According to Banff classification, borderline rejection in kidney allografts exist as an entity between no acute rejection and acute cellular or T cell-mediated rejection type IA. Its criteria were arbitrarily set with mild tubulitis (t score equal to or less than 1) and/or mild interstitial mononuclear cell infiltrate (i score equal to or less than 1). Borderline rejection is problematic because it does not tell whether rejection is occurring in the grafts. Acute tubular necrosis, a common cause of delayed graft function, can also have similar histologic features.
de Freitas et al used microarray technique to differentiate 40 cases of borderline rejection into T cell-mediated rejection (TCMR)-like and nonrejection-like and found that the former were 13 (33%) and the latter were 27 (67%). Decision tree analysis indicated the i-total >27% and tubulitis extent >3% as the cut off point for most of the cases with molecular signal of TCMR-like.
The i-total represents the total interstitial mononuclear cell infiltrate in both scar and non-scar areas while the standard i represents only infiltrate in non-scar. Further evaluation found that not considering infiltrate in scar areas was the major cause of those with molecular signal of TMCR-like but were histologically borderline rejection.
The t score was not different between rejection and non-rejection-like molecularly, but extent of tubulitis or t-extent was.
Using both molecular and histological techniques in evaluating renal allografts biopsies are not practical in most institutes. However, using the i-total and t-extent rather than original t and i score have a promising prospect in eliminating borderline rejection.