Secondary Logo

Share this article on:

Foretelling Graft Outcome by Molecular Evaluation of Renal Allograft Biopsies: The GoCAR Study

Yazdani, Saleh PhD; Naesens, Maarten MD, PhD

doi: 10.1097/TP.0000000000001512
In View: Game Changer
Social Media Collection

1 Laboratory of Nephrology, Department of Microbiology and Immunology, KU Leuven, University of Leuven, Leuven, Belgium.

2 Department of Nephrology and Renal Transplantation, University Hospitals Leuven, Leuven, Belgium.

Received 13 September 2016.

Accepted 14 September 2016.

The authors are funded by the BIOMARGIN European research network (Collaborative Project), supported by the European Commission under the Health Cooperation Work Programme of the 7th Framework Programme (Grant agreement n° 305499).

The authors declare no conflicts of interest.

Correspondence: Maarten Naesens, Department of Nephrology and Renal Transplantation, University Hospitals Leuven, Herestraat 49, 3000 Leuven, Belgium. (

More than 2000 years ago, Cicero quoted Cato the Elder: “I wonder, that a soothsayer doesn’t laugh when he sees another soothsayer”, and added: “For how many things predicted by them really come true? If any do come true, then what reason can be given why the agreement of the event with the prophecy was not due to chance.”1

Nothing much has changed since then, one could think, although with modern technologies and the use of predictive modeling statistics one may hope for an improvement. In organ transplantation, there is an undeniable need for innovative tools for precision medicine and theranostics. We lack approaches to reliably forecast outcomes after kidney transplantation, and to prevent bad outcomes with timely and targeted interventions. We wish to avoid invasive biopsies, in addition to providing clinicians with tools to accurately detect early injury while anticipating ongoing chronic or irreversible damage.2,3

Graft biopsies remain the gold diagnostic standard, despite being invasive. Histological grading is not reproducible, impacted by expertise and sampling error. Although quantitative approaches exist, thresholds for treatment decisions are blurry. Moreover, most lesions in kidney transplant biopsies are insufficiently specific and sensitive. What does interstitial fibrosis and tubular atrophy really tell us in regard to etiology or mechanisms of injury processes? Still, interstitial fibrosis and tubular atrophy remain a main contributor to late graft loss.4 Hence, attention has to be paid to expand our understanding of the immune and nonimmune mechanisms of chronic allograft injury. Fortunately, the molecular signature of progressive chronic injury linked to extensive immune alterations, can be detected long before overt histological lesions occur.5

Taking advantage of the progress in transcriptomics research, Phil O'Connell, Barbara Murphy and co-workers of the prospective Genomics of Chronic Allograft Rejection (GoCAR) consortium published an exciting translational study in The Lancet 6 on a 13-gene panel in 3 months' biopsies that allowed a prognosis of chronic injury by 1 year, which was assessed by the Chronic Allograft Damage Index [CADI]. This link has been confirmed in publicly available data from our previous publication,5 in addition to a small validation cohort of the GoCAR study.

“…transplantomics has come a long way and prognostic biomarkers in kidney transplantation are becoming reality…”

The 13-gene panel did not only provide a prognosis of progressive chronic injury, the GoCAR group also demonstrated a superior prognostic ability for graft failure compared with routine clinical and histopathological markers. After complex data analyses, 9 of 79 (11.4%) patients in the high-risk group, assessed by the 13- gene set risk score, experienced graft failure within the first 3 years. This incidence compared with 2 of 80 (2.5%) patients who lost their graft in the low-risk category. In contrast, clinical, histological, and demographic factors by 3 months did not allow a prognosis of graft outcome, making the molecular risk assessment a better tool, with an area under the receiver operating characteristic curve (AUC of the ROC curve) of 0.84 for graft failure by 3 years.

The prognostic ability for graft failure of this 13-gene set risk score was then validated against another publicly available data set,7 and confirmed the good performance of the test with an AUC under the ROC curve of 0.87 for association with graft loss by 1 year after a for-cause biopsy. This comparison illustrates that the 13-gene set risk score identifies also a subgroup of patients with a poor prognosis at the time of clinical graft dysfunction, with almost 40% graft failure by 3 years, compared with the low-risk group with less than 10% chance of graft failure by 5 years.

It remains remarkable that the 13-gene set risk score identifies both patients at increased risk of graft failure at time of clinical quiescence (3 months protocol biopsies with stable graft function), and at time of graft dysfunction (for-cause biopsies). The GoCAR study provides therefore not only a first tool for prognosis of future chronic injury, but could also help identifying patients at increased risk of graft failure. At a time when we are in search for precision medicine in the field of organ transplantation, the tool kit presented by the authors in The Lancet publication may pave the way towards novel targeted therapeutic interventions to avoid progressive injury and chronic graft loss.

Nevertheless, it has to be emphasized that the 13-gene set is a prognostic and not a predictive biomarker (Figure 1).8,9 The gene set can be used to identify a high-risk patient subgroup that is in need for specific treatments to prevent chronic injury. The 13-gene panel, however, does not predict which treatment should be used in which patient. The precise biological relevance of the genes included in the 13-gene panel remains largely unclear and there is no relation between this marker and any therapeutic intervention. O'Connell and Murphy provide some biological interpretation of the 13 targets that were derived from the association with progressive chronic allograft nephropathy, but the 13 genes resulted from a complex bioinformatics approach and have not been prioritized on mechanistic aspects of progressive fibrosis. Based on this selection process, it seems unlikely that the 13 genes will serve as relevant targets for intervention.



Knowing which patient will benefit most from which intervention needs predictive biomarkers that are often closely related to the mechanism involved.

Thus, the GoCAR gene set could become valuable in the identification of whether a patient actually needs additional therapeutic intervention to prevent fibrosis (“who to treat”); however, the set is unlikely to become helpful in decisions on which therapy to use (“how to treat”).

This should not discourage from further developing this test. It appears important for the medical industry and health authorities to note that the development and approval of a novel prognostic biomarker should not be delayed based on the presumed necessity to be predictive. A validated prognostic biomarker has great potential as part of a therapeutic algorithm, because prognostic biomarkers identify patients at the highest risk, with the greatest potential benefit of additional therapeutic interventions, whatever these may be.

The authors emphasize that extensive further validation of the 13-gene set remains necessary. Although the validation in independent cohorts with different time points of assessment suggests that the biomarker is robust, it remains unclear how reproducible the test is, what its kinetics are in repeated biopsies, how it works in patients with established fibrosis, and whether the test results are influenced by the immunosuppressive regimen or ongoing active disease processes. Furthermore, validation of the prognostic value for progressive chronic injury in much larger cohorts seems necessary, because both the test and the validation cohorts of the GoCAR study have been relatively small. The patient cohorts of our own previous study5 have also not been accurate to estimate the true performance of the test, because this study had a highly selected case-control study design, with inherent overestimation of the actual diagnostic performance. The previous study by Einecke and co-workers7 evaluated graft failure and not chronic histology, thus not allowing to evaluate whether progressive chronic injury had been an independent predictor of graft failure in this study.

Finally, the need for an invasive biopsy seems a drawback of the 13-gene set. Obviously, a noninvasive test (eg, blood-based liquid biopsy) predicting early or progressive fibrosis would be preferable. However, until now, attempts at finding noninvasive markers of intrarenal fibrosis have been very disappointing, and the 13-gene test seems the promising approach.

In aggregate, The Lancet publication by O'Connell and Murphy will hopefully fuel further validations of this gene set. If thoroughly validated and further developed as a prognostic biomarker, this test could be used to identify those patients in need for novel therapies aimed at halting or reversing progressive chronic injury.

Clearly, there is more than the pessimistic view communicated by Cicero. Indeed, transplantomics has come a long way, and prognostic biomarkers in kidney transplantation are becoming reality, despite the many difficulties and hurdles that lie ahead. Or said with the words of Madame Leota, the soothsayer from The Haunted Mansion: “You try. You fail. You try. You fail. But the only true failure is when you stop trying.”10

Back to Top | Article Outline


1. Cicero M. De Senectute De Amicitia, De Divinatione/With An English Translation By William Armistead Falconer [e-book]. Cambridge, Mass: Harvard University Press; London W. Heinemann; 1923:1923 Available from: Harvard Library Bibliographic Dataset, Ipswich, MA. Accessed September 29, 2016.
2. Naesens M, Sarwal MM. Molecular diagnostics in transplantation. Nat Rev Nephrol. 2010;6:614–628.
3. Anglicheau D, Naesens M, Essig M, et al. Establishing biomarkers in transplant medicine: a critical review of current approaches. Transplantation. 2016;100:2024–2038.
4. Naesens M, Kuypers DR, De VK, et al. The histology of kidney transplant failure: a long-term follow-up study. Transplantation. 2014;98:427–435.
5. Naesens M, Khatri P, Li L, et al. Progressive histological damage in renal allografts is associated with expression of innate and adaptive immunity genes. Kidney Int. 2011;80:1364–1376.
6. O'Connell PJ, Zhang W, Menon MC, et al. Biopsy transcriptome expression profiling to identify kidney transplants at risk of chronic injury: a multicentre, prospective study. Lancet. 2016;388:983–993.
7. Einecke G, Reeve J, Sis B, et al. A molecular classifier for predicting future graft loss in late kidney transplant biopsies. J Clin Invest. 2010;120:1862–1872.
8. Majewski IJ, Bernards R. Taming the dragon: genomic biomarkers to individualize the treatment of cancer. Nat Med. 2011;17:304–312.
9. Lo DJ, Kaplan B, Kirk AD. Biomarkers for kidney transplant rejection. Nat Rev Nephrol. 2014;10:215–225.
10. Buena Vista Pictures. The Hauted Mansion [film]. Burbank, CA: Walt Disney Pictures; 2003.
Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.