1. Gene expression testing in cardiac and lung transplantation
Banff Congress
Jay Wohlgemuth MD
July 16, 2005

2. Scientific assumptions
Gene expression profiling of the immune system may anticipate tissue injury
Multiple genes from multiple pathways are required to overcome complexity and variability
Complex multi-pathway signals can be reduced to simple, clinically actionable test result(s)
Use of genomic information may enable proactive therapy, reduction in un-necessary immunosuppression and monitoring procedures
Confidential, 11/3/2018

3. The transplant patient management challenge
Confidential, 11/3/2018

4. Clinical need for molecular testing in cardiac transplantation
Monitoring for rejection
Rejection rates are very low (2-3% for Grade 3A)
Biopsy has limitations for patients and physicians
Reduction in burden of immunosuppression
Complications of IS are a major cause of M & M beyond the 1st year post-transplant
Minimization may be facilitated with molecular testing
Clarification of uncertain clinical and pathological cases
Mild rejection
Need for augmentation or change in Rx
Confidential, 11/3/2018

5. CARGO Clinical Study
Goal: discover, develop and validate gene expression testing for rejection and quiescence in cardiac transplant recipients
Multi-center observational study initiated in (centers represent 22% of US cardiac transplants)
Followed 650 patients during > 5500 post- transplant encounters
Microarrays used for gene discovery, real- time PCR for development and validation of a multi-gene, multi-pathway molecular test
Prospective, blinded validation study of 20 gene algorithm demonstrated ability to distinguish rejection from quiescence
Confidential, 11/3/2018

6. CARGO Study Overview Candidate gene selection Phase I Exploratory
Leukocyte microarray derived from 25K cDNAs and human genome information
285 CARGO samples used in microarray experiments
Database and literature mining
Identification of 252 candidate genes
Phase I
Exploratory
Phase II
Development
Algorithm development
Sensitive and reproducible real-time PCR methods
Development of a 20-gene algorithm to distinguish rejection from quiescence (AlloMap)
We have followed a rigorous development process which comprises thre distinct phases.
During the exploratory phase we researched and identified gene candidates from the genome and over the course of a three year clinical trial with eight centers across the USA, which collectively represent more than 20 % of the heart transplants performed every year we collected over 5000 samples from more than 600 patients.
We selected relevant biomarkers from about 8,000 gene probes that we had evaluated on microarrays and created a multi-gene panel quantitative kinetic PCR test system.
As a consequence of this work we have filed for several patents for which some of the claims have already issued.
Finally we have validated our complete system through a prospective, blinded, statistically-powered validation study which is more analogous to the pharmaceutical industry than to the traditional diagnostic.
One of the challenges we encountered was the fact that the reference method, biopsy is less than perfect. Therefore we had to improve biopsy results by having all bipsy results read by three different pthologists to increase the consistency of the biopsy results.
The results and findings of our development and validation have been summarised and a manuscript submitted to a major peer-reviewed medical journal.
Validation
Prospective, blinded, statistically-powered clinical study (n = 270)
Additional samples were tested to further define performance (n > 1000)
Phase III
Validation Study
Confidential, 11/3/2018

7. Rejection Associated Gene Expression Pathways
Of 252 PCR-assayed genes, 68 genes correlated with rejection (p < 0.01) and/or have a median ratio more than ±25%.
Measuring both gene expression and shifts of cell populations
CD8 T cell and NK markers
Markers of hematopoiesis up-regulated with rejection
Activated Macrophage / PMN
Steroid responsive
Megakaryocyte
Hematopoiesis
Cytokines, IFN induced
T lymphocyte
68 genes were shown to correlate with high grade rejection. The major categories include:
T cell regulation and migration including PDCD1 and ITGA4 that are known to upregulate with rejection.
Quote from manuscript: “Genes involved in T cell regulation and migration (PDCD1 and ITGA4), were also increased in rejection. In the case of PDCD1 a role for this molecule in late events post-transplant in an animal model has been described36. “
Genes of the Erythroid lineage including WDR40A, MIR, and ALAS2 suggesting that rejection is associated with active hematopoesis
A steroid responsive pathway was also identified. This pathway overlaps with one of the informational genes used in the rejection algorithm or AlloMap score.
B lymphocyte
Confidential, 11/3/2018

8. Monitoring Multiple Pathways Associated with Rejection
Platelet Activation
Inflammation
PF4, G6B
IL-6
T cell Priming
PDCD-1, ITGA4
Lymph node
Mobilization of Hematopoietic precursors
WDR40A, MIR
Lymph and Lymphocytes
Naïve T cell
Dendritic cell
Monocyte
Primed T cell
Rejection Rx
IL1R2, ITGAM, FLT3
Confidential, 11/3/2018

9. AlloMap Diagnostic Algorithm
4 x Gene 1
+ 1 x Metagene 1
– 2 x Metagene 2
– 3 x Metagene 3
+ 5 x Gene 2
+ 6 x Gene 3
+ 7 x Gene 4
0 40 Q R
68 genes were shown to correlate with high grade rejection. The major categories include:
T cell regulation and migration including PDCD1 and ITGA4 that are known to upregulate with rejection.
Quote from manuscript: “Genes involved in T cell regulation and migration (PDCD1 and ITGA4), were also increased in rejection. In the case of PDCD1 a role for this molecule in late events post-transplant in an animal model has been described36. “
Genes of the Erythroid lineage including WDR40A, MIR, and ALAS2 suggesting that rejection is associated with active hematopoesis
A steroid responsive pathway was also identified. This pathway overlaps with one of the informational genes used in the rejection algorithm or AlloMap score.
Confidential, 11/3/2018

10. CARGO Study Results Summary
AlloMap distinguishes grade ≥3A rejection from grade 0 at all times post-transplant (p <0.0001)
Grade 1B samples have high algorithm scores on average, grades 0, 1A and 2 are indistinguishable
Patients with low scores have very low risk of moderate-severe acute rejection
Confidential, 11/3/2018

11. CARGO Study Results Summary
AlloMap correlates more closely to centralized than local pathology
Algorithm predicts future rejection and graft dysfunction in grade 0 cases
Pediatric samples look qualitatively similar
CMV signature identified which does not confound the AlloMap test result
Confidential, 11/3/2018

12. AlloMap Score Increases with Decreased Steroid Dose
AlloMap score and steroid dose vs. days post transplant
AlloMap
AlloMap score, quiescent samples
Prednisone dose [mg/day]
The Allomap score increases over time and paralells the prospective reduction in corticosteroids that occurs. Note that these data are from quiescent samples only thus showing the presumed impact of “steroid responsive” genes on the AlloMap score
Prednisone
Days post transplant
Confidential, 11/3/2018

13. Steroid Responsive Genes and Pathways
5 of 11 informative algorithm genes significantly correlate with steroids
Steroid and rejection gene expression responses are opposite
Predominance of monocyte and PMN expressed genes
ITGAM and IL1R2 are most responsive to steroid dose
Steroid Gene Rejection Response Description Cell Type
ITGAM Integrin, alpha M Monos, PMN, NK
IL1R2 Interleukin 1 Monos, PMN receptor
G6b lg superfamily Hemotopoietic cell lines
FLT fms-related lymphoid/myeloid tyr kinase progenitors
ITGA integrin, alpha 4 Monos, PMN, lymphocytes
Proposed text for this:
5 of 11 informative genes are also steroid responsive. ITGAM and IL 1R2 are quantitatively the most responsive to steroid dose. Note that the rejection response is opposite to the steroid response.
Questions for clarification:
I need to understand this process better: Red means increased expression of a gene so for REJECTION ITGAM, IL 1R2, G6b, FLT3 are all with decreased expression and ITGA4 gene is upregulated. For steroid response 4 genes are upregulated and only ITGA4 is down regulated. Please confirm.
Quote from manuscript: The predominant genes showing increased expression with rejection were activation markers of T-cell/NK cells and CD8+ T-cells, (e.g., perforin and granulysin) and erythropoiesis markers (e.g., ALAS2, WDR40A, MIR). These results are consistent with up-regulation of CD8+ markers during rejection observed previously Additional genes showed reduced expression with rejection which may represent decreased gene expression, decreased proportions of cell types or migration of cells from the blood to the graft with rejection.
Confidential, 11/3/2018

14. Individual Patients Have Varied Responses to Steroids: Steroid Resistant Rejection and Steroid Sensitivity
Gene expression based estimate of steroid dose
Quiescent gold
Rejection blue
Multiple linear regression model was utilized to fit gene expression values of the 5 algorithm steroid responsive genes with reported steroid dose .
A functional gene response to steroid treatment is measurable
(logistic models also tested, but largely linear Ct-dose relationship found); Regression coefficients derived from training sample set and applied to independent Validation samples;
The rejectors are shown in blue. The episodes shown in the blue circle 7 epidsodes of rejection where the actual dose is much higher than the dose predicted by the 5 steroid responsive genes and may represent “steroid resistent rejection”
Perhaps more importantly clinically, all of the quiescent episodes plotted to the left of the line of identity represent patients where the functional steroid gene response is greater than predicted by the actual dose; these patients may be candidates for a reduction in corticosteroids.
More off-diagonal HR samples than Q samples, indication of inadequate response to steroids in HR patients?
Confidential, 11/3/2018

15. Identification of Quiescence
Samples below threshold are unlikely to have 3A or higher biopsy NPV > 99%
Samples above threshold are enriched for concurrent biopsy ≥3A
12X increased risk for 3A rejection vs. low scores
Still, low PPV relative to biopsy – Why?
Below threshold
(high NPV)
Above threshold
(moderate PPV)
68 genes were shown to correlate with high grade rejection. The major categories include:
T cell regulation and migration including PDCD1 and ITGA4 that are known to upregulate with rejection.
Quote from manuscript: “Genes involved in T cell regulation and migration (PDCD1 and ITGA4), were also increased in rejection. In the case of PDCD1 a role for this molecule in late events post-transplant in an animal model has been described36. “
Genes of the Erythroid lineage including WDR40A, MIR, and ALAS2 suggesting that rejection is associated with active hematopoesis
A steroid responsive pathway was also identified. This pathway overlaps with one of the informational genes used in the rejection algorithm or AlloMap score.
Confidential, 11/3/2018

16. Molecular Testing: Correlation to Biopsy and Graft Dysfunction
Humoral Rejection
Molecular
Rejection
Cellular
Rejection
Graft
Dysfunction
AlloMap Test
(early)
Biopsy
(late)
Graft Failure
(too late)
Confidential, 11/3/2018

17. Risk of graft dysfunction with high score, negative biopsy
Risk of graft dysfunction (PCW >20) within 45 days
RR = 6.8
p = 0.03
Low
High
Confidential, 11/3/2018

18. Cardiac biopsy interpretation variability contributes to discordance between molecular and pathological results
Pathology panel (Billingham, Marboe and Berry) re-read all biopsy slides for the study (n = 827) Marboe et al., JHLT 2005
The maximum concordance between two central pathologists for grade ≥3A rejection was 77%
The average concordance between the local and central pathologists grade ≥3A rejection was 40%
Local pathologists call grade ≥3A rejection 50% more frequently than central pathologists
Quilty lesions cause significant uncertainty and overcalls for rejection
Confidential, 11/3/2018

19. Quilty lesions cause over diagnosis of ISHLT Grade 2 and 3A rejection
Serial sections of 18 cases performed
All cases involved local Grade 2 or 3A
All cases had been identified as likely Grade 0-1 and Quilty B by centralized panel
17 of 18 confirmed to be Grade 0-1
12/12 Local Grade 2s
5/6 Local Grade 3As
Confidential, 11/3/2018

20. Local Pathologist: Rejection
Confidential, 11/3/2018

21. Consensus: Quilty B
Confidential, 11/3/2018

22. Discordance between molecular and pathological results
Positive biopsy, low molecular score
>50% of Grade 3As after year 1 may resolve without therapy
Quilty lesions or other causes of over diagnosis by biopsy
Molecular test and biopsy measure different processes which may be discordant
Negative biopsy, high molecular score
Early, focal rejection, negative on biopsy
Sensitivity of test for humoral rejection?
Patients may have peripheral alloimmune activity, no cellular rejection
Risk of vasculopathy?
Confidential, 11/3/2018

23. AlloMap Scores by ISHLT Grade
472 samples ≥ 6 months post-transplant
Grade 1B sample scores were significantly higher than Grades 0 (p=0.02) and 1A (p=0.002)
Grade 1B scores were not significantly different than grade ≥3
Grades 0, 1A and 2 scores were not significantly different
Mild rejection (0-2) with high scores have significant increased risk of progression to grade 3A on next biopsy (p = )
AlloMap Score
Box plot showing 25% - 75% quartiles and line connecting medians
Local ISHLT grade
Confidential, 11/3/2018

24. Molecular response to rejection therapy
Confidential, 11/3/2018

25. Clinical uses of molecular testing in cardiac transplantation
Rejection surveillance
Stable outpatients with low scores are low risk: biopsy reduction
Access issues
Immunosuppression titration
Guide weaning of immunosuppression
Follow after rejection therapy
Risk stratification
Mild rejection or possible Quilty on biopsy
Uncertain clinical picture
Confidential, 11/3/2018

26. CARGO – Ongoing Studies, CARGO II Study
Prediction of clinical outcomes
Immunosuppression / Rejection Rx monitoring
Humoral rejection
Vasculopathy
Pediatrics
Infection
Confidential, 11/3/2018

27. Endpoints: Acute rejection, BOS, infection
LARGO Study
Ongoing 14 center international study of molecular testing in Lung transplantation
Endpoints: Acute rejection, BOS, infection
Infectious complications may be addressed with development of new molecular information
Common relevant pathways for multiple solid organ transplant settings will be sought
Confidential, 11/3/2018

28. Thank You
Confidential, 11/3/2018

29. CARGO Centers and Collaborators
Transplant cardiology programs:
Cleveland Clinic Randall Starling, MD, MPH
Columbia University Mario Deng, MD, Helen Baron, MD
Columbia University (peds) Seema Mital MD, Linda Addonizio, MD
Ochsner Clinic Mandeep Mehra, MD
Stanford University, PAVAMC Sharon Hunt, MD, Fran Johnson MD
Stanford University (peds) Daniel Bernstein, MD
Temple University Howard Eisen, MD
UCLA Medical Center Jon Kobashigawa, MD
University of Florida James Hill,MD, Dan Pauly, MD, PhD
University of Pittsburgh Srinivas Murali, MD, Adrianna Zeevi, PhD
University of Pittsburgh (peds) Steven Webber, MBChB
Centralized reading of biopsy pathology:
Gerald Berry, MD (Stanford)
Margaret Billingham, MD (Stanford)
Charles Marboe, MD (Columbia)
Confidential, 11/3/2018