CONCLUSIONS
Our study examined outcomes of patients bridged to dLVAD with ECMO for severe cardiogenic shock, compared to less ill patients bridged with IABP or no tMCS. ECMO patients had a longer preoperative length of stay, more frequent mechanical ventilation, higher transaminases, lower albumin and hemoglobin, higher PCWP, lower PAPi, and more concomitant procedures. That said, operative and CPB times were similar amongst groups and reoperation rates were low. The postoperative length of stay in the ECMO group was slightly longer, though acceptable. We observed two perioperative ECMO deaths (18%), with exceptionally low early mortality in the other groups. Differences were not significant, but this highlights the greater severity of illness in ECMO recipients. Of note, longer-term outcomes of patients bridged with either form of tMCS were acceptable and comparable to those with no tMCS.
Use of ECMO as bridge to dLVAD remains controversial, as evidenced by the recent INTERMACS analysis of Ton et al.2 They studied 2013-2017 registry data, which included 1138 patients bridged to dLVAD with ECMO and 3901 bridged with IABP. Their conclusion was that the exceptional acuity of ECMO-bridged dLVAD patients merits an even more severe profile (“INTERMACS 0”).
Three-month survival in that study was 76% ECMO, 88% IABP, and 91% no tMCS, while 12-month survival was 67% ECMO, 79% IABP, and 82% no tMCS. We observed similar survival in our IABP and no tMCS cohorts (100% and 94% at 3 months; 88% and 86% at 12 months), but saw improved survival in the ECMO cohort (81% at 3 and 12 months). They observed increased early bleeding (19 EPPM) and ischemic CVA (2.8 EPPM) in ECMO patients. By contrast, our ECMO cohort had an early bleeding rate similar to that of non-tMCS patients (10.7 and 9.8 EPPM) and lower than the INTERMACS no tMCS cohort (13.2 EPPM). We observed more early bleeding (19.9 EPPM), driven by increased GI bleeding (10.7 EPPM), in the IABP cohort. No early ischemic CVA occurred in ECMO patients, compared to 2.8 EPPM in the INTERMACS analysis. One early hemorrhagic CVA occurred in an ECMO patient, for a rate of 3.55 EPPM. Late events were similarly rare in our analysis, with the exceptions of more infections in the IABP group and more hemorrhagic CVAs in the no tMCS group.
Poor ECMO survival in the INTERMACS series is likely driven by the high prevalence of biventricular support (22% of ECMO patients vs. 5% IABP and 3% no tMCS). Patients with biventricular support had reduced 3-month (61-69%) and 12-month (50-57%) survival. By comparison, none of our patients required durable BiVAD. Two patients (1.2%; 1 ECMO, 1 non-tMCS) required temporary RVAD, one of which expired in the perioperative period.
Similarly, Tsyganenko11 reported a large single-center series (100 ECMO to dLVAD patients) with 38% operative and 57% 12-month mortality. One third required an RVAD. Shah9described 68 patients bridged with non-IABP tMCS, including 22 ECMO. They found that ECMO outcomes were similar to non-tMCS INTERMACS 1 patients despite improved hemodynamics and end-organ function in the ECMO group. In this series, 21% of tMCS and non-tMCS INTERMACS 1 patients required RVAD, vs. 2% of INTERMACS 2-3. Twelve-month survival was 70% tMCS, 77% non-tMCS INTERMACS 1, and 82% INTERMACS 2-3 (p<0.001).
In contrast, Han2 showed very good 12 month survival for both ECMO-bridged (78%; N=18) and non-ECMO INTERMACS 1 (88%; N=17; 47% IABP), which they attribute to 46% of patients being transplanted within <12 months. Our 12-month transplantation rates were 32-41%.
ECMO is our primary tMCS for severe cardiogenic shock because we are a high-volume (~200 cases/yr) center with protocolized management. Patients are routinely extubated and mobilized7,8, allowing rehabilitation prior to dLVAD. We routinely await renal and hepatic recovery prior to dLVAD, even if it prolongs ECMO support. Our median ECMO duration was 10 days, and 4 patients (36%) were supported >14 days, all of which survived to discharge.
We acknowledge this may be contrary to current literature. Cheng12 reported better early survival in patients transitioned to durable MCS after <4 days versus longer or not at all. Tsyganenko11 found >7 days of ECMO was an independent risk factor for mortality. Durinka13 reported using longer-duration support (mean 12.1 days) to await normalization of end-organ function before dLVAD. However, they found much poorer survival for patients supported >14 days (25% vs. 92% <14 days).
We acknowledge that ECMO is not risk-free. Complications include bleeding, limb and spinal-cord ischemia, strokes, compartment syndrome, and cannulation site infection14. However, we believe some of the morbidity and mortality reported in the above-cited studies is due to prolonged intubation, which may impair RV function. In the studies that report mechanical ventilation for ECMO-bridged patients, the incidence ranges from 59-86% with RVAD usage ranging from 21-29%10,11,15. In our cohort, 2 patients (18%) were intubated at dLVAD insertion. One of these expired early from multiorgan failure, and the other required VV ECMO due to persistent hypoxemia. These patients were supported six and eight days, but may not have been sufficiently optimized. Finally, one ECMO patient (9%) required an RVAD. This patient was high risk due to prior chest radiation, three prior sternotomies, and biventricular dysfunction (pre-ECMO PAPi 1.2).
We also believe standardizing to the LTHS approach improved outcomes in ECMO-bridged dLVAD recipients. It has been postulated that leaving the heart in its natural position, avoiding RV compression, and leaving the pericardium intact reduce post-LVAD RV distention and failure16. We have previously shown that the LTHS approach improves outcomes in patients with RV failure17.
Other studies evaluating minimally-invasive dLVAD for patients in cardiogenic shock include Wert16, who compared LTHS HVAD placement to sternotomy in INTERMACS 1 patients. In this series, ≥90% of patients in both groups were on ECMO, and half had prior cardiac surgery. Seventy percent of LTHS patients remained on ECMO post-LVAD, for mean 3.5 days. In this very sick cohort, mean ICU stay was 16, and total hospital stay 31 days (p>0.05 vs. sternotomy). Significantly fewer LTHS patients required an RVAD (6% vs. 22%), and 3-month (27%) and 12-month (30%) mortality were significantly better than the 50% mortality at both timepoints in the sternotomy group.
Sagebin15 compared outcomes for patients on ECMO who received a Heartmate 3 (Abbott, Abbott Park, IL) via a complete sternal-sparing (CSS) technique to those implanted via sternotomy. Median ECMO support was 8 days, and 59% were on mechanical ventilation. Eighteen percent of CSS patients required an RVAD, vs. 31% of sternotomy patients (p=0.08). Median ICU stay (12 vs. 11 days) and total stay (22 vs. 34 days) were similar. Six-month survival was 89% CSS and 68% sternotomy.
The same group18 presented a large series comparing CSS to sternotomy for Heartmate 3. There was a high prevalence of INTERMACS 1 (41% CSS, 34% sternotomy) and ECMO (22% CSS, 13% sternotomy). They found a lower incidence of reoperation for bleeding (5% vs. 20%) and RVAD use (5% vs. 16%), and shorter median length of stay (15.5 vs. 21 days) for CSS. Six-month survival was 93% CSS and 77% sternotomy.
Our IABP patients tended toward worse renal and similarly poor RV function indices as ECMO patients. IABP has been shown not to improve survival in cardiogenic shock complicating myocardial infarction19. IABP also does not provide biventricular support. Our IABP patients had less improvement in creatinine during tMCS. Their higher incidence of early bleeding may have been partially due to ongoing congestion and renal dysfunction. However, their overall survival was excellent and none required an RVAD.
In summary, this report augments the growing body of evidence of improved outcomes with minimally-invasive dLVAD insertion. High-risk patients with cardiogenic shock were able to safely undergo LTHS dLVAD implantation after stabilization with ECMO or IABP, with acceptable short- and long-term outcomes. Perioperative outcomes and complication burden were also comparable to a less severely ill cohort who did not require tMCS.
Author Contributions: Sorensen: concept/design, data collection, data analysis/statistics, drafting manuscript; Griffith: concept/design, critical revision; Feller: concept/design, critical revision; Kaczorowski: concept/design, data analysis, critical revision, approval.