Bradycardia in Recent Heart Transplant: Will the Microscope Illuminate the True Answer?CME

Transplant recipients are at risk of developing rejection that may cause significant morbidity and mortality following transplantation The clinical presentation of rejection may be atypical, leading to difficulties in diagnosis and management especially in cases with a nondiagnostic biopsy specimen. The emergence of artificial intelligence may aid in clinical decision making when traditional techniques are inconclusive.


INTRODUCTION
With the increased volume of heart transplants, it is crucial for providers to be aware of potential complications patients face and their inherent management challenges. One such complication is acute rejection, although improved immunosuppressive regimens in recent years have helped lower the incidence of rejection and hospitalization for heart transplant patients. 1

A) Cardiac magnetic resonance imaging B) 2-dimensional echocardiogram C) Endomyocardial biopsy D) Molecular Microscope Diagnostic System
ANSWER 1

(C) Endomyocardial biopsy
According to the most recent International Society for Heart and Lung Transplantation (ISHLT) guidelines, the current standard of care for all adult orthotopic heart transplant (OHT) recipients is to undergo periodic endomyocardial biopsy (EMB) 6 to 12 months postoperatively for surveillance of heart transplant rejection or at any point if there is a clinical suspicion for rejection. 2 After this time, periodic EMB surveillance every 4 to 6 months is recommended for heart transplant recipients at higher risk for late acute rejection. Beyond a period of 5 years post heart transplantation, the routine use of EMB is optional depending on clinical judgment and the risk of late allograft rejection. 2

e14
Although EMB is the ISHLT's method of choice for primary assessment and surveillance when transplant rejection occurs, 2 its utility in detecting rejection is restricted when sampling and reporting issues arise and can thus lower the sensitivity for accurate diagnosis. With nondiagnostic EMB results, physicians have newer options in place such as donor-derived cell-free deoxyribonucleic acid (DD-cfDNA) and the Molecular Microscope Diagnostic System (MMDx TM , One Lambda, Inc.). However, without proper guideline support for these tools, care teams must collaborate creatively to manage patients with suspicion for rejection.

CASE HISTORY
A 56-year-old African American male with left ventricular noncompaction cardiomyopathy and ventricular tachycardia on amiodarone underwent orthotopic heart transplantation (OTH). The postoperative hospital course was complicated by intermittent asymptomatic junctional rhythm in the 50s. The patient was discharged on terbutaline after intermittent sinus rhythm improved to the 80s. His immunosuppression regimen consisted of tacrolimus, azathioprine, and prednisone. Six weeks after transplant, the patient had normal routine surveillance studies (Table 1) except for a new diagnosis of donor-derived coronary disease based on intracoronary intravascular ultrasound. Outpatient Holter monitoring showed evidence of heart rate between 35 bpm while sleeping to 80s when awake, with no evidence of pauses or patient-triggered alarms.
Four months after transplant, the patient was admitted for loss of consciousness. Extensive workup for seizures and COVID-19 testing was negative. Outpatient tacrolimus levels varied between 8.2 and 15.2 ng/mL prior to admission. The patient was noted to have a junctional rhythm in the 30s that required dopamine. Repeat diagnostic workup for rejection is shown in Table 1

CASE CONTINUED
At this time, we performed MMDx imaging (Figures 1 A, B) and EMB (Figures 2 A, B).  Endomyocardial biopsy.

QUESTION 3
Based on the results of the MMDx (Figure 1 A, B), what is the diagnosis for our patient? The result from our patient's biopsy specimen (yellow triangle) is represented among 889 reference biopsies (remaining circles) that are distributed by their molecular-rejectionrelated measurements in a three-dimensional data cloud. 3 Figure 1 A shows the main variation (principal component [PC], x-axis, normal vs. abnormal) compared to the second aspect of variation (PC2, y-axis, that separates ABMR from TCMR). 3 Figure 1 B rotates the data cloud to show the third axis of variation (PC3, separating acute injury from rejection). 3 Thus, our patient has TCMR (Figure 1 A) with injury secondary to TCMR (Figure 1 B) without evidence of ABMR.

QUESTION 4
The molecular microscope diagnostic system incorporates machine learning algorithms to assign probability of disease state by isolating and analyzing which of the following? MMDx isolates mRNA, measures gene expression with 99% precision using gene chips, and uses machine-learningderived algorithms to express diagnostic probabilities of each new biopsy compared to a reference set. 4,6 In addition, machine learning overcomes errors in sample labeling, such as those seen with biopsy diagnoses. 5,6 QUESTION 5 Figure 2,  After a multidisciplinary team approach to manage the patient, including high-dose steroids, and switching his antimetabolite from azathioprine to mycophenolate mofetil, the patient's symptoms resolved.

Based on the EMB shown in
Ultimately, the patient was discharged in sinus rhythm without terbutaline. Outpatient heart monitoring revealed no further bradycardia. In subsequent visits, he had normal heart rates and improved surveillance studies without further changes in his medication regimen.
The utility of artificial intelligence when used by cardiac care teams should be evaluated further. As mentioned earlier, the ISHLT guidelines recommend a few alternatives to EMB for rejection monitoring, such as gene expression profiling (eg, Allomap, CareDx, Inc.) and ventricular evoked potentials monitoring, while not recommending the use of various laboratory markers and noninvasive imaging. 2 However, little guidance is suggested for suspected rejection when there are nondiagnostic or inconclusive biopsy results. This is especially relevant because human error plays a significant role in biopsy results. One study even suggests that pathologists agreed with each other on cell-mediated rejection approximately 50% of the time. 6 Tools such as MMDx, a system that uses machine learning to compare gene expression to a given data set, can assist in managing transplant rejection. Although mainly studied using kidney transplant rejection, MMDx has been shown to provide more diagnostic accuracy than histological biopsy results in kidney transplant patients with antibody-mediated rejection and more accurately diagnose results that were left equivocal with standard histology. 4,6 Further studies should be considered involving cellular rejection in heart transplants and the role MMDx and other machine-learning technologies can have when compared to the current gold standard. This could play a vital role in supplementing or replacing the information provided by EMB.

CONCLUSION
The use of machine-learning tools like MMDx proved to be invaluable in the case of our patient, who had no evidence rejection on EMB. With the information these resources provide, cardiac care teams can effectively impact the management trajectory of transplant patients with complications. Transplant standards should reflect the increased utility of these tools to ensure providers have more guidance when suspecting rejection when biopsy shows otherwise.