This study demonstrates the reproducibility of MMDx and the lack of significant disagreement between MMDx sign outs.
The study describes in detail where MMDx and histopathology disagree. There is relatively good agreement between histological ABMR and MMDx, but MMDx diagnoses more ABMR than histology. Additionally, there is reasonable agreement in “no rejection” but there is very significant disagreement in TCMR and “borderline for TCMR.”
Diagnostic platforms require precision (reproducibility), accuracy (represent the true disease state), and standardization. Histology has well-described limitations.
The Molecular Microscope Diagnostic System (MMDx) consists of an automated report combining rejection and injury-related scores with a graphic representation of the molecular position of the current biopsy, which is compared to a reference set, and a categorical sign-out by an expert or an algorithm (AutoMMDx).
This study compared the results of 1679 indication biopsies from the 1120 INTERCOMEX trial subjects and compared various discrepancies: within MMDx; between MMDx sign-outs; and between MMD and histology.
The results showed that MMDx outputs (rejection classifiers, etc.) are nearly identical between technical replicates, i.e., a biopsy core divided in two and processed separately gives a 99% correlation. Similarly, MMDx sign-outs between two experts and the automated algorithm were highly correlated. One comparison revealed 6% discrepancies, of which only 1.4% occurred in clear categories. Conversely, MMDx and histology sign-outs disagreed 37% of the time, broken down into 22% in ABMR, 61% in TCMR, and 24% in no rejection.
The histology diagnosis of “borderline for TCMR” was rarely TCMR by MMDx (ABMR 24/128, no rejection 83/128, TCMR 9/128). 49% of MMDx diagnosed ABMR was not ABMR by histology, and although MMDx diagnosed similar numbers of TCMR as by histology, there was extensive disagreement. 68/123 MMDx diagnoses of TCMR were not reported as TCMR by histology. MMDx may offer particular value where pathology is ambiguous, i.e. borderline or suspicious.
If not sending MMDx on all biopsies already, the authors recommend storing a 3mm section of biopsy core in RNA later for use in certain circumstances, such as where MMDx may add clarification:
It would also be sensible to utilize MMDx where patients are re-biopsied because the original biopsy was ambiguous or where the clinical course has not unfolded as expected.