Brain abscess + pulmonary AVM + pulmonary htn

Earlier this week, the inestimable Rob Weber presented a case of a middle-aged man with acute neurologic symptoms (facial droop, dysarthria and abnormal arm movements). Head imaging revealed a large frontal lobe mass, and the ultimate diagnosis was anaerobic brain abscess in the setting of a pulmonary AV malformation. 

Lurit did a phenomenal job facilitating a rich case discussion — one of my favorite AM reports of the year!

I’ll focus on a few of the interesting points from the case. 

Brain Abscess:

Harry Hollander reminded us that brain abscesses are either the result of direct spread from periodontal or sinus infection or due to hematogenous spread. The anatomy of the abscess can be a clue: typically, direct spread of infection from the teeth or the ear can cause an abscess in either the frontal lobe or the temporal lobe respectively. On the other hand, if there is hematogenous spread, you’re more likely to see multiple abscesses, scattered along the distribution of the largest artery in the brain, the middle cerebral artery (this is described in further detail in the UpToDate page). 

This patient did not have signs of systemic inflammation — no fever, no white count, & normal inflammatory markers. HH reminded us that fever is not sensitive for brain abscess (may be missing in 50% of cases). 

Right to left shunts are a known risk factor for brain abscess. The pulmonary capillary beds play a key role in “disinfecting” venous blood — both because of the lymphatic system and because anaerobic bacteria can’t survive the high oxygen tension in this environment. In fact, up to 10% of patients with pulmonary AV malformations can develop brain abscess. 

Pulmonary AVMs and Hereditary Hemorrhagic Telangiectasia (HHT)

Interestingly, the majority of patients with pulmonary AVMs actually have HHT — aka, Osler-Weber-Rendu Syndrome (Rob did present the case, but he tells us it’s a different Weber). Although there were no obvious mucocutaneous telangiectasias on exam,  this patient reported a history of frequent nose bleeds, which even further supported the case for HHT. Lurit and Rob taught us about the Curaçao criteria for diagnosis, and the patient is undergoing genetic testing. 

Chest CT revealed a pulmonary AV malformation, which is strongly associated with HHT.

Mechanism of Pulmonary HTN in HHT (for the Physiology Geeks)

This patient had a TTE to rule out endocarditis as a cause for brain abscess. There were no valvular vegetations, but interestingly, there was RV pressure and volume overload, and pulmonary HTN was confirmed on RHC (mean PA pressure 38 mmHg, PVR 8.1 Woods, PCWP 10 mmHg, consistent with “pre-capillary” etiology). 

I was puzzled about the mechanism of elevated R-sided pressures here. On one hand, we see elevated L-sided filling pressures in systemic AV-shunts — this is, in fact, “high output” heart failure: low SVR A.) causes a compensatory increase in LV output (to maintain MAP) and B). induces RAAS, which increases volume retention and preload; if you have low ventricular reserve, the increased circulatory volume results in elevated filling pressures despite the high output. 

Could a pulmonary AVM do something similar for the right heart —  i.e. could the patient have “Right sided high output heart failure”? It’s an intriguing idea, but the pathophysiology doesn’t quite make sense because the PVR here is *high* not low (remember: low SVR is critical to the pathophysiology of left-sided high output failure). 

The all-star cardiology fellow on this case, Megan McLaughlin (who, mind-blowingly, considered HHT on the differential within seconds of hearing about this patient), was similarly dissatisfied. She dug up an amazing article which actually explains that there are two distinct mechanisms of pHTN in HHT. First, these patients can get post-capillary pulmonary hypertension due to high output heart failure from *systemic* AV malformations (usually in the liver); this patient did not have hepatic AV malformations, and, as RHC numbers above suggest, he had pre-capillary disease. Fascinatingly, the other mechanism of pHTN in HHT is due to arterial remodeling, similar to what is seen in PAH (i.e. group 1 pHTN), and may be related to some of the genetic mutations implicated in HHT (Faughnan et al ERS 2009). 

Yes, I do love extracting cardiac physiology at every opportunity — but this is actually important for the management! Patients with brain abscess and pulmonary AVMs should get an endovascular embolization of the AVMs — this gets more complicated in a patient with pHTN, since plugging up the AVM will cause acute R-sided pressure overload — which could therefore worsen the pHTN. This patient was started on pulmonary vasodilators to drop the PVR before planning embolization.

Wow, what a case! The “Weber Archives” (as HH dubbed them) never disappoint. 

Bacterial infections and the glomerulus

On Monday, Anna Fretz presented a case of a middle-aged man who presented with anasarca and was found to have persistent MSSA bacteremia, hematuria, and heavy (but not quite nephrotic-range) proteinuria. 

Interestingly, he did not have endocarditis or another identifiable focus of infection, despite +MSSA cultures on two admissions separated by several months. He had a negative ANCA, and low C3 levels. Kidney biopsy showed glomerulonephritis, with +IgA IF staining. 

Glomerulonephropathy gets complicated really quickly — both because of the confusing anatomy of the glomerulus and because of the complex immune phenomena (or should I say, immune complex phenomena, lol). The terminology is more confusing still. 

It was a fascinating case, and I’m so glad Harry Hollander and Gurpreet Dhaliwal were present to help navigate the complicated territory. This case inspired me to take a dive into a narrow slice of GNs, and focus just on how bacterial infections can cause glomerular disease. 

Post-streptococcal glomerulonephritis:

Remember from medical school? Post-streptococcal glomerulonephritis typically involves a child who gets strep throat or a GAS skin infection. After some time, some of the antibodies against strep antigens cross-react with glomerulus antigens (molecular mimicry) and incite inflammation; this is why usually post-strep GN happens in the weeks after an infection (takes time for the antibodies to develop, deposit, and cause a reaction). (I’ve oversimplified the pathophysiology: there’s also deposition of circulating immune complexes, and there’s direct deposition of positively charged strep antigens which get trapped in the glomerulus —  since the electrical filter in the glomerulus is negatively charged— where the antigens incite an in situ immune response). 

Anyway, I’ve never seen a case of this: first off it’s more common in kids, and second the incidence is decreasing significantly in developed countries because we’re so good at treating strep throat before the immune system has time to misbehave. 

In contrast, to post-streptococcal GN, where there is a post-infectious immunologic phenomenon, there are also GNs that result from immunologic sequela of active infection. 

Most classically, this includes endocarditis-related GN and shunt nephritis. A bug actively infects either a valve or shunt, and then incites an immune response (bug antigens+antibody complex) that can migrate and deposit into the glomerulus. I remember Harry Hollander teaching me about shunt nephritis as an intern — it’s less common now because ventriculo-arterial shunts have been replaced largely by ventriculo-peritoneal shunts (so there’s less direct access for the bug/immune complex hanging out in the biofilm to migrate to the kidney). 

It seems that there is now broader recognition of infection-related glomerulonephritis from sources other than a shunt or endocarditis — and most commonly, the implicated bug is Staph aureus. In fact, there’s even an entire UpToDate page on staph-associated glomerulonephritis. This is relevant to Anna’s patient, who didn’t didn’t have a shunt or any other hardware, and they had no obvious valve vegetation (TTE or TEE). 

This review article describes the pathophys of infection-related GNs (IRGN). The figure below shows the sites and bugs (staph aureus most commonly) that can cause IRGN.

From Dr. Sami Nasr et al, Review in Kidney International, 2013

There’s limited data on how best to treat these patients: there’s no evidence to support immunosuppressants, and seems like the strategy involves eradication of the infection with antibiotics and source control. Anna’s patient ultimately got 6 weeks of IV antibiotics. Unfortunately, even after eradication on infection, many patients don’t recover renal function in infection-related GN. 

One last point — we mentioned that kidney biopsy showed + staining for IgA in Anna’s patient. While IgA deposition is commonly found on biopsy for infection-related GN, it is distinct from primary IgA nephropathy. Primary IgA nephropathy can present as nephritic syndrome, nephrotic syndrome or both, and it is one of the most commonly identified findings on renal biopsy in the developed world. The pathogenesis of IgA nephropathy is extraordinarily complex, and is thought to involve genetic predisposition, and possibly infectious trigger. IgA vasculitis (aka Henoch-Schonlein), if it involves the kidney, looks the same as IgA nephropathy. 

Yikes! No wonder glomerulonephropathies are so hard to learn… 

Evidence for PFO Closure in Cryptogenic Stroke

On Tuesday morning, Akshay Ravi (PGY3) and Stephanie Kim (MS4) presented a case of a man in his 20s with acute aphasia and hand tingling. Imaging suggested embolic stroke affecting the frontal and parietal lobes, and, after extensive workup, the team felt this patient had a paradoxical embolus via a patent foramen ovale. Interestingly, chest CT showed acute subsegmental PEs, which further supported venous clot with R—>L embolus as the mechanism. 

Akshay and Tim Dyster (who facilitated) did an amazing job teaching about the work up for cryptogenic stroke. The patient was referred for PFO closure, and I was curious about the evidence in support of this procedure in patients with cryptogenic stroke and PFO. Here we take a little dive into the data. 

First off — PFOs are super common — present in 25% of people. Interestingly though, people noticed (and reported in case-control studies from the ‘80s-‘00s), that PFOs are much more common (as high as 50%) in patients who have a “cryptogenic” stroke (i.e., no slam dunks: no afib, no valvular lesion, carotid stenosis, LV thrombus, HLD, HTN etc). 

It makes perfect sense: people get clots in their veins, and, if there’s a hole in the atrium, sometimes those clots get direct access to the brain. So, being the plumbers that they are, cardiologists thought: let’s plug the hole and we’ll solve the problem, right??! 

Well…not so fast. Three randomized trials were published in NEJM between 2012-2013, CLOSURE I (wins award for best trial name from this list), PC, and RESPECT, which randomized patients with cryptogenic stroke to either receive PFO closure + an antiplatelet agent vs medical therapy alone. None of these trials showed a significant difference in stroke between the two groups after a few years of followup. 

But between 2017 and 2018 a few more trials were published in NEJM (REDUCE and CLOSE) which did show a reduction in new stroke in patients who had PFO closure after initial stroke. 

So what changed? First off, the original trials allowed use of anticoagulation in patients in the “medical therapy” group, at the discretion of individual providers (some patients even got antiplatelet therapy and anticoagulation). This is not standard of care for secondary stroke prevention (that would be an antiplatelet agent), and you could imagine that, since anticoag agents are great at preventing venous thrombosis, they would provide protection against paradoxical emboli. In fact, it’s possible that anticoagulation is just as good as PFO closure + anti-platelet at reducing stroke; we just don’t know, since there hasn’t been a RCT to study this question (more on this important point in the takeways). The new trials were also more stringent about inclusion criteria — they had further workup to truly ensure that “standard” causes of stroke had been ruled out (e.g, patients had more imaging to rule out small-vessel disease) and patients generally had larger PFOs. The plugging technology improved, and follow up time was longer as well (in fact one of the original trials, RESPECT, showed a benefit for PFO closure when they reported data after 6 years of follow up compared with the original 2.6 years). 

One issue is that the end points for these studies (new stroke after original stroke) were pretty rare events. That’s why longer follow up time and meta-analyses are useful. Generally, meta-analyses have shown that, between 3 and 6 years of follow up, the risk of new stroke with an antiplatelet agent is 5%, whereas with PFO closure and an anti-platelet agent, the risk comes down to about 2%. Roughly, this means you need to place a PFO closure device in 30 patients, to prevent one stroke over the study period. Note that many of the patients in whom you close the PFO will never have had a stroke even without the procedure; additionally, many of the strokes that people have may not be disabling. 

Also consider that randomized trials assessing this question were not blinded — patients (and clinicians) knew who got the procedure and who got medical therapy alone — this should always make us consider the possibility of bias. 

Fortunately, the risk of complication after a PFO closure is quite low (main adverse effect is post-op AFib) — this is helpful to know when making a shared decision with patients on risks vs. benefits of closure. 

My takeaways: 

  1. PFO closure provides the most benefit in young patients with cryptogenic stroke where you’ve REALLY done a thorough investigation to ensure there’s no other primary cause. 
  2. These patients should also have good evidence of R—>L shunt on bubble study. (Recall, shunt can be transient with positional changes, Valsalva, etc — you don’t need chronic elevation in R sided pressure to get R—>L shunt.) 
  3. There are additional features (e.g. larger size of PFO, morphology, presence of atrial septal aneurysm) which increase risk of paradoxical embolus and affect amenability to closure — these are important to consider. 
  4. If a patient is found to have evidence of hypercoaguability (e.g. APLS), then anticoagulation is critical. In patients who you’ve committed to  anticoagulation, we don’t know whether PFO closure is useful. For this, we’d need a new randomized trial, with long followup, which compares patients with antiplatelet+PFO closure (or anticoagulant + PFO closure) to anticoagulation alone. The net clinical benefit endpoint would have to incorporate both the stroke reduction and the increased bleeding on anticoag. This is a crucial data free zone, and important to keep in mind (and share with patients) when discussing risks vs benefits of PFO closure. 
  5. I’m interested to see how this is all integrated into AHA/ACC Guidelines, which currently have not been updated to reflect the latest data on PFO closure. 

As always, please let me know if I’ve reported something incorrectly/imperfectly, or if you have any other takes!


Protein Losing Enteropathy — Think: “Nephrotic Syndrome of the Gut”

Earlier this week at AM report, Akshay Ravi presented a fascinating case of a patient with years of diarrhea, malnutrition and volume overload who was ultimately diagnosed with protein losing enteropathy (PLE) due to primary lymphangiectasia (abnormal dilation of lymphatics) in the gut. 

Akshay did an amazing job describing the differential for PLE — and I realized I really don’t have a good mental model for this process. Rabih Geha cited an analogy that I found really helpful: “Think of PLE as nephrotic syndrome of the gut.”

First let’s consider that your small bowel mostly absorbs protein in the form of broken down amino acids, from dietary intake. 

Turns out there’s also a small component of protein loss and turnover — normally, the gut loses a little bit of protein into the lumen (through the wall and through lymphatics).  Unlike in the kidney, this is *irrespective of the size of the protein* — that is to say, unlike the glomerulus, the gut is not acting like a mesh filter — you lose proteins of all sizes. 

Once in the gut lumen, most proteins dissolve away in the proteolytic environment and are converted into amino acids (that are either reabsorbed or secreted). The liver responds to the gut loss by replenishing production of proteins (albumin, etc). Under normal circumstances, this accounts for ~10% of protein turnover in the body (UpToDate article on PLE has more detail on this). 

Protein Losing Enteropathy occurs when there is excessive loss of protein in the gut — it’s a *process* that occurs in a number of different diseases. These diseases can broadly be divided into a few big buckets — thanks Akshay for brilliantly highlighting these buckets!

Gut loses protein because…

1. Mucosa is more leaky: think about inflammatory and infiltrative disease of the gut — IBD, sarcoid, GI malignancy, etc. 

2. Lymphatics are leaky – either because they’re congenitally dilated (as was the case in our patient) or because the hydrostatic pressure in the lymph system is really elevated — often as a result of elevated *venous* pressure, which backs up into the lymphatics (remember, lymph system is connected to venous system). This is why patients with bad right heart failure, or patients with congenital disease s/p Fontan (high central venous pressure) can develop PLE as a complication. 

Here’s a nice chart (thanks Harry Hollander, for sharing/linking) with an overview of the diseases on the differential:

It’s helpful to return to the model of “nephrotic syndrome of the gut” when thinking of some of the complications/manifestations of PLE. There’s decreased oncotic pressure, so patients may have edema and effusions. As in nephrotic syndrome, there’s a lost of paraprotein, so patients may have humoral immunodeficiency. Also (as in nephrotic syndrome) there may also be an increased risk of thromboembolism, maybe due to combination of loss of clot-busting proteins, as well as association of PLE with states of high venous stasis due to low cardiac output (think of the right heart failure patients). 


Akshay mentioned that his team sent a stool Alpha-1-Antitrypsin level to diagnose PLE. Why A1A…? The answer is absolutely brilliant. Most protein that leaks into the gut gets cleaved away into amino acids. A1A — being a protease inhibitor — is robust, and resistant to cleavage! So you can still detect it in the stool.

If it’s abnormally high in the right clinical context, would be concerning for PLE! 

Physiology of Constrictive Pericarditis

Last week, Stephanie Chen and the amazing UCSF IM CPS Journal Club Crew (Shout out: Smitha Ganeshan, Anna Fretz, Dan Minter, Jack Penner, Jessie Holtzman, Jenn Davis, Andrew Kim, Jordyn Silverstein, Lakshmi Subbaraj) presented a case of an elderly man originally from Southeast Asia who had progressive dyspnea on exertion and leg edema. Cross sectional imaging showed a thickened pericardium, and a TTE and heart cath showed evidence of constrictive physiology. He was ultimately diagnosed with constrictive pericarditis from tuberculosis.

Harry Hollander was our expert discussant, and (no surprise here) was impossible to stump! Some of the residency applicants were able to join, which was really special because the CPS nights are such a great showcase of our IM Residency’s inquisitive (but laid back) spirit.

It was a really fun case, but I realized I had a pretty incomplete understanding of constrictive physiology. This led me down a rabbit hole. Below, I’ve tried to distill the physiology into a few key points. This gets a bit into the weeds, but is hopefully worthwhile…

In constriction, there is rapid ventricular filling in early diastole, followed by a plateau 

A hallmark of constriction is the rapid ventricular filling during early diastole, followed by sharp plateau once the ventricles smack against the rigid pericardium: if you measure the pressure in the RV and LV, this manifests as the “square root” or “dip & plateau” sign. This is nicely visualized in Panel A in the figure below: the dip represents rapid early filling, the plateau reflects equalization of the LV and RV pressures in mid-diastole once the ventricles hit the rigid pericardium. Note that this filling pattern is in contrast to tamponade where the “squeeze” of the pericardial fluid pressure blunts even early filling.

Why is there rapid filling in constriction?  Ventricular filling is driven by the gradient between atrial and ventricular pressure (larger gradient = more rapid filling). So if there’s a relative rise in atrial pressure (compared to ventricle) or a drop in ventricular pressure (compared to atria), you’ll get more filling; in constrictive pericarditis, there is both an increase in the atrial pressure and a decrease in ventricular pressure during early diastole. Below we explore each of these two concepts:

  1. Elevated Atrial pressure:  This is a bit confusing, and I’m honestly not entirely sure why this occurs in constriction. Braunwald mentions that low CO from left heart (discussed more in the section below) induces RAAS which increases effective circulatory volume, which raises atrial pressure. Is it just me or does this feel a bit unsatisfying? I’m not sure if it’s the only mechanism. (I understand the intuition of increased *right* atrial pressure during *inspiration*, as there’s more flow, which we’ll explore more lower in this post, but I’m not certain why there are generally increased left and right atrial pressures). Would love to know if anyone has a more precise explanation, but nevertheless, it is empirically observed that RA pressures are higher in constrictive pericarditis (note the elevated atrial pressure is in panel B of the figure above).
  2. Augmented/accelerated “suction” of ventricle. There is a rapid drop in ventricular pressure in early diastole because of augmented suction. Why is there augmented suction of the ventricles in constriction? Braunwald explains that this is due to the really small end systolic ventricular volume (since systolic function is preserved in constriction); I presume that this is a consequence of the “spring”-like nature of the heart? The more it is stretched, the stronger/faster the force of squeeze (i.e. the Frank Starling mechanism), and so as a corollary, the more that it is compressed, the higher the force of expansion? I’ve also heard explanations that the augmented suction is due to the fibrous pericardium “pulling” the pericardium out at the end of systole/beginning of diastole, but I don’t know if there is any evidence for this). Again curious if I have this wrong, or if anyone has a better explanation of this.

OK, so now we’ve explained rapid early filling followed by plateau with equalization of ventricular pressures.

Another core feature of constrictive physiology is exaggeration of ventricular interdependence.

There is an exaggeration of ventricular interdependence in constriction. 

First let’s review normal physiology. During inspiration, you generate a negative intra-thoracic pressure. This drop pressure is transmitted to the heart, and so the right heart acts like a vacuum, sucking in blood from the IVC. What’s going on on the left side? That negative pressure acts like a vacuum to keep the blood in the lungs, so you get under-filling of the left ventricle. So you now have a situation where the RV overfills a bit, and the LV underfills; this allows the ventricular septum to bulge slightly to the left (this is referred to as “ventricular interdependence”), and obliterate the LV cavity a bit. The combination of these two effects: decreased LV filling and leftward septal bowing leads to a slightly lower LV cardiac output, and your blood pressure drops a few mmHg when you inspire. This is normal.

In constriction, the normal physiology (increased RV filling, leftward bulge of septum, and decreased LV filling) is exaggerated. To understand why, we have to look at what’s happening to the pressures inside the heart.

This is really nicely illustrated in the figure below (from a great JACC Review on Constrictive Physiology). In constrictive pericarditis, the thick pericardium prevents the atria and ventricles from “feeling” the negative pressure in the thorax during inspiration; importantly, the IVC and pulmonary veins are external to the constricted pericardium, so they do feel the negative intrathoracic pressure (just like normal). In the illustration below, there is a pressure gradient between the pulmonary veins and the left atrium of 10-4=6 favoring filling of the LA during apnea. During inspiration, the intrathoracic pressure drops (in this illustration from 0 –> -5) and the pressure drop gets perfectly transmitted to the pulmonary veins (pressure drops from 10 –>5) and somewhat imperfectly tansmitted to the left atrium (in this illustration, there’s a 4 point drop from 4 –>0 ). So in the normal heart, the pressure gradient between the pulmonary veins and the left atrium during inspiration drops from 10-4=6 to 5-0=5.

In constriction, during inspiration the LA wall doesn’t feel any of the drop in intrathoracic pressure (stays from 4 –>4), but the pulmonary veins “feel” the drop (from 10 ->5, just like normal). So in a constricted heart, the pressure gradient between the pulmonary veins and the left atrium is 5-4=1 (compared to 5 in the normal heart), and a lower gradient means less filling compared to normal (which allows even more septal bulging to the left). You can use the same logic to work out why left sided filling actually improves with expiration. In constrictive pericarditis, there is an exaggeration of the drop in LV filling during inspiration.

You may recall that, in tamponade, you also get an exaggeration of RV filling, LV under-filling and leftward septal bulging during inspiration, which manifests as a greater than normal drop in arterial pressure with inspiration: pulsus paradoxus; apparently the pulsuss paradoxus phenomenon is less common in constrictive pericarditis even though the physiology is the same (according to Lily this is because the bowing effect is not as pronounced as in tamponade). (If you do see pulsus paradoxus in a patient with constrictive physiology, it may be due to effusive-constrictive disease…which is a combination of constriction/tamponade physiology…but let’s not complicate things…)

Phew. Ok really getting into the weeds here. So now we’ve explained the square root sign and the exaggeration in ventricular interdependence; if you agreed with the logic described above, you should be convinced that, because there’s more RV filling during inspiration, you should see a drop in JVP with inspiration. But, alas, no. This is where constriction gets even more confusing! There is a RISE in JVP during inspiration in constriction, known as the Kussmaul sign. What the heck?!

There is a positive Kussmaul Sign 

This is counter-intuitive, but, again, has to do with the fact that the atria and ventricles (including on the R side) are “shielded” from the drop in intrathoracic pressure during inspiration in constrictive physiology. In contrast, the SVC is in the thorax and does feel the drop in pressure; so if you measure the gradient between the SVC and the R atrium you’ll see that blood will be “stuck” in the pressure “sink” of the SVC during inspiration– recall that since the internal jugular vein is a branch of the SVC, it will swell up and increase during inspiration — this is the Kussmaul sign. Note though that there’s still a net increase in total R-sided filling (confusing!), but it comes from the IVC; during inspiration, the diaphragm squeezes down on the abdomen, which squeezes the IVC, which forces blood into the Right atrium. See this great physiology paper from the ‘80s if you’d like to ponder this more. (Note that on Echo, there is holodiastolic flow several in the hepatic veins during expiration, which is fairly sensitive/specific for constriction; you can work through the logic for this by using the same intuition as above).

Distinguishing Constriction vs Restriction

One more thing that often comes up is that restrictive physiology can sometimes look similar to constrictive physiology (both can have square root sign). There’s a lot of nuance to this (echo helpful as well), but one way to distinguish the two is with simultaneous pressure measurements of the LV and RV during catheterization. I’ve shown an example below, taken from a great Grand Rounds presentation at UW in 2004. Note that the left panel shows what happens in restrictive physiology: LV and RV pressures drop together (known as “concordance”) with inspiration. This is because of negative intra-thoracic pressure generated by inspiration, and transmitted to both ventricles. Note that the pressures rise together with expiration (because of increased intra-thoracic pressure). In contrast, on the Right panel, note that in constrictive physiology, RV pressure increases with inspiration (because of more filling and not “feeling” the intrathoracic pressure) but LV pressure drops with inspiration (less filling and septal bulge), and vice-versa with expiration. The fact that the RV and LV do opposite things means they are “discordant.” 

Phew…OK…. that was a lot. This topic has always been super confusing to me, but hopefully this post helps. The articles that I linked in the post are also really helpful, as well as this fantastic episode of Cardionerds, which had some really great visualizations.

Coronary ct angiography: A mini review

Last week Alissa Kleinhenz presented a case of a patient she took care of in primary care clinic who presented with a 2 year history of chest pain. His pain seemed pretty classic for ischemic heart disease: he was the right age, the pain was substernal, it worsened with activity and it improved reliably after a few minutes of rest. Given that he had such a high pre-test probability of coronary disease, he was referred for a diagnostic left heart catheterization. He was hesitant to get a LHC, but was amenable to coronary CTA.

Despite having such a good story for angina, his CTA was remarkably normal — and coronary disease was essentially excluded.

We spent some time talking about the use of coronary CT as an adjunct (or alternative?) to stress testing when evaluating a patient with new stable chest pain. Here we take a little bit of a deeper dive.

Who is coronary CTA appropriate for?

Let’s start by taking a look at the AHA/ACC Guidelines (see below).

Note that this is actually from the 2014 Guidelines on Valvular Disease, in the subsection about assessing for CAD before consideration of valvular surgery. Unless I’m mistaken, the 2012 AHA/ACC Guidelines on Management of Stable Ischemic disease have not yet been updated with guidelines on the use of CTA. Importantly, these guidelines were written before two of the largest randomized trials (see below) were published.

What is the evidence?

There are two recent large RCTs which have evaluated the use of coronary CT in stable chest pain: PROMISE (2015) and SCOT-HEART (2018).

In the PROMISE trial, there was no difference in the primary endpoint (a composite of death, non-fatal MI, hospitalization for unstable angina, or procedural complications) when patients with stable chest pain were evaluated with Coronary CT compared with stress testing (one criticism of the study though, is that event rates were pretty low — only about 3% of the patients in both arms had an event at 2 years, so there may have been some power issues).

In the SCOT-HEART trial, the investigators asked whether the use of coronary CT plus standard care (which included stress testing in ~85% of the patients in both arms) to workup stable chest pain improved a composite endpoint (death from coronary disease or non-fatal MI) compared with standard care alone. The patients in the coronary CT arm had a statistically significant reduction in the composite outcome (seems that this was most driven by a reduction in non-fatal MI). The thought is that more of the patients in the coronary CT arm would have been “identified” to have coronary disease (because it’s such a sensitive technique), and therefore more of these patients were started on aspirin/statin, thus leading to lower rates of CV death / MI. (Interestingly though, John Mandrola counter-argues that while this is a tempting explanation for the difference in event-rates, in reality the number needed to treat on preventive therapy is much higher [~50] than what would be needed to account for the reduction seen in this study.)

So what?

I find it hard to operationalize the results of these trials, but I think this Pro and Con series from the ACC helps highlight some of the strengths and limitations in using coronary CT as an initial test to evaluate patients with stable chest pain.

One of the major criticisms of coronary CT is that, unlike in functional/exercise stress testing, you don’t know whether the lesions that are identified anatomically at rest are physiologically meaningful when the heart is under higher demand. This may be evolving as people develop some interesting computational fluid dynamics approaches to compute non-invasive FFR on coronary CT lesions.

Coronary CT is very sensitive test which can detect lesions that can be missed on a nuclear study (see figure below). But one cost of increased sensitivity is that the use of coronary CT tends to lead to more invasive tests downstream (both invasive coronary angiography and revascularization procedures). Proponents of coronary CT point to the fact that there are more revascularization procedures because more patients with chest pain are correctly phenotyped (i.e. they have more intervenable lesions identified) while skeptics argue that more re-vascularization is not useful (potentially harmful and more costly), if it doesn’t lead to a reduction in a more robust endpoint like CV mortality.

Excited to see how this debate evolves as the technology improves and we get more data!

Taken from the “Pro” article referenced in the text. This figure illustrates two scenarios (“balanced” ischemia and non-obstructive but “high risk” plaque) where a nuclear study would be normal, but a CTA would identify pathology).


Last week at “Physiology Report” we shared a case of a 50 year old man with a history of hypertension who was brought to the ED with confusion and chest pain. His EKG showed an anterior STEMI, and he was taken to the cath lab, where he was found to have vasospasm of his left anterior descending artery. His ST elevation resolved with administration of intracoronary vasodilators, and he was taken to the CCU.

In the CCU, an ultrasound showed a large pericardial effusion and aortic regurgitation, so the team was concerned about dissection. A stat CT with arterial contrast revealed a Stanford Type A aortic dissection (a coronal slice from the patient’s CT below shows the dissection flap in the arch).

The patient was referred for emergent surgical repair. We focused on the medical management of a patient with an aortic dissection — either while awaiting surgical management or in the case of a Type B dissection, where medical management is the first line.

We spoke in particular about how the mainstay of management isn’t quite blood pressure control — more specifically, the shear forces in the aorta that propagate an intimal tear seem to be related to the change in pressure over time — the “dp/dt“. This was demonstrated in in vitro physiology experiments from the 1970s (see abstract below).

This is an aortic pressure tracing. The shear force in the aorta is determined by the change in pressure over time (dp/dt) during the upstroke.

This is an important point to understand — not just because it serves as justification for including calculus in our premed requirements (lol), but because it informs why dropping the blood pressure alone is not sufficient. In fact, it can make things worse! In the picture below, the plot on the left demonstrates this point. Compared to our old tracing (red), the purple tracing reflects what happens when you administer a vasodilator in isolation (e.g. nitroprusside). Note that the blood pressure goes down, but the dp/dt actually increases — this is because vasodilators induce a reflex chronotropy (i.e. tachycardia) and reflex inotropy in the LV in response to the drop in pressure. On the other hand (right sided panel), note that beta blockers blunt the dp/dt by slowing the heart rate.

The plot on the Left shows that the administration of vasodilators increases the dp/dt, while the panel on the Right shows that beta blockers drop the dp/dt.

The Takeaway: in a hypertensive patient with aortic dissection, for medical management start with a beta-blocker (esmolol is a good choice since it has quick on-off) to lower the shear stress.

Target a HR of 60 and a systolic BP less than 120 (see: 2010 ACCH/AHA guidelines). If you need additional agents to get there, it’s ok to add on a vasodilator like nitroprusside, only if you’ve got a beta blocker on board.

When do we start antiretroviral therapy in HIV patients with opportunistic infections?

On Monday and Wednesday, Megan and Jack presented cases of patients with uncontrolled/newly diagnosed HIV who presented with PJP pneumonia.  One question that came up was, when do we start antiretroviral therapy (ART) in these patients?

For the most part, the answer is immediately.  Early ART initiation results in less AIDS progression and death with little increase in adverse events.  An RCT published in 2009 in PLoS ONE evaluated almost 300 subjects who presented with either an AIDS-defining opportunistic infection (OI) OR serious bacterial infection (BI) + CD4 < 200 (1).  People were randomized either to immediate start of ART or deferred start after treatment of OI was completed.  The breakdown of infections among patients was: 63% PJP, 12% cryptococcal meningitis, and 12% bacterial infections.  The study found that 14% of patients in the early arm had progression of disease or died, compared to 24% of patients in the deferred arm.  Median time to ART was 12 days after OI treatment in the early arm and 45 days after OI treatment in the deferred arm.  Rates of confirmed immune reconstitution inflammatory syndrome (IRIS) were at 5.7% for the early arm and 8.5% for the deferred arm and was not statistically significant.

The primary exception to starting immediately is in a patient with cryptococcal meningoencephalitis.  The trial described above was not powered to look solely at patients with cryptococcal meningitis, and other trials looking at early versus delayed initiation of ART in patients with cryptococcal CNS disease consistently lower rates of survival and increased rates of IRIS with early initiation (2-4).  ART is also frequently delayed in the setting of other CNS opportunistic infections (ex. CNS tuberculosis) although evidence is less robust here.

Takeaway: Initiate ART immediately in patients presenting with opportunistic infections (especially PJP) EXCEPT for those who have CNS cryptococcal disease, and consider delaying in other CNS infections such as tuberculosis.  Have a low threshold to send a serum cryptococcal antigen in patients with HIV before starting ART, particularly those with neurological symptoms.


  1. Zolopa A, Andersen J, Powderly W, et al. Early antiretroviral therapy reduces AIDS progression/death in individuals with acute opportunistic infections: a multicenter randomized strategy trial. PLoS ONE. 2009;4(5):e5575.
  2. Makadzange AT, Ndhlovu CE, Takarinda K, et al. Early versus delayed initiation of antiretroviral therapy for concurrent HIV infection and cryptococcal meningitis in sub-saharan Africa. Clin Infect Dis. 2010;50(11):1532-8.
  3. Boulware DR, Meya DB, Muzoora C, et al. Timing of antiretroviral therapy after diagnosis of cryptococcal meningitis. N Engl J Med. 2014;370(26):2487-98.
  4. Bisson GP, Molefi M, Bellamy S, et al. Early versus delayed antiretroviral therapy and cerebrospinal fluid fungal clearance in adults with HIV and cryptococcal meningitis. Clin Infect Dis. 2013;56(8):1165-73.

When should we think about diffuse alveolar hemorrhage and how do we treat it?

On Monday, Jackie presented a case of a middle aged woman with anti-phospholipid syndrome who presented with cough and dyspnea and was ultimately diagnosed with diffuse alveolar hemorrhage (DAH).

When should we consider diffuse alveolar hemorrhage on the differential? (1)

DAH is usually acute onset but can sometimes have a stuttering course over weeks.  The most common symptoms are cough (hemoptysis is present in about 2/3 of cases), fever, and dyspnea.  There are usually diffuse or patchy opacities on CXR and diffuse GGOs +/- consolidations on CT chest.  Patients usually have a hemoglobin drop on lab check.  DAH can be diagnosed on bronchoalveolar lavage if serial lavage aliquots become progressively bloodier.

Three different types of disease processes can lead to DAH, reflected in the histology: pulmonary capillaritis, bland pulmonary hemorrhage, and diffuse alveolar damage (the histologic companion to acute respiratory distress syndrome).  Thinking through these categories is helpful in determining the treatment course, as we’ll see below, but it also helps you reason through who is at risk of developing DAH.  Pulmonary capillaritis can be a complication of systemic vasculitis, rheumatic disease, and various drugs.  Bland hemorrhage is usually related to bleeding disorders or blood thinners, although it can also be associated with elevated left ventricular end diastolic pressures and some connective tissue disorders.

Takeaway: Suspect DAH when a patient with specific risk factors (elevated bleeding risk, rheumatologic disease, or systemic vasculitis) presents with dyspnea, a hemoglobin drop, and diffuse GGOs/opacities on imaging.  The presence of hemoptysis is helpful but this symptom can be absent in many cases.

How do we treat DAH 2/2 pulmonary capillaritis?

Patients who present with DAH that is presumed 2/2 pulmonary capillaritis should be pulsed with glucocorticoids.  The choice of additional immunosuppressive therapy differs depending on the underlying disease, but usually involves cyclophosphamide, rituximab, and/or plasma exchange.  Most of the evidence for treating DAH 2/2 pulmonary capillaritis comes from patients with ANCA-associated vasculitis, mixed cryoglobinemia, and SLE and is extrapolated to other causes of capillaritis.

A randomized controlled trial published last month in NEJM evaluated the efficacy of plasma exchange and two different dosing regimens of glucocorticoids in severe ANCA-associated vasculitis, either with diffuse pulmonary hemorrhage or severe kidney involvement (2).  There were 4 treatment arms: plasma exchange + standard dose steroids, plasma exchange + reduced dose steroids, no plasma exchange + standard dose steroids, and no plasma exchange + reduced dose steroids.  All patients in the trial received induction therapy with either cyclophosphamide or rituximab (>80% got cyclophosphamide) and high dose IV methylprednisolone for 1-3 days.  Notably, only about 8-9% of participants in each arm had severe pulmonary hemorrhage.  Rates of the primary composite outcome (death from any cause and end stage kidney disease) were similar and non-inferior in all groups.

Takeaway: Treat DAH 2/2 capillary capillaritis with pulse dose steroids and usually an additional immunosuppressive agent (most evidence is with cyclophosphamide) if severe disease.  IVIG and plasma exchange can be considered in severe cases as well, although plasma exchange may not be effective in severe ANCA-associated vasculitis.


  2. Walsh M, Merkel PA, Peh CA, et al. Plasma Exchange and Glucocorticoids in Severe ANCA-Associated Vasculitis. N Engl J Med. 2020;382(7):622-631.


Recent Updates in HTN

This week, we’re highlighting a couple of recent papers on treatment of hypertension which have come up in discussions.

The first is the Hygia Chronotherapy Trial (1) which was published in October of 2019 in the European Heart Journal.  It was a multi-center, randomized controlled trial in Spain that evaluated whether taking anti-hypertensives at bedtime or upon awakening was better for cardiovascular disease risk reduction in adults with essential hypertension.  In short, patients taking the medications at bedtime had a much lower hazard ratio for the composite primary outcome of cardiovascular death, MI, coronary revascularization, heart failure, and stroke (HR 0.55).  All of the individual components of the primary outcome were also significant, including CV death alone (HR 0.44).  There are certainly limitations to this trial: it was non-blinded and it was unclear why the analyses were adjusted, since randomization usually washes out confounders and renders adjusting analyses unnecessary.

Takeaway: Despite the limitations, the effects of this intervention are very robust and given the minimal risks in taking anti-hypertensives at night (except for maybe diuretics), it makes sense to ask patients to take their blood pressure medications at night.

The second is an observational cohort study just published in JAMA IM last week comparing cardiovascular outcomes and safety of hydrochlorothiazide versus chlorthalidone (2).  They used multiple databases to study over 700,000 patients and used propensity scoring to try to control for confounders and match the two groups.  The authors found that there were no significant differences in the risk of MI, hospitalized HF, or stroke, but chlorthalidone was associated with much higher rates of electrolyte abnormalities (hypokalemia, hazard ratio, and hyponatremia) and renal disease.  Important to note that the median “at risk,” or exposure, period was only 92 days. The overall exposure period in this study was short and there may not have been enough follow up time to show cardiovascular benefit.  This is also an observational cohort study which is prone to bias — patients who were higher risk at baseline may have been prescribed chlorthalidone by providers who believe in its benefit over HCTZ.

Takeaway:  There was previously little direct evidence that HCTZ specifically improves cardiovascular outcomes.  The ALLHAT trial in 2002 (3) studied chlorthalidone, not HCTZ, and the 2017 AHA/ACC guidelines suggest chlorthalidone as the preferred thiazide diuretic.  This paper suggests that chlorthalidone has a worse side effect profile and may not have the benefits over hydrochlorothiazide that we previously believed, but a follow up RCT is needed!


  1. Hermida RC, Crespo JJ, Domínguez-sardiña M, et al. Bedtime hypertension treatment improves cardiovascular risk reduction: the Hygia Chronotherapy Trial. Eur Heart J. 2019;
  2. Hripcsak G, Suchard MA, Shea S, et al. Comparison of Cardiovascular and Safety Outcomes of Chlorthalidone vs Hydrochlorothiazide to Treat Hypertension. JAMA Intern Med. 2020
  3. Major outcomes in high-risk hypertensive patients randomized to angiotensin-converting enzyme inhibitor or calcium channel blocker vs diuretic: The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT). JAMA. 2002;288(23):2981-97.