Abstract

The median survival of FL is 10 years, but has a heterogeneous clinical course and some patients die rapidly from disease, while others survive for decades. Global gene array of FL diagnostic LNs has distinguished immune cell-associated signatures associated with good and poor prognosis. The goal of this study was to attempt to validate the gene array data at the protein level as well as assess the ability to discriminate prognostic groups based upon immunohistochemistry of diagnostic lymph nodes. TMAs were constructed of 1mm cores from initial diagnostic LNs from two groups of FL patients. The first group selected were 35 patients who survived <5 yrs from diagnosis (median survival 2 yrs; median age 61 yrs; median stage 4; median FLIPI 3). The second group selected were 25 patients who survived >15 yrs from diagnosis (median survival 20 yrs; median age 46 yrs; median stage 3; median FLIPI 2). Patients in both groups received a variety of standard treatments. Immunohistochemisrty was performed on the TMAs using a panel of antibodies detecting antigens associated with T-cells, including (CD4, CD8α, CD7, CD25, TIA-1, CD45RO, FOXP3) and macrophages including CD68, CD163. The immune infiltrates were scored for immunophenotype, frequency and peri- and inter-follicular distribution. Of the panels completed to date, incidence of FOXP3, perifollicular CD4 and perifollicular CD7 showed greatest discrimination between the two patient groups.

CD4, CD7 and FOXP3 in prognostic patient groups

Antigen expressionGood Prognosis PtsPoor Prognosis Pts
Perifollicular CD4 >5 cells/hpf 18/23 cases (78%) >5 cells/hpf 18/35 cases (51%) 
 <5 cells/hpf 5/23 cases (22%) <5 cells/hpf 17/35 cases (49%) 
Perifollicular CD7 >5 cells/hpf 22/23 cases (96%) >5 cells/hpf 26/34 cases (76%) 
 <5 cells/hpf 1/23 cases (4%) <5 cells/hpf 8/34 cases (24%) 
FoxP3 >5 cells/hpf 21/24 cases (87%) >5 cells/hpf 21/32 cases (66%) 
 <5 cells/hpf 3/24 cases (13%) <5 cells/hpf 11/32 cases (34%) 
Antigen expressionGood Prognosis PtsPoor Prognosis Pts
Perifollicular CD4 >5 cells/hpf 18/23 cases (78%) >5 cells/hpf 18/35 cases (51%) 
 <5 cells/hpf 5/23 cases (22%) <5 cells/hpf 17/35 cases (49%) 
Perifollicular CD7 >5 cells/hpf 22/23 cases (96%) >5 cells/hpf 26/34 cases (76%) 
 <5 cells/hpf 1/23 cases (4%) <5 cells/hpf 8/34 cases (24%) 
FoxP3 >5 cells/hpf 21/24 cases (87%) >5 cells/hpf 21/32 cases (66%) 
 <5 cells/hpf 3/24 cases (13%) <5 cells/hpf 11/32 cases (34%) 

To further validate this training set, we interrogated this panel on a further test set of 33 diagnostic and relapse samples from FL patients to assess predictors of outcome on a TMA. Increased detection of CD4 or CD7 cells in a perifollicular pattern correctly predicted better outcome in 67% and 58% cases respectively. Here we demonstrate that not only the cell number and cell type, but also the distribution of the immune infiltrate has prognostic significance. Moreover, unlike gene expression profiling, immunohistochemistry can readily be applied in a diagnostic setting in routine histopathology. It is highly likely that the ability to discriminate between these two prognostic groups at diagnosis will influence patient treatment algorithms.

Author notes

Corresponding author