Biomarkers and Kidney Transplant: Time for a New Paradigm?

    loading  Checking for direct PDF access through Ovid

Excerpt

The search for an ideal biomarker in kidney transplant has been a daunting journey.1 Initially, attention was centered around biomarkers that could correlate accurately with early acute T cell–mediated rejection (TCMR).2 Though these studies showed early promise, adoption as a clinical diagnostic tool has not been universally accepted. As the incidence of acute TCMR has gone down dramatically over the last several years, the development of a reliable biomarker for acute TCMR has been further hampered by low positive predictive values (PPVs). Also, the challenges of thorough validation3 continue to present challenges to the discovery of solid associative data, let alone true predictive and, perhaps more importantly, true surrogate biomarkers.
In this issue of Transplantation, Faddoul et al4 have used data from the Clinical Trials on Transplant (CTOT-1) trial along with GoCar (CTOT-17) to address the issue of late graft loss. This is a much needed and refreshingly rigorous attempt to tackle perhaps a far more important issue than early TCMR, namely, late graft loss; important particularly in light of questions regarding the direct causal link of acute TCMR and graft loss.5 For this, the authors should be given our gratitude. However, the challenges that surround valid biomarker detection of TCMR are even more apparently difficult when addressing late graft loss.
In the study by Formica et al, molecular diagnostics, namely, CXCL9, Elispot, and so on, had no association with either 5-year graft outcomes or changes in GFR. The major finding of this study was a modest association of loss of renal function between 6 to 24 months and 5 year graft loss. Caution must be used in interpreting these results, for the following reasons.
Perhaps, the greatest lesson from this exhaustive study by Formica et al is rethinking the concept of graft loss. Graft loss is often a continuous decrement in renal function, not a binary event. The pitfalls in treating something inherently continuous, such as graft loss, as a binary event can possibly lead to an absurd situation. If a graft is considered “lost” at an arbitrary cut off, that is, eGFR of less than 20 ml/min, then enormous differing degrees of injury may be lumped together; conversely, small differences in degree of injury may be dichotomized and misleadingly be addressed statistically as identical. For example, a patient may actually lose graft function by going from 50 to 20 mL/min and yet by definition will escape the cutoff. This patient thus would be treated similarly as those without any significant change in renal function at all. Both would fall under “graft survival” despite vastly different degrees of functional loss and injury. This binary cutoff does not reflect degree or mechanism of injury. Graft loss encompasses a heterogeneity and spectrum of injury; current biomarkers are not designed to detect an arbitrary cutoff.
One wonders, then, if rather than an arbitrary endpoint to determine graft loss, should we not be looking more at the degree of injury, or clinical loss of function? This would certainly make more physiologic sense and would avoid the above mentioned statistical anomalies.
We are grateful to the members of CTOT-17 for their valuable contribution and for highlighting the need for us, as a community, to be open to rethinking old paradigms.
    loading  Loading Related Articles