Patients with PE don’t die as a direct result of the clot, but rather due to right ventricular (RV) failure. One of the contributing factors to decompensation appears to be pulmonary vasoconstriction, which increases the right ventricular afterload and thereby exacerbates RV failure.
Administration of an inhaled pulmonary vasodilator is therefore a very logical intervention for (sub)massive PE. Severe PE appears to cause depletion of normal nitric oxide levels. Exogenous nitric oxide inhalation might have the following effects:
- Dilation of pulmonary vasculature, thereby reducing the RV afterload and avoiding decompensated RV failure (likely primary mechanism of benefit).
- Preferential vasodilation of capillaries in the best ventilated areas of the lung may improve ventilation-perfusion matching and thereby improve oxygen saturation (potential secondary mode of benefit, but oxygenation generally isn’t a big problem in these patients).
- Nitric oxide has anti-platelet effects which might augment anticoagulation and thereby prevent extension of clot (theoretical).
Inhaled nitric oxide (iNO) is generally regarded as safe (rapid degradation of nitric oxide prevents undesirable systemic vasodilation). Methemoglobinemia can occur, but this seems to be rare. The main drawback is cost.
Case series describe the successful use of inhaled nitric oxide to stabilize patients with submassive PE.1–3 A new RCT finally offers us some higher quality evidence on this.
iNOPE trial: Inhaled nitric oxide to treat intermediate risk pulmonary embolism: A multicenter randomized controlled trial
This was a multi-center RCT investigating the effect of iNO among patients with submassive PE who were treated with heparin anticoagulation.4 76 patients were randomized to placebo vs. iNO at a dose of 50 parts per million for 24 hours. Inclusion criteria were as follows:5
Patients were well matched at baseline. Median blood pressures were ~125/95, 70% of patients had RV hypokinesis on echocardiography, and the median estimated pulmonary artery systolic pressures on echocardiography were ~52 mm.
Here is where things get unusual. The primary and secondary endpoints were:
- Primary endpoint: Composite requiring both normal RV on echocardiography and high-sensitivity troponin <14 pg/mL.
- Secondary endpoint: Composite requiring both brain natriuretic peptide concentration <90 pg/ml and a Borg dyspnea score <=2.
These endpoints are problematic for several reasons:
- BNP and troponin are products of both the right and left ventricle, with clearance by the kidney. Therefore, neither test is purely an evaluation of the right ventricle. Troponin may take a while to normalize, even after the right ventricle has improved.
- Both endpoints are composites that combine different sorts of endpoints. Ideally a composite should combine roughly equivalent items, but these composites don’t. For example, in the first composite, a normal RV is clinically important, whereas the exact value of the high-sensitivity troponin is less important.
- To make matters a bit worse, the definition of a “normal RV” is actually a composite endpoint of several sophisticated echocardiographic measurements. So the primary endpoint is a composite of a composite that requires a total of five components be satisfied:
- RV size <42mm in diastole
- Tricuspid annular plane systolic excursion >16 mm
- RV index of myocardial performance <0.4 using spectral Doppler or <0.55 using tissue Doppler
- RV fractional area of change >33%
- High-sensitivity troponin <14 pg/mL.
- All endpoints involve a conversion of a continuous variable (e.g., troponin or BNP) into a dichotomous variable, depending on whether the variable is below a specific cutoff. Conversion of a continuous variable into a dichotomous variable causes a loss of information, making it harder to determine the presence of a signal.
These endpoints are not the endpoints that one would typically choose to optimize the likelihood of obtaining a positive study. For comparison, let’s look at the ULTIMA trial, a study which was able to obtain a positive primary endpoint despite including only 59 patients with submassive PE.6 ULTIMA’s primary endpoint was the difference in RV/LV ratio on echocardiography over 24 hours, allowing ULTIMA to achieve a p-value below <0.001. This primary endpoint was a wise choice for several reasons:
- It’s not a composite – it’s looking at a single variable.
- It’s a continuous This allows for detection of small differences in a statistically significant fashion.
- It’s expressed as a change from baseline, which tends to neutralize differences between groups at baseline. This facilitates detection of subtle treatment effects.
Primary endpoint: iNOPE
More patients in the iNO group met the primary endpoint (24% vs 13% of patients), but this wasn’t statistically significant (p=0.4). Meanwhile, fewer patients in the iNO group met the secondary endpoint (13% vs. 34%), but this difference also wasn’t statistically significant (p=0.1).
It’s hard to know what to make of this, because these endpoints are so unusual. It’s possible that the combination of numerous cutoff values decreased the signal strength, causing the composite endpoint to be under-powered.
Hemodynamic & safety endpoints: iYEP
Let’s move on to the echocardiographic data, which is easier to make sense of. For the sake of this discussion, “normal RV” is defined as a right ventricle without dilation or hypokinesis (which is likely how most clinicians would define this).
- iNO increased the proportion of patients with a normal RV after 24 hours (19/35 vs. 10/36; p =0.03, fragility index of one).
- Only some patients had a baseline echocardiogram. Among such patients the RV was normal in 4/22 patients in the iNO group and 4/24 patients in the placebo group. Tracking these patients over time, after 24 hours the RV was normal in 11/22 patients in the iNO group compared to 5/24 patients in the placebo group (p=0.01). So iNO caused interval improvement in RV dysfunction, whereas placebo patients remained largely stable over time.
No adverse events were noted among patients receiving iNO. This is reassuring and is generally consistent with prior research and experience demonstrating that iNO is a safe agent.
Right study, wrong endpoint: What now?
This is a common situation: investigators perform an excellent study using the right cohort of patients and the right intervention, but choose an unfortunate primary endpoint. The study is technically “negative,” but secondary endpoints are positive (secondary endpoints which arguably ought to have been the primary endpoint from the beginning).
Technically the study should be considered negative and repeated. In a universe with infinite resources this would make sense. However, the truth is that this is the best study on iNO that we have, or that we’re likely to have anytime soon. Therefore, it may be reasonable to consider the study as weakly positive overall, overlooking the primary endpoint.
Potential roles of iNO in PE?
1) Bailout of crashing patient
The most obvious role of iNO is to stabilize a crashing PE patient. iNO may cause immediate hemodynamic improvement, stabilizing the patient and bridging them to some other definitive therapy (e.g. medical thrombolysis or surgical thrombectomy).
Historically, this has been the primary role of iNO. Evidence from the iNOPE trial supports this and implies that it should perhaps be more widely considered.
2) Pre-emptive stabilization of the submassive PE patient
Another less dramatic role could be to pre-emptively stabilize a patient with a submassive PE at risk for decompensation (e.g. perhaps a high-risk submassive PE patient with contraindications to thrombolysis). With anticoagulation, the clot will generally break up eventually, but this may take some time. iNO could potentially be used to hemodynamically stabilize the patient while waiting for the clot to resolve, thereby reducing the risk of hemodynamic deterioration.
The iNOPE trial promotes this concept, but not strongly enough for routine utilization. Since these patients aren’t crashing, the benefit/risk ratio is less clear than in the crashing patient above.
A word on cyclic GMP, pulmonary HTN, and sildenafil (um, yeah, Viagra)
Nitric oxide is extremely expensive and logistically difficult to deploy (many smaller hospitals don’t have it at all). Oral sildenafil has a similar mechanism of action that works to reduce pulmonary vasoconstriction, but it’s cheaper and more widely available. Sildenafil has been shown to cause significant improvements in pulmonary pressures within a few hours of initiation (24613188).
This raises the possibility that oral sildenafil could be used to temporarily treat pulmonary hypertension due to acute PE, as a bridge to endogenous thrombus resolution. Theoretically this could help stabilize patients with sub-massive PE and prevent further deterioration, buying additional time to allow for anticoagulation and endogenous thrombolysis to degrade the clot. Sildenafil has long been used as therapy for chronic thromboembolic pulmonary hypertension. Its use in acute thromboembolic pulmonary hypertension is supported by animal models and case reports.7–16
Thoughts from the study's primary investigator:
I e-mailed Jeffrey Kline to elicit any further comments he has on this, and this is his response:
What I would want to respond to is the criticism of my composite endpoint. Patients don’t care if their RV:LV ratio is 0.99 or 1.17 (the magnitude of difference that Nils found). They should care if their echo is normal and they have no necrosis because they predict good quality of life.
I would absolutely agree that a normal echo and lack of necrosis is a more important clinical endpoint than RV/LV ratio. The issue is selecting an endpoint which can investigated with adequate statistical power in a rare condition with a limited sample size. Any study of submassive PE faces a tough choice between selecting the most clinically important endpoints (e.g. mortality, echocardiographic normalization) versus endpoints which are easier to test definitively (e.g. echocardiographic RV/LV ratio).
- The use of an inhaled pulmonary vasodilator is a logical strategy for stabilization of PE patients (especially nitric oxide, which may be depleted in this situation). Previously inhaled nitric oxide has only been supported by case series.
- iNOPE is a multi-center placebo-controlled RCT which demonstrated that iNO is safe and that it improved hemodynamics (causing improved RV function). The study was too small to determine whether this translated into an improvement in clinical endpoints (e.g. fewer episodes of hemodynamic deterioration).
- iNOPE utilized a strange composite primary endpoint, which was not different between both groups. For this reason, it may technically be regarded as a “negative” trial.
- Inhaled nitric oxide may be a very useful therapy to stabilize the crashing PE patient and bridge them to further therapies. Although proving this in an RCT may be nearly impossible, iNOPE provides some evidence to support this therapy.
- PubMed to primary publication of the study here.
- Paper describing study protocol is freely available here.
- Inhaled nitric oxide: A tool for all resuscitationists? (Cliff Reid, Resus.Me)
- Eight pearls for the crashing PE patient (PulmCrit)
- The critical PE patient (Salim Rezai, RebelEM)