Blog entry by Arif Khwaja

Anyone in the world
There is an interesting paper published in this weeks NEJM by Suthanthiran and colleagues evaluating the predictive value of urinary biomarkers in acute cellular rejection. They collected 4300 urine samples from 485 kidney grafts from day 3 post-transplantation to 12 months. They measured absolute levels of  urinary mRNA of CD3e, perforin, granzymeB, proteinase inhibitor 9, CD103, Inteferon-inducible protein -10 (IP-10), CXCR3 and TGF-beta1 and 18S ribosomal RNA (rRNA). They came up with a three gene signature of CD3e, IP-10 and 18S rRNA to discriminate acute cellular rejection from those not displaying rejection.
They demonstrated an area under the curve (AUC) of 0.85, p<0.001 by ROC curve analysis for this three gene signature discriminating aucte cellular rejection from non-acute cellular rejection. In the external data validation set the AUC was lower at 0.74 though this apparently wasnt significantly different from the AUC in the primary data set. Interestingly in sequential urine samples there was a significant incraese in the 3 gene signature in the 30 days prior to the development of acute rejection, raising the possibility that this signature could be used to identify incipient patients who are about to develop rejection. There is a long history of groups looking for non-invasive markers of rejection (either for mRNA or a proteomic approach) and this is certainly an interesting addition to that literature.
The statistical methodology in this paper is complex and is beyond my area of expertise but my immediate thoughts are:
 
i) is an AUC of 0.74 to  0.85 really good enough for a diagnostic test? Probably not!
ii) In an era of daycase US-guided renal biopsies are we overplaying the risk of transplant biopsies? The risks of biopsies are small (but not zero) and its not clear to me that there is a clinical need to replace a gold-standard test with a non-invasive test. A biopsy will of course provide other invaluable therapeutic information such as evidency of viral infections and drug toxicity which maybe difficult for a non-invasive test to do.
iii) the increase in the trajectory of the gene signature prior to the diagnosis of rejection is interesting as it raises the tantalising possiblity that such an approach may help diagnose rejection earlier. However no comparative data is presented to show what happened to either GFR or creatinine in the 30 days prior to the diagnosis of rejection. i.e. the paper didnt really answer the question whether this test was any better than measuring the serum creatinine
iv) clearly prospective studies in other data sets comparing to conventional measures of graft dysfunction will be helpful- till then I think this kind of approach to diagnosing rejection has a long way to go
 
[ Modified: Thursday, 1 January 1970, 1:00 AM ]