I’ve mentioned before that I think the PET-CT is an exciting new addition to the workup of S.aureus bacteremia. Well, I recently did some research on the relative value of TTE and TEE for the same indication, and was surprised by what I found (or rather, what I didn’t find), so I thought it merited the discussion below.
Consider the following: You’re asked to see a patient for MRSA bacteremia. He’s in his mid-80s and has diabetes complicated by peripheral neuropathy and peripheral vascular disease. He presents with a several-days’ history of fever and a painful purulent plantar right foot wound; MRI shows underlying abscess and osteomyelitis. Orthopedic surgery performs incision and drainage + wound debridement and cultures from pus, tissue, and bone all grow the same MRSA. Blood cultures clear on day 3 after surgery; however, he continues to have low-grade temperatures and leukocytosis for another two days. Eventually he’s found to have a small deep right thigh abscess that is drained by interventional radiology, after which his temperature and WBC count normalize. After a thorough history and physical examination you’ve been unable to identify any other source of infection. The transthoracic echocardiogram you ordered did not identify any vegetation or cardiac abscess. Should you request a transesophageal echocardiogram?
As a fellow I was all over the place with regards to my approach to echocardiography in SAB, in part because my mentors had different approaches to the issue. We weren’t alone in this: a recent study of variation in North American ID consultants’ practice in treating SAB found that 19% of clinicians obtain a TEE in all cases while 10% choose not to obtain any echocardiography in at least some cases. Even among the majority, who decide on a case-by-basis, there is only consensus that TEE should be obtained when the patient has had persistently positive cultures (86%) or has not timely clinical improvement (81%).
So, if you have someone with a complicated S. aureus bacteremia and a negative TTE, should you order a TEE? Before we look at the literature, let’s consider the specific question we ought to be asking. I can think of two:
1. In patients with complicated S.aureus bacteremia in whom a TTE is negative, does a obtaining a TEE lead to important changes in the management of the patient’s infection?
2. In patients with complicated S.aureus bacteremia in whom a TTE is negative, is getting a TEE associated with improvement in outcomes that matter to patients (e.g. lower rates of mortality, rehospitalization, or serious complications of infection)?
Neither of these questions are “does the TEE find more valvular lesions.” More on that later.
On to the literature: the 2011 guidelines for the management of infections due to Methicilin-resistant Staphylococcus aureus recommend echocardiography and preferential use of transesophageal echocardiography whenever possible. This recommendation is based on four small observational studies by Fowler, Sullenberger, Abraham, and Reynolds, which reported rates of vegetation/endocarditis identification by TEE and/or TTE. Three of these studies showed that TEE detected vegetations at a higher rate, with absolute increases in the positivity rate ranging from 11% to 14% (the fourth study, by Abraham, does not compare the rates of IE diagnosis with TTE vs TEE, and I’m not convinced it actually supports the statement that references it in the 2011 guidelines). The reported relative sensitivity of the TTE for infective endocarditis versus the TEE, given in the studies by Fowler and Reynolds, was only 32% and 55%, respectively (on the other hand, the NPV for IE with TTE, only calculable from the data from Fowler, was 88%, because most folks with bacteremia don’t have endocarditis).
So, TEE finds many more vegetations and makes more diagnoses of IE. This begs the question – are there any data indicating that these additional diagnoses matter? Recently, I asked twitter why it was worth performing an echocardiogram in complicated S.aureus bacteremia. Here’s how folks replied:
For those who answered “help set the duration of treatment,” I’ll point out the 2011 guidelines recommend a 4-6 week course of therapy for complicated bacteremia. Moreover, there are no studies comparing the outcome of either S.aureus complicated bacteremia OR endocarditis with four versus six weeks of therapy. Also, complicated S.aureus bacteremia is plenty deadly on its own: the one-year mortality rate of S. aureus bacteremia has consistently been reported at 20-60%, and endocarditis has not been consistently reported as a risk factor for mortality among patients with SAB (see work by Ekeson, Yilmaz, Yahav, Yildiz, Gotland, Jacobsson, Yaw, and Hanses). So, I’m not sure that deciding to prescribe 4 weeks for complicated bacteremia vs 6 weeks for endocarditis on the basis of an echocardiogram actually reflects evidence-based practice.
There were a few great additional thoughts in the comments, which ranged from “it’ll help define the need for dental prophylaxis” (OK, but for the purposes of this discussion that’s small potatoes. Figure that a third of these people are going to be dead in a year, and TTE has a NPV of 88% for IE – what’s the NNT going to be for getting a TEE to determine prophylaxis to prevent one dental misadventure?) to “it helps determine the prognosis” (OK, but the prognosis of S. aureus bacteremia is terrible anyway, and as those studies above indicate, the main predictors of outcome are other comorbidities and sepsis / acuity of illness during the hospitalization) to “it annoys cardiology” (props for honesty!).
I think the most compelling answer, chosen by nearly 80% of respondents, was to determine the need for surgery (ongoing bacteremia or fever despite antibiotics beyond 7 days being an indication for valve surgery in IE based on the 2016 AATS consensus guidelines). What are the criteria for valve surgery in infective endocarditis? According to those same guidelines:
· Severe valve dysfunction resulting in symptoms of heart failure
· Prosthetic valve endocarditis
· Paravalvular abscess, cardiac fistulas, and other penetrating/destructive lesions
· Recurrent systemic embolizations
· Large mobile vegetations (>10mm)
· Persistent bacteremia or fever after 7 days of appropriate antibiotics, or recurrent bacteremia after a full course of treatment with no other identified source of infection
· IE due to highly resistant organisms including fungi
Ok, so what are the data on the incremental value of a TEE over a TTE for identifying an indication for surgery? Here we’re really talking about severe valve dysfunction, cardiac/perivalvular abscesses or other penetrating lesions, and large mobile vegetations. The good news for the TTE is that it’s got a much higher specificity for large vegations (see Reynolds et al, cited above). On the other hand, the same study found that TEE revealed indications for surgery missed on TTE in 5/51 (10%) of patients with positive TTEs – one mitral valve leaflet aneurysm with aortic valve prolapse, one torn chordae with flail mitral leaflet, one aortic valve perforation, and two pacer wire vegetations (the latter two are arguably not relevant given that the pacer wires ought to have been removed based on the clinical history anyway). In the study by Fowler, TEE mostly identified small vegetations missed by TTE, but did find surgical criteria missed by TTE in 4/103 patients with S. aureus bacteremia (2 aortic root abscesses, 1 valve perforation, and one new severe valvular dysfunction); however, whether these patients went to surgery and their ultimate outcomes were not reported. A small retrospective observational study out of Thailand found that TEE identified perivavular abscesses in 5/24 patients who had previously undergone TTE. More recently, in a study of 196 patients with endocarditis who underwent TTE and TEE, 14/196 (7%) had newly detected perivalvular abscesses, and 75% of the prosthetic valve endocarditis cases (18/24) were only detected by TEE; however, this doesn’t tell us the incremental value of TEE over TTE in all comers with S.aureus bacteremia.
It’s also worth noting that valvular lesions evolve over time, and that repeating a TTE will also catch endocarditis missed on an additional study. So another question worth asking - which I could find no published data to answer - is how better, if at all, a TEE is to waiting a few days and just getting another TTE.
What about mortality? Here too there is a scarcity of data; moreover, what data does is exist is confounded by the “bundle intervention” phenomena: there are studies showing improved outcomes with ID consultation in S. aureus bacteremia, wherein more TEEs are performed, but patients managed by ID also have more echocardiograms done in general, more followup blood cultures, indentification and removal of more noncardiac sources of infection, more frequent use of beta-lactams for MSSA, etcetera (see work by Turner). The only study I could find showing an independent patient benefit of undergoing TEE was just published this month by Holland et al out of Australia, who reviewed 204 cases of patients ultimately diagnosed with infective endocarditis (less than half of which were shown to be due to S.aureus), finding that lack of TEE was an independent predictor of the composite outcome of death + rehospitalization (HR 2.12; 95% CI 1.27-3.53).
So, should every patient with complicated S. aureus bacteremia get a TEE? Maybe? It certainly seems to find more vegetations, but data is scant that getting a TEE after a TTE leads to more cardiac surgery / source control, or more critically, that it reduces rates of mortality or major infectious complications. This is an important distinction because TEE is a not only costly but invasive procedure with an ill-defined risk profile, so if the NNT for meaningful benefit is low it might be outweighed by the NNH for serious procedural complication.
I’m uneasy with the ID community’s uncritical embrace of the TEE (note that performing a TEE was proposed as a quality measure in suspected or proven complicated S.aureus bacteremia by a recent Delphi panel paper published in JAC), particularly because PET-CT – a safer procedure that actually does have data indicating that it leads to more source identification and reduced mortality in S.aureus bacteremia – has not seen the same interest. This is especially irksome because the major rate-limiting step in increasing access to / insurance coverage of PET-CT for S.aureus bacteremia is getting it recommended by the guidelines of major medical societies.
(This is my way of saying it’s going to be a crying shame if the next iteration of the IDSA S.aureus guidelines don’t recommend PET-CT as part of the workup in cases of persistent bacteremia and/or suspected disseminated infection)
I’ll end by answering my own question: in which patients with S.aureus bacteremia would I demand a TEE? First, among those with uncomplicated bacteremia, there are several studies that have identified a low-risk subset for whom TEE has little additional diagnostic value (I won’t go into those here, but for reference, see these papers by Showler and de la Vega and these reviews by Holland and Heriot). The only caveat I’ll add here is that I would probably still get a TEE in a low-risk patient if the TTE was reported as technically difficult or obtaining only limited windows, because that suggests that the sensitivity of the study is compromised and the incremental benefit of a TEE is likely to be more than it would otherwise.
Among those with complicated S.aureus bacteremia, it seems to me that the main reason to get an echo is to find something to cut on, so I don’t think it makes sense to insist on a TEE unless you’ve established that (1) the patient would be a candidate for cardiothoracic surgery and (2) the patient would actually be willing to undergo said surgery if you found something. But the data sure are lacking. We need studies similar to the ones by Fowler and Holland, studies that examine large cohorts of patients with S. aureus bacteremia and a negative TTE and identify how often and in whom the TEE leads to surgical intervention, and whether the improvement in meaningful outcomes like rehospitalization and death outweigh the TEE’s complications and cost.
And now, on to the literature: (article descriptions to come - for now I must fly to IDWeek 2019!)