I have been asked to talk about the diagnosis of massive PE for the upcoming EUSEM congress in Glasgow this month. We have blogged on this before1, but it’s a thorny issue and one that is probably worth revisiting in the context of recent further evidence. We keep seeing these cases clinically; it’s always worth using the best simulator (your brain) to go over how you would manage serious and common cases, so that you are prepared when they arrive.
We are not going to talk about diagnosis or PE today per se,but how you diagnose a massive PE, and the advantages of doing so. Some definitions of massive PE are simple and based entirely on non-invasive blood pressure measurements. But is this the best way to classify this disease? I think we have highlighted this before, but for me these cases require system 2 thinking.
Let’s take this 54 year old woman for example, who walks in to your department on New Years Eve. She has had a cough for 5 days, beginning to expectorate a little phlegm. She is febrile. She has no past medical history of note. She does however, have observations that are remarkable – a pulse of 117, a BP of 127/63 and saturations of 73% on room air. These rise to the low 90’s on 60% oxygen. She is breathless at a rate of 28, but alert and lucid. A CTPA reveals a moderate left lower lobe pneumonia, but also a central pulmonary embolism extending into all branches throughout the right sided arterial tree. There is also notable right sided cardiac dilatation and evidence of strain.
How would we classify this? The American Heart Association2 would be using the phrase submassive here. European Society of Cardiology guidelines3 would be declaring this an ‘intermediate risk’ PE based on the absence of shock or hypotension, which they define as a systolic as <90 or a systolic drop of >40mmHg for more that 15 minutes in the absence of an alternative cause. NICE would be asking us to think about whether this patient had ‘haemodynamic instability’4, as that is the only fine print they use to discuss severity of disease. The PESI score5 would be clocking in at somewhere in the mid 90’s, telling us this patient has a 30 day mortality risk of somewhere between 3.2 and 7.1% at 30 days. None of these classification systems really scream ‘massive’ PE at us in this context. Does that mean this patient is safe, nothing to worry about?
What does ‘massive’ even mean? It’s certainly a crude word to describe the complex pathophysiology of a serious disease. The definition appears to originate from estimation of thrombus burden in the 1970’s, using the Miller index6, where massive was a clear descriptor of size. However, following this work, clinical registry data 7noted an association between increased 90-day mortality in patients with acute PE and a systolic blood pressure <90mmHg at presentation, compared to others. Similar European registry data8 mirrored these results and clinical prediction research also began to highlight the link between hypotension/shock and increased risk of adverse outcome. The AHA consequently proposed a definition for massive PE in 2011, that centred on sustained hypotension, without reference to clot burden. They also included dependence on ionotropic support interestingly, which is an element lost in other guidelines.
But is that all there is to it? Or, is this more complex. Really what we want from any classification system, is either a way of succinctly communicating complex information, or identifying accurate risk prediction. And we want to predict risk, because that inherently guides our threshold for intervention. The higher the risk, the more justified the intervention. But do we achieve this with simple definitions of massive and submassive? High, intermediate and low risk? We have already heard that the definition of massive PE was proposed based on registry (rather than RCT) data and the numbers of hypotensive PE patients in ICOPER was small 9(108 patients only). The main evidence10 from a systematic review linking haemodynamic instability to improved outcomes with thrombolytic treatment, predates this definition. As such we have found ourselves administering thrombolysis based only on a dumbing down of clinical gestalt to a simple objective measurements. This is arguably the case for decision making in submassive PE as well, with the recent PEITHO trial solely focussed on including patients meeting the ESC definition of upper intermediate risk. We know that some PE patients have an adverse outcome in relation to comorbidity. We know that there are other factors in play when we review patients on the shop floor and make decisions on tiered escalation. There is more to this disease than the first non-invasive blood pressure.
Rather than using single word definitions to tailor our interventional threshold, what we need to really ask ourselves for each case, at each time point, is ‘do I need to do more than anticoagulate this patient’. Therapeutic dose anticoagulation is established and effective therapy for this disease. A ‘no brainer’ apart from in the most complex of cases. Systemic thrombolysis, catheter directed lysis, surgical embolectomy, EKOS and ECMO are all potentially accessible additional treatments, but for which the evidence base is far more vulnerable. What we need from a diagnosis of massive PE, or any classification system, is information that supports doing more than just anticoagulating.
And therein lies the rub. No single simple definition will give you this information. If you want to properly risk stratify, you need to think hard and carefully evaluate. An acute pulmonary embolus can interact on multiple organ systems and disrupt physiology in many ways. But there are clear areas on which to focus, that should inform your decision to escalate care above the norm.
So, this is the bread and butter of it. If there is clear haemodynamic compromise then additional interventional strategies must be considered, so say NICE, ESC, AHA, ACCP etc.. However, the arbitrary definition of hypotension defined by these groups disparages the increasing evidence base and technological expertise available. When discussing CVS instability, we could potentially do better than a systolic of <90, which can often be too late.
Some authors suggest clinical features such as presyncope, dizziness and syncope can provide clear evidence of clinically relevant hypotension11, even if the numbers do not fit the box. Pulse rate is an indirect marker of RV reserve and included in multiple risk scoring systems as a result. Blood pressure is of course highly relevant, but non-invasive measurement in diaphoretic patients can be unreliable; arterial line insertion and waveform analysis can provide reliable, objective and temporal data on blood pressure which can be compared to baseline if available. This can also facilitate regular blood sampling to look for data supporting a diagnosis of biochemical shock, such as rising lactate levels and increasing metabolic acidosis. There is reasonable data 12 to suggest the former correlates directly with mortality, independent of hypotension, RV dysfunction or myocardial injury.
Real time evaluation of cardiovascular function is now also readily available with point of care echocardiography. Several findings which indicate RV dysfunction13 have been identified as independent predictors of an adverse outcome in acute PE, although standardisation is challenging14. Also, in those patients who are otherwise haemodynamically stable, the positive predictive value is low15. Undoubtedly though, this gives you a real window to the stability of the heart and the degree of compromise, both of the left and right ventricle. These images can be supported by biomarker evaluation of myocardial injury using, highly sensitive troponin or natriuretic peptide measurements. Occasionally, you will also gain bespoke information to stratify your patient such as the presence of a clot in transit.
In addition to echo, RV dilatation on CT has previously been shown to be an independent predictor of increased mortality 16 in both haemodynamically stable and unstable patients. These modalities provide real time evaluation of cardiac function. A picture is worth a thousand words (or numbers)….
Hypoxaemia is an interesting phenomena in pulmonary embolism. Most clinicians think of PE as a disease of the lung; it is of course principally a cardiovascular event and so lung function should remain preserved in the main. This takes us right back to Barcroft’s classification of Hypoxia, and the recognition that dead space, rather than shunt is the initial issue in PE. The question of why we get hypoxic with PE gets trickier though depending on the stage of disease. Redistribution of blood flow to underventilated lung units, pulmonary oedema and subsequent shunt at the site of emboli/later infarcts can all play a role. Andy Neil has done a great deep dive on this 17 if you want to hear about iatrogenic clots in mice and what it does to their lung function…..
Hypoxaemia is included in the PESI criteria as a poor prognostic sign, but there is no clear degree that appears to differentiate high from intermediate risk PE regarding the evidence. However, understanding the physiology in itself can be helpful. Hypoxia in PE will not occur with small peripheral emboli or even unilateral segmental clots, due to the adequate ventilation throughout and ability of the lungs to redistribute blood supply.
Hypoxia is a particularly late sign in acute PE, implying either very large/bilateral clot burden, underlying pulmonary comorbidity or an alternative pathology. Imagine having a pneumonia on one side, with a physiological shunt to the contralateral side, which was unfortunately full of emboli. Remind you of anyone? This understanding can influence our treatment decisions. For example, what do you think intubation and ventilation will do to our patient here? Will filling those lungs with 100% oxygen help in the absence of any blood to pick it up? What will the consequent reduction in preload do to the RV? Refractory hypoxia can and should factor in to your treatment decisions with pulmonary emboli, as sometimes clearing the pipes rather than blowing up the balloon is the answer to this problem.
Things get even more interesting with clot burden. We get a tremendous amount of information from the CTPA in addition to the headline of ‘VTE positive’, which we would be fools to ignore. We mentioned earlier the initial definition of massive PE came from the Miller index18, a simplified measure of clot burden and an attempt to link outcome to the size/distribution of clot. There are multiple other clot load scores, such as the Walsh, Qanadli and Mastora indices19. All of these scores have the same aim. Previous retrospective work has described a >10 fold increase in mortality risk20 when the obstruction index approaches >40%, backed up by more recent regression analyses demonstrating a significant correlation between clot burden and 30 day mortality outcomes21.
In addition to size, there is also location location location. Papers looking at anatomical clot location and short term outcomes have suggested that central (saddle or at least one main pulmonary artery) emboli are associated with an increased risk of all cause death or clinical deterioration in haemodynamically stable patients22. This again suggests the sum of the parts should be examined to guide decisions on intervention.
And what about the legs? They are a fair way away from the lungs, but none the less seem to remain relevant. A prospective cohort study found that the presence of concomitant DVT in patients with a new diagnosis of PE doubled the hazard ratio for all cause mortality and quadrupled the hazard ratio for PE specific mortality. These findings were externally validated in the RIETE registry23, suggesting the presence of concomitant DVT to be an independent predictor of death in the first three months following diagnosis of PE.
This is not rocket science, but of course clinical progress has to impact on this thought process. Benefit from systemic reperfusion therapy has been reported up to 14 days from symptom onset24. Making a decision to do more in someone who has not got better with anticoagulation, is a different and easier decision that in it is in someone who has not received any treatment at all. In itself, a failure to improve on anticoagulant therapy can in many ways be considered a diagnostic criterion of a high risk PE, provided dual pathology has been excluded.
Decisions like these are often best made in a critical care environment, where invasive monitoring is readily available and staffing allows for enhanced frequency of observations and blood testing. Sequential scoring using validated tools with such as the PESI and Geneva prognostic score which allow for incorporation of objective progress, alongside inherent risk, can be a very useful addition to gestalt evaluation. These scoring systems are recommended by most modern guidelines and provide helpful support in uncertainty.
We have always been enthusiastic in this area about prediction of clinical deterioration and justification of tiered intervention. There has previously been less focus on bleeding risk. This is a crucial part of the classification; it guides decisions to escalate therapy at an equal level to cardiorespiratory features and it determines which alternative therapies may be most suitable. Analysis of pooled data using different thrombolytic regimes and agents suggest a haemorrhagic stroke rate of between 1 and 2%25.This rightfully mandates caution. But new data is emerging that can perhaps help us determine levels of risk within these patients.
In PEITHO for example26, age appears to play a significant role in the risk of bleeding. While haemorrhagic stroke occurred in 12 patients, 75% of these events occurred in patients over the age of 70. If you look at extracranial bleeding rates, you can also see a reduction in the hazard ratio for those <75.
Further literature continues to come through on predictors of haemorrhage following thrombolysis for PE. Studies like this one27 and this one28 describe high risk characteristics on multivariate regression, while others such as this29 one highlight lower risk groups from longtitudinal population based studies. There are extrapolations from thrombolysis in MI and Stroke with proposed scoring systems, but care is needed here due to the unknown generalisability.
What we can be sure of is that bleeding risk with thrombolysis is not global. As always, risk depends on individual characteristics.
Shared decision making
When you have this information in front of you, then of course it should be shared. ‘No decision about me, without me’ is an excellent mantra that I will remember following a recent public/patient research engagement meeting. Patients should be counselled about the likely potential for clinical deterioration, the available options for therapy and their consequent risks. We should be clear about what we know and what we do not know. We are often uncomfortable with this, but it is an absolute pre-requisite to shared decision making. If you get to the end and they ask ‘what would you do doctor’, that is the time to offer opinion. Prior to that, it should be the facts.
What did we do in the end?
Well, on discussion this particular patient had unfortunately suffered a very memorable family bereavement due to venous thromboembolic disease. When they understood the gravity of the situation and the potential options, they were keen to pursue more aggressive therapy than standard anticoagulation. With a low bleeding risk, an out of hours presentation and no on-site interventional radiology cover, we opted together for systemic thrombolysis with alteplase using a 2 hour regimen.
It worked a treat. She was stepped down from ICU the next day to the ward and discharged home 3 days later. Follow up at respiratory clinic 1 month later was unremarkable.
If you take away only one thing from this talk, strangely, and with a nod to Shrek, I think it is the idea that you should consider a big PE like an onion30; ignore the crude definitions on the useless skin, and peel back the layers. By doing this you can gather information that will highlight the risk of serious deterioration, the risks of bleeding with advanced treatments and patient preferences. Combine this with your experiential learning and the wealth of evidence and you will be able to present your patient with educated options. They will thank you for it.
This was a long one I know. Thanks for sticking with me if you are still here.
- Subscribe to the blog (look top right for the link)
- Subscribe to our PODCAST on iTunes
- Follow us on twitter @stemlyns
- PLEASE Like us on Facebook
- Find out more about the St.Emlyn’s team
- Come join us at our conference in October 2018