Susan Jeffrey

November 21, 2014

CHICAGO, IL — Results of a new trial suggest supplemental oxygen therapy in patients with ST-elevation MI (STEMI) may actually be harmful for patients who are not hypoxic[1].

The Air Versus Oxygen in ST-Elevation Myocardial Infarction (AVOID) trial compared supplemental oxygen vs no oxygen unless O2 fell below 94%.

"The AVOID study found that in patients with ST-elevation myocardial infarction who were not hypoxic, there was this suggestion that, potentially, oxygen is increasing myocardial injury, recurrent myocardial infarction, and major cardiac arrhythmia and may be associated with greater infarct size at 6 months," lead author Dr Dion Stub (St Paul's Hospital, Vancouver, BC, and the Baker IDI Heart and Diabetes Institute, Melbourne, Australia) concluded.

"These findings certainly need to be confirmed in larger randomized trials that are powered for hard clinical end points, but the AVOID study investigators would really question the current practice of giving oxygen to all patients and certainly to those who have normal oxygen levels to begin with," he concluded.

The results were presented here at the American Heart Association (AHA) 2014 Scientific Sessions.

Effects on Infarct

Following the first report in 1900 of supplemental oxygen relieving angina, pre- and in-hospital oxygen "has really been a fundamental component of first-aid management of patients with suspected acute myocardial infarction, and this is done the world over," Stub said. International guidelines differ on who should be given oxygen, he noted, "but all guidelines recognize that this fundamental of practice has very limited evidence behind it in a randomized clinical-trial fashion."

However, recent physiologic data suggest that even as little as 15 minutes of oxygen can cause hyperoxia, leading to a reduction in coronary blood flow, increased coronary vascular resistance, increased oxygen free radicals, and disturbed microcirculation, he said, "and this all may contribute to increased reperfusion injury, myocardial injury during acute coronary syndromes."

AVOID was an investigator-initiated randomized, controlled, multicenter trial with the aim of comparing supplemental oxygen therapy with no oxygen in STEMI patients with oxygen saturation in the normal range.

It was a "pragmatic" trial coordinated by the research division of Ambulance Victoria, in conjunction with nine tertiary-care centers in Melbourne, Australia. "A key component of the trial is that all patients were randomized by the paramedics prehospital," he noted.

Patients were included if they had symptoms suspicious of MI for less than 12 hours, normal oxygen levels, defined as O2sat>94% measured by pulse oximeter, and a diagnostic prehospital ECG with ST-elevation on two or more contiguous leads. Patients were excluded if oxygen saturation was below 94%, they were in an altered conscious state, they received oxygen prior to randomization, or there was planned transport to a nonstudy hospital.

Patients with confirmed STEMI randomized to the oxygen arm (n=218) received 8 L/min of O2 given prehospital right through to admission to the coronary cath lab for primary angioplasty and until they were stable on the ward. Patients in the no-oxygen arm (n=223) were given no oxygen unless they became hypoxic (O2sat<94%).

The co–primary end point was myocardial infarct size based on cardiac enzymes and other markers (mean peak creatinine kinase, mean peak troponin I, area under the curve of creatinine kinase and troponin I).

Patient characteristics in the groups were well-matched. The oxygen was appropriately administered, with 99.5% receiving oxygen via face mask in the prehospital and in-hospital setting, he said. In the no-oxygen group, 4.5% received oxygen prehospital for low oxygen saturations, 7.7% were treated during the procedure in the cath lab, and over 20% while on the ward.

This resulted in significant differences in oxygen saturation throughout the study, Stub noted. Cardiac arrest and cardiogenic shock occurred similarly between groups. Time from paramedic arrival on the scene to hospital arrival was approximately 55 minutes in both groups. Interestingly, they found no indication of symptomatic benefits of oxygen, with pain scores and administration of analgesics also similar in both groups, he pointed out. Details of the procedures were also not different between the study arms.

On the primary end point, they found a significant 25% increase in creatine kinase (CK)—"so the suggestion of increased myocardial injury in those delivered oxygen," Stub noted.

For the troponin I co–primary end point, the curves were similar, but with wider confidence intervals. "We had one site that had issues early in the trial with regard to ascertaining the standardized troponin assay, so the confidence intervals were a little wider, and this was a nonsignificant result," Stub said. "So on the one hand, you had a highly significant CK result, with a nonsignificant troponin."

AVOID Primary End Point: Infarct Size on Cardiac Enzymes

Measure Oxygen No oxygen Ratio of means (oxygen/no oxygen) P
Creatinine kinase (U/L)
Geometric mean peak (95% CI) 1948 (1721–2205) 1543 (1341–1776) 1.26 (1.05–1.52) 0.01
Median peak (IQR) 2073 (1065, 3753) 1727 (737, 3598) 0.04
Troponin I (µg/L)
Median peak (IQR) 65.7 (30.1, 145.1) 62.1 (19.2, 144.0) 0.17
Geometric mean peak (95% CI ) 57.4 (48.0–68.6) 48.0 (39.6–58.1) 1.20 (0.92–1.55) 0.18
IQR=interquartile range

A secondary end point was cardiac MRI (CMR), "the gold standard of final infarct size," which was done in a subgroup of 135 patients at 6 months, he said.

The suggestion of increased infarct size and myocardial injury was again seen. "When we looked at late gadolinium enhancement, there was a significant difference between the oxygen and the no-oxygen group. When this was normalized for left ventricular mass, it was just a nonsignificant trend," Stub added, but it still suggests the potential for increased myocardial injury.

AVOID: Infarct Size on CMR

Infarct Size Oxygen No Oxygen Ratio of means (oxygen/no oxygen P
Median (IQR, g) 20.3 (9.6, 29.6) 13.1 (5.2, 23.6)   0.04
Geometric mean (95% CI), g 14.6 (11.3–18.8) 10.2 (7.7–13.4) 1.43 (0.99–2.07) 0.06
Median (IQR) proportion of LV mass 12.6 (6.7, 19.2) 9.0 (4.1, 16.3)   0.08
Geometric mean (95% CI) proportion of LV mass 10.0 (8.1–12.5) 7.3 (5.7–9.3) 1.38 (0.99–1.92) 0.06
IQR=interquartile range

The study wasn't powered to look at major adverse cardiac events, he noted. Mortality was similar between the groups, but significant increases were seen in recurrent MI and in significant arrhythmias in the oxygen group. No significant differences were seen at 6 months on the clinical end points, Stub said.

AVOID: Clinical End Points

End point Oxygen (%) No oxygen (%) p
Hospital discharge
Mortality 1.8 4.5 0.11
Recurrent MI 5.5 0.9 <.01
Stroke 1.4 0.4 0.30
Major bleeding 4.1 2.7 0.41
Significant arrhythmia 40.4 31.4 0.05
6 mo
Mortality 3.8 5.9 0.32
Recurrent MI 7.6 3.6 0.07
Stroke 2.4 1.4 0.43
Repeat revascularization 11.0 7.2 0.17

"Hypothesis-generating" subgroup analysis showed that females, those with the longest symptom-to-intervention times (>180 minutes) and preintervention TIMI 2 or 3 blood flow all favored no oxygen therapy.

During discussion of the trial at a press conference, Stub stressed that these patients were normoxic. "We certainly are not proposing that all patients should not be given oxygen," he said. "Clearly if you're having an acute coronary syndrome or any condition in which you're hypoxic, then oxygen clearly is a lifesaving drug."

Stub noted that this issue is being further studied by researchers with the Swedish Coronary Angiography and Angioplasty Registry (SCAAR), a national registry of consecutive patients from 29 hospitals in Sweden where angiography and PCIs are performed. The registry was established in 1989 and is independent of funding from industry.

"I think this will be a fantastic study," Stub told heartwire , and at about 5000 patients, it will have the power to look at clinical end points like mortality. Results are expected to be available in a couple of years.

Dr Robert Harrington (Stanford University, CA), chair of the program committee for this year's AHA Scientific Sessions, who moderated the press conference, pointed out the SCAAR group conducted the recent TASTE trial of thrombectomy in MI.

"They're now doing a series of trials, and one of them is a randomized trial of oxygen and mortality as the primary outcome, so kudos," he said. "They have a system where they can do this very readily."

Breaking Up With MONA?

The invited discussant for this trial was Dr Karl B Kern (University of Arizona, Tucson), who pointed out that all cardiologists are familiar with MONA, which stands for morphine, oxygen therapy, nitrates, and aspirin.

"We were all taught that MONA is our friend anytime we met a cardiac patient with ischemic chest pain," Kern said. "But this may be, with the AVOID trial, the beginning of her demise."

Every good trial has to start with the mnemonic, he said wryly, "and I congratulate the AVOID trial. Should we avoid oxygen after the AVOID trial? Maybe so."

Results of a Cochrane review on oxygen therapy released last year[2] combined the small studies that have been done to date, he said, "and when combined, the data were clearly inconclusive. It actually suggested harm but was underpowered, so no real conclusion could be made."

The current study took on that question, and in the prehospital setting, which makes it "even more remarkable," Kern said. "They found, as they prespecified, that their primary end point of infarct size was significantly less without oxygen. That's an astounding finding, and really one that I think will cause many cardiologists and physicians to take note and perhaps step back."

That's an astounding finding, and really one that I think will cause many cardiologists and physicians to take note and perhaps step back."

He cautioned, though, that infarct size using biomarkers is still a surrogate end point, not hard clinical outcomes, and the use of biomarkers, "although admirable, is perhaps not today the most accurate." What is accurate, on the other hand, is the use of CMR, "and the data held up at 6 months as well—with oxygen therapy there was a larger infarct."

He raised a few issues, though, with the study. One is that they used 6 to 8 L/min of oxygen, while in the hospital phase "we would use less," 2 to 4 L/min, particularly for this population of patients who were not hypoxic, Kern noted. "The effect of that extra oxygen is not known."

Although the curves for oxygen saturation clearly separated in the trial, he said he would be interested to see blood gas measures, but these were not available since it was done out of hospital.

Finally, while they were secondary end points, there was an increase in significant arrhythmias and recurrent infarction in those given oxygen. "Certainly the arrhythmias could be explained by microvascular damage and ongoing ischemia, but perhaps not quite so with the recurrent infarctions, which are more typically plaque rupture during the hospital course before discharge," he said.

"But I really congratulate the authors on this very provocative, if not in fact definitive, trial," Kern said, and it will be of interest to see if there will ultimately be a mortality difference.

"So back to MONA," he concluded. "Should we divorce her, or as Neil Sedaka said, at least break up? I guess I'm not quite ready to do that, but I'm certainly willing to date her less often."

The study was funded by Alfred Hospital Foundation, FALCK Foundation, and Paramedics Australia. Stub reports research grants from the Royal Australian College of Physicians and Cardiac Society of Australia and other research support from St Jude Medical. The coauthors report no relevant financial relationships.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.