 21, 2013

San Diego, California — The SPOTRIAS (Specialized Program of Translational Research in Acute Stroke) trial of oxygen treatment for acute ischemic stroke was stopped prematurely after 85 patients were enrolled because of a higher death rate in the oxygen group than the control group, the lead investigator reported.

The study was presented here by Aneesh Singhal, MD, Brigham and Women's Hospital, Boston, Massachusetts, at the American Academy of Neurology (AAN) 65th Annual Meeting.

He pointed out that most of the deaths in the study were due to withdrawal of care requested by the relatives and were not believed to be due to oxygen treatment.

"These were extremely ill patients who had had massive strokes, and family members decided to withdraw care," Dr. Singhal told Medscape Medical News. "We were just unlucky that more of these patients ended up in the oxygen group."

During the discussion it was suggested that the treatment group had many more comorbid conditions at baseline, which caused a randomization bias. Dr. Singhal replied, "Yes, possibly. But when only 85 patients have been enrolled there is going to be an imbalance."

"After adjusting for baseline factors there was no significant difference in death rates between the 2 groups," he added. "And apart from the deaths, there was no difference in any other adverse events between the 2 groups."

"The trial was prematurely terminated when it was barely one third through," he added in an interview. "There was no evidence of oxygen toxicity based on clinical progression, adverse events, or infarct volume. The imbalance of deaths appears to be a chance finding."

He said the data and safety monitoring board (DSMB) did the right thing to stop the trial to look more closely at the data, but he added, "I might have considered temporarily suspending the trial while looking into what was happening."

Asked whether any recommendations could be made on oxygen therapy now, he replied, "We can't recommend oxygen right now as there is too little data to analyze. And it could be challenging to convince people to study it further."

But he added that a secondary analysis showed some evidence for efficacy with oxygen and no evidence for toxicity in an MRI voxel-based analysis of "tissue fate," which he said was "an objective, automated, and statistically more powerful method to detect change."

"Bad Luck Sunk This Study"

Commenting on the study, Louis R. Caplan, MD, from Beth Israel Deaconess Medical Center, Boston, said he thought giving oxygen in the setting of stroke is a "great idea."

"It's cheap, has no known safety issues, and can be given at the scene," he told Medscape Medical News. "Bad luck sunk this study. They chose a very severe group of patients and there was an imbalance in baseline factors that led to more patients in the oxygen group having care withdrawn. This obviously led to a higher death rate in the treatment group, which made the DSMB stop the study. Perhaps a better option would have been for the DSMB to suspend the study while they investigated the cause of the higher death rate."

Presenting the study at the AAN meeting, Dr. Singhal noted that oxygen has many desirable properties: It is globally available, can be started by paramedics, is well tolerated, easily diffuses across blood–brain barrier, and has multiple mechanisms of action. "The hypothesis is that oxygen may be a simple, feasible therapy that salvages acutely ischemic brain tissue and extends the time window for tPA [tissue plasminogen activator]."

The SPOTRIAS trial had the goal of recruiting 240 patients (mostly tPA-ineligible) with symptoms of acute ischemic stroke for less than 9 hours, confirmation by imaging, and a National Institutes of Health Stroke Scale (NIHSS) score greater than 4. They were randomly assigned to normobaric oxygen or room air (control), delivered for 8 hours. NIHSS score and MRI were obtained at baseline, 4 hours (during therapy), 24 hours, 48 hours, and 3 months. Two investigators were involved for each patient: One, who was unblinded, delivered the gas, and the other, who was blinded, recorded the results.

The DSMB stopped the trial after 85 patients enrolled because of a high mortality rate in the oxygen group. This is the first presentation of the more detailed results.

The primary prespecified safety and efficacy outcomes did not significantly differ between groups.

Table 1. SPOTRIAS: Main Safety and Efficacy Endpoints

Endpoint Air Oxygen
Safety endpoint: change in NIHSS at 0 to 24 h (95% CI) 0.52 (–0.82 to 2.8) 1.65 (–0.8 to 4.1)
Efficacy endpoint: change in NIHSS at 0 to 4 h (95% CI) 0.83 (–1.2 to 2.9) 3.14 (–0.1 to 6.3)

CI = confidence interval.

Dr. Singhal reported that 17 patients died in the oxygen group versus 7 in the room air group. The mortality curves diverged very early on — in the first few days. However, on closer inspection, it was found that 15 of the 24 total deaths were due to withdrawal of care because of massive baseline infarcts or moribund admission status.

Results showed that 17 patients died in the first 20 days: 15 because of withdrawal of care, 1 because of pneumonia, and 1 because of congestive heart failure. Of the 15 withdrawal-of-care deaths, 11 were in the oxygen group and 4 in the control group.

Seven late deaths occurred: 2 due to cancer, 2 due to cardiac conditions, and 3 nursing home deaths that were said to be expected because of comorbid conditions or stroke.

Analysis of data from patients who died versus those who survived showed that those who died were older, had worse baseline NIHSS scores and infarct volumes, and had more baseline comorbid conditions.

Table 2. Mortality Analysis

Factor Dead Alive P Value
Age (y) 82 71 <.001
Baseline NIHSS score 16.3 10.5 <.001
Baseline infarct volume (mL) 78 32 .002
Charlson Comorbidity Index score 7.4 4.3 <.001
Mean prestroke modified Rankin scale score 1.24 0.43 .005


Dr. Singhal also noted that there were many baseline imbalances in the air versus oxygen groups, including the percentages women, non-Hispanic patients, those with a prestroke modified Rankin scale (mRS) score of 3 to 5, and patients with prior chronic pulmonary disease or chronic obstructive pulmonary disease (COPD).

Table 3. Baseline Characteristics by Group

Characteristic Air (%) Oxygen (%) P Value
Women 40 63 .05
Non-Hispanic patients 86 98 .05
Prestroke mRS score, 3 to 5 12 28 .10
Prior COPD 2 16 .06


In addition, prior cancer, dementia, chronic renal failure, peptic ulcer disease, and several other preexisting conditions were all more frequent in the oxygen group, although not statistically significantly so. Estimated 10-year survival rates based on the composite Charlson Comorbidity Index score were 42% in the control group and 29% in the oxygen group (P = .14).

Dr. Singhal also pointed out that 37% of control patients had occlusion of the internal carotid or middle cerebral artery versus 57% of oxygen patients (P = .08). And baseline infarct volumes and perfusion lesion volumes were approximately 20% to 30% larger in the oxygen group, although the difference was not statistically significant.

"These imbalances are not unexpected given the small number of subjects who were randomized before the trial was terminated," he commented to Medscape Medical News. "The most important point is that the logistic regression models accounting for these baseline imbalances did not find that oxygen predicted death," he said. "Moreover there was no 'pattern' of death, and there was no worsening of NIHSS scores or infarct volume growth, or intercurrent complications like pneumonia, suggesting that the withdrawal-of-care decisions were not due to immediate worsening during or after treatment."

American Academy of Neurology (AAN) 65th Annual Meeting. Abstract S02.001. Presented March 19, 2013.