Wednesday, April 27, 2011

EKG - Sensitive for MI?

Motivation: During my final clerkship in medicine, I was plagued by chest pain.  Almost every day, a patient complained of chest pain, and the usual reflex was to order a 12-lead EKG.  Most of the time, the EKG was unchanged, but I did not know what to make of the normal result.  Could this person still be having a myocardial infarction?  The answer was, after all, yes.  At the same time, an unchanged EKG presumably decreased the likelihood of the patient suffering an MI.  So, how sensitive is EKG or serial EKGs for MI?

Paper: Usefulness of Automated Serial 12-Lead ECG Monitoring During the Initial Emergency Department Evaluation of Patients With Chest Pain. Fesmire, F. et. al. Annals of Emergency Medicine (1998), (31): 3-11.

Methods: A prospective observational study in 1000 patients with chest pain admitted at an academic center.  Patients received an initial ECG and automated sequential ECG 20 minutes apart.  The ECG information was compared against the final diagnosis at discharge.  Acute myocardial infarction was diagnosed on the basis of cardiac enzymes, new Q-wave formation, or death within 24 hours of presentation.  Unstable angina was diagnosed if the in-hospital attending diagnosed the chest pain as likely ischemia related.  Investigators examining the ECG were blinded to final diagnoses.  Exclusion criteria for patients included cocaine use, tachyrhythmia, presence of pacemaker, and discharge from ED.

Results:
Acute Myocardial Infarction: Among patients with final diagnoses of acute MI, the initial ECG had a sensitivity of 55.4%.  Serial ECG twenty minutes apart had sensitivity of 68.1%.  The difference in sensitivity between serial ECG and initial ECG was statistically significant (p < 0.001).  The specificity of initial ECG and serial ECG monitoring for acute MI were comparable (94.6% vs. 94.8%, difference not significant).
Acute Coronary Syndrome (ACS): For patients with final diagnoses of ACS (which consisted of MI plus unstable angina), the sensitivity of an initial ECG was 27.5% while the sensitivity of serial ECG was 34.2% (difference statistically significant with p < 0.001).  Serial ECG had higher specificity at 99.4% compared to 97.1% for initial ECG (p < 0.01). 
Mortality: In the study, 17 patients died (8 with initial diagnosis of acute MI, 8 with initial diagnosis of unstable angina, and one patient from non-ACS causes).  17.6% of deaths occurred in patients with no changes in serial ECG.

Discussion: Overall, a one-time ECG is a pretty insensitive tool for detecting acute myocardial infarction (sensitivity of 55.4%) and an even poorer tool for diagnosing ACS (sensitivity of 27.5%).  Serial ECG had higher sensitivities, but the difference to me was not overwhelming.  Rather, I think that the study highlights that if ACS is a real concern, the diagnoses still hinges on a good story and cardiac enzyme monitoring.  The ECG can be used as a triage tool to identify patients with STEMI, but for all patients with ACS in hosptial, the ECG is not an appropriate tool to rule out concern.

Some of the methodological deficiencies in the study may have inflated the calculated sensitivity of ECG.  The study only considered patients admitted to the hospital, but it is known that myocardial infarctions are often missed in the emergency department.  So, it is possible that some patients with MI and normal ECG were discharged.  With inclusion of these patients, the sensitivity would be even lower.

Thursday, April 21, 2011

Machine CPR

Motivation: In the past few days, I realized that doing CPR is exhausting and doing quality CPR is even more tiring.  But, even with our Herculean efforts, critical organs like brain and kidney teeter at the edge of survival.  Today, as we were drilling through CPR, I wondered wouldn't it be great if we could hook up the circulation to a machine and focus on restarting the heart? Such a machine, of course, already exists and is used in children - the ECMO or extracorporeal membrane oxygenation.  ECMO assisted CPR for adults is currently in the experimental stages.  Given the dire nature of CPR, no randomized trials have been conducted, but a group in South Korea recently implemented an ECMO assisted CPR protocol and reported retrospective results.


Paper: Extracorporeal cardiopulmonary resuscitation in patients with inhospital cardiac arrest: A comparison with conventional cardiopulmonary resuscitation.  Shin, TG et. al., Crit. Care Med. (2011) 39: 1-7.


Methods: The study considered patients between ages 18-80, who had witnessed in-hospital cardiac arrests.  Patients were considered for ECMO only if regular CPR was conducted for greater than 10 minutes without return of spontaneous circulation (ROSC).  Contraindications for ECMO assisted CPR (E-CPR) included previous neurologic damage, other terminal disease, or arrest of septic origin.  During chart review, the authors also excluded patients who had cardiac arrest from known trauma, hemorrhage, asphyxia, or non-cardiac cause.  Importantly, during the study, the decision to invoke ECMO was left to the physician (in other words, there were no standardized criteria).


Results:
Study Population:  In the six year period reviewed, 406 patients met the inclusion criteria.  Of these, 321 patients received conventional CPR (C-CPR) while 85 patients received ECMO assisted CPR (E-CPR).  Of the 85 patients on E-CPR, successful cannulation was performed in 81 patients.  At baseline, patients receiving E-CPR were more likely to have primary cardiac disease (74.1% vs 56.7%, p = 0.004) and located in the ICU/OR/cath lab (63.5 vs 31.8%, p < 0.001).  Other baseline values such as age, gender, other cardiovascular risk factors, and initial rhythm at arrest were equivalent.
Outcome: Without adjusting for baseline differences, ROSC was achieved more frequently in E-CPR (75.3 vs. 52.05, p < 0.001).  Intact neurologic survival at discharge was also higher in E-CPR (28.2 vs 7.8%, p < 0.001).  Importantly, at six months post-arrest, intact neurologic status was higher in patients receiving E-CPR versus C-CPR (28.2 vs 7.5%, p < 0.001).
Adjusted Data: Since the E-CPR and C-CPR groups were not randomized and had baseline differences, the authors matched the two groups to be equivalent in age, gender, comorbidites, cause of arrest, rime of arrest, initial rhythm, and other baseline parameters.  Even after adjusting, E-CPR group had better intact neurological survival at discharge (23.3 vs 5.0%, = 0.013) as well as six month overall survival (26.7 vs 8.3%, = 0.019).   


Discussion:  Although the study has many weaknesses, I was very impressed by the promise of E-CPR.  With ECMO, an astonishing 28% of patients suffering from cardiac arrest had minimal neurologic damage.  Also, at baseline, the patients getting ECMO tended to be sicker since they were more likely to be in the ICU (though this difference was adjusted for).  Also, ECMO was used as a last resort by physicians in prolonged arrest situations, and the effects of ECMO may be even better if it is applied as soon as possible.  One of the promises for ECMO is that if survival in arrest situations can be prolonged, then more definitive interventions like PCI or thrombolytics can be administered to save patients after cardiac arrest.


While promising, the paper has many weaknesses.  The primary weakness in the design is that there were no standardized criteria when E-CPR should be applied.  This introduces a whole host of possible selection bias that may not have been adjusted for.  Also, the authors considered a very select group of patients - in-hospital witnessed cardiac arrest of presumed cardiac etiology.  The applicability of this result to the broader population of patients (such as a septic patient suffering from PEA arrest) is unclear.  However, given the promising results, the next step is likely a prospective study.  In ten-fifteen years from now, we may all be practicing ECMO-CPR and give up on lengthy, tiresome chest compressions!

Friday, April 15, 2011

Disulfiram & Activation of Latent CD4 Reservoir

Xing et al. (April 2011), Disulfiram reactivates latent HIV-1 in a Bcl-2 transduced primary CD4+ T cell model without inducing global T cell activation. JVI.

Highly active antiretroviral therapy (HAART) can reduce plasma HIV-1 levels to below the detection limit. However, due to the latent reservoir in resting CD4+ cells, HAART is not curative. Elimination of this reservoir is critical to curing HIV-1 infection. Agents that reactivate latent HIV-1 through nonspecific T cell activation are toxic. In the above paper by Xing et al--which isn't even fully out in print-- disulfiram has been shown to reactivate latent HIV-1 without global T cell activation. The extent to which disulfiram reactivates latent HIV-1 in patient cells is unclear, but the drug alone or in combination may be useful in future eradication strategies.