December 3rd, 2013
Estimating Prosthesis-Patient Mismatch for Patients Undergoing Bovine Pericardial AVR
Jyothy Puthumana, MD,MS
Harlan Krumholz and John Ryan interview Jyothy Puthumana about his research group’s Circulation: Cardiovascular Imaging study, for which they compared prosthesis-patient mismatch (PPM) prevalence and its impact on survival.
A total of 614 patients with normal systolic function undergoing bovine pericardial aortic valve replacement (AVR) were evaluated using 3 modalities: (1) the ASE guidelines-suggested algorithm (ASE PPM); (2) the manufacturer-provided charts (M PPM); and (3) the echocardiographically measured, body surface area-indexed, effective orifice area (EOAi PPM) measurement. PPM prevalence using the algorithmic-ASE approach was low and correlated well with manufacturer-provided PPM. Independent of assessment method, PPM was not associated with medium-term mortality.
Krumholz and Ryan: PPM is defined as an effective orifice area of the implanted prosthetic valve that is too small for the patient, causing high gradients through a normally functioning valve.
By the ASE algorithm (peak gradient > 3 m/s, dimensionless index > 0.25, and acceleration time <100 ms), you estimate that about 4% of patients have severe (effective orifice area ≤0.6 cm2/m2) PPM. You also find no implication for outcomes. Should we be concerned about mismatch and make it a never event – or is it not really a problem? Or do we not know since you only had 22 patients with severe mismatch?
Puthumana: In our population of patients with normal systolic function who had a pericardial tissue valve implanted (a very homogenous group), the prevalence of PPM is very low by either the ASE algorithm estimates or by the manufacturer estimates. This correlated very well with excellent intermediate-term clinical outcomes, therefore, we feel that clinically relevant PPM is not being missed using this approach.
EOAi estimates for this cohort identified moderate and severe PPM in a significant number of patients (63% of the cohort), without prognostic relevance, causing us to question EOAi as a routine method of PPM assessment in patients with normal systolic function who undergo AVR with a pericardial tissue valve.
Krumholz and Ryan: You evaluated through three different modalities. In an echo lab that can only do one modality, which would you recommend and why?
Puthumana: Based on our outcomes in patients with normal LVEF who undergo AVR with pericardial tissue valves, we feel that the ASE algorithmic approach to quantify the severity of PPM has the best clinical relevance. This correlated with the manufacturer provided estimates, which makes valve selection in the OR easy and practical, and this approach would significantly decrease the risk of PPM and its associated complications during follow up.
EOAi estimates, on the other hand, seemed to overestimate the number of the patients who were identified to have PPM, without adverse prognosis during intermediate duration of follow up.
Krumholz and Ryan: You submitted your paper on May 2, 2013, and it was not accepted until July 26, 2013. What took the paper so long to be accepted?
Puthumana: After initial submission, based on reviewers’ comments, we updated the paper to include near complete follow up for an additional two years and included re-operation for AVR as one of the outcomes. These changes warranted collection of additional data on a few patients, through telephone follow up, etc.
Once this was complete, we performed the data analysis that reconfirmed excellent clinical outcomes for a longer term of follow up and re-submitted the paper.
I have not yet read the article. However, it appears that the greater than sign for the DI and the less than sign for the acceleration time should be reversed for the definition of severe mismatch. The fact that the echocardiographically derived AVA overestimated the prevalence of PPM suggests that the LVOT measurements may have been under measured ( as the AVA is simply the dimensionless index multiplied by LVOT area). How often was the acceleration time < 100 but the DI < 0.25?
The outcome follow-up is obviously still short. How about LVH regression or diastolic function?