Interferon signaling patterns in peripheral blood lymphocytes may predict clinical outcome after high-dose interferon therapy in melanoma patients

Archived peripheral blood mononuclear cells (PBMCs) from 14 Stage IIIB-C melanoma patients (a total of 28 PBMC samples, 14 acquired pre- and 14 acquired post-HDI treatment) were analyzed for STAT1-Y701 phosphorylation (pSTAT1) levels by phosflow cytometry. Patient demographics and clinical details are shown in Table 1.

These patients participated in a clinical trial for neoadjuvant therapy with HDI and their treatment has been thoroughly described [12]. Briefly, the HDI induction phase consisted of IFN-α2b 20 million units (MU)/m² per day intravenously, 5 days per week (Monday to Friday) for 4 weeks. Following surgical resection, patients were placed on a HDI maintenance regimen consisting of IFN-α2b 10 MU/m² per day subcutaneously, three times per week (Monday, Wednesday, Friday) for 48 weeks. For each patient, blood samples were taken before (Day 0-pre) and after the 4-week induction phase of HDI (Day 29-post). Due to adverse events, 5 patients underwent 1/3 dose reductions and two of these patients required dose reductions twice (Table 1). Phosphorylated STAT1 levels were measured in lymphocytes, T cell subsets (both CD4 and CD8) and B cells with or without stimulation of IFN-α for each of these two time points. All patients signed informed consent and the study was approved by the University of Pittsburgh (Pittsburgh, PA) Institutional Review Board.

Interim responses were determined using WHO criteria [26] and patients were classified as clinical responders or non-responders based on a measure of tumor reduction both clinically and histologically over the 4-week induction phase of HDI therapy[12]. Among these patients, 9 showed a complete response (n = 1) or partial response (n = 8) and both were grouped into a single responder group (R). Five patients did not exhibit a clinical response and were grouped as non-responders (NR).

For long-term clinical outcome, patients were further classified as exhibiting no evidence of disease (NED) or metastatic disease (MET) based on their status at the time of follow-up (range 9-86 months) after completing a maintenance regimen of 48 weeks. Six patients were classified as NED and exhibited no evidence of metastases at a minimum follow-up of 4.5 years. In contrast, all of the 8 MET patients developed metastatic disease within 3 years and three of these were disease free for up to 1 year (Table 1).

IFN-α stimulation and detection of pSTAT1-Y701 in cancer patients have been previously described [16] with modifications. Briefly, cryopreserved PBMCs were thawed and rested overnight in IMDM 10% FBS at 37°C 7% CO₂. Cells were ficolled, resuspended to 2 × 10⁶ cells per 50 μl in IMDM 5% human AB serum (HS) and stained with mouse anti-human CD3 FITC, CD8 PE-Cy7, CD4 PE-AF700 and CD19 PE-TR (Caltag-Invitrogen) for 30 minutes. IMDM 5%HS was added to each tube and 1 × 10⁶ cells were aliquoted per test. PBMCs remained unstimulated or were stimulated with IFN-α (NIAID Reference Reagent Repository) to a final concentration of 1000 IU/ml and incubated at 37°C 7%CO₂ for 15 minutes. Cells were fixed by formalin and incubated at 37°C 7% CO₂ for 10 minutes. PBMCs were washed twice with 1× PBS, resuspended in 1 ml of 1× Custom Perm Buffer (#643435, BD Biosciences), and incubated for 30 minutes at room temperature. Cells were washed in wash buffer (1× PBS, 2% FBS, 0.09% sodium azide), resuspended to exactly 50 μl and incubated with mouse anti-human STAT1-pY701 Alexa Fluor^® 647(BD Biosciences) for 1 hour at room temperature. Cells were washed in wash buffer and put on ice until analyzed by flow cytometry on a LSRII flow cytometer (BD Biosciences). Paired pre-and corresponding post- PBMC samples were assayed on the same day.

Flow cytometry FCS files were analyzed using FlowJo 8.5.3 (Treestar, http://​www.​treestar.​com). Gating strategy for the selection of lymphocytes, T cells and B cells are shown in Additional File 1 Figure S1. The mean fluorescent intensity (MFI) of STAT1-pY701 Alexa Fluor^® 647 was calculated for all stimulated and unstimulated samples and fold changes were determined by dividing the MFI of stimulated samples by the MFI of the corresponding unstimulated samples. Basal levels of STAT1-Y701 were determined by the MFI in unstimulated cells. Kaplan-Meier survival curves were generated and the correlation of IFN-α induced pSTAT1 with disease-free and overall survival was estimated using the log-rank test. For the purpose of these comparisons, a ratio for each patients' lymphocytes were determined by dividing the fold change in pSTAT1 post-treatment by the fold change in pSTAT1 pre-treatment. A median of these ratios was generated using all patients in the study (n = 14) and patients were segregated according to whether they fell within a range of ± 0.1 around the median. Data was analyzed using Graphpad Prism 5.00 and the R statistical package 2.7.1 http://​www.​r-project.​org. P-values, estimated differences and 95% confidence intervals were calculated with R software from the Comprehensive R Archive Network using nonparametric unpaired or paired two-sided Wilcoxon-Mann-Whitney T-tests. The False Discovery Rate (FDR) was calculated in R and used to adjust for multiple comparisons testing [27]. Adjusted P-values < 0.05 were considered significant. Coefficient of variations (CV) was calculated by dividing the standard deviation with the mean of the fold changes multiplied by 100.

Initially, we compared STAT1-Y701 (pSTAT1) activation from patients who underwent IFN dose reductions to patients who did not undergo dose reductions (Table 1 data not shown) before and after the HDI induction phase. Both unpaired and paired analyses showed no significant changes in STAT1 activation between the patients who had dose reductions and patients who did not undergo dose reductions, and from day 0 to day 29 with HDI therapy. Subsequently, we addressed whether Type I IFN signaling differed between HDI clinical responders and clinical non-responders at day 0 or day 29 after HDI therapy. The induction of pSTAT1 from IFN-α stimulation was assessed by phosflow in PBMCs by examining the overall median fold change. The median fold change of IFN-α induced pSTAT1 in PBL from responders was lower than non-responders on day 0 (Figure 1A), which was observed in both CD4 and CD8 T cells, but these differences were not statistically significant. In CD19 B cells, a statistically significant difference was observed in the median fold change of pSTAT1 induced by IFN-α in responders and non-responders on day 0 (Figure 1A). The median fold change of pSTAT1 induced by IFN-α in lymphocytes and lymphocyte subsets showed little or no difference between responders and non-responders on day 29 (Figure 1A).

We further investigated within the HDI-responding and non-responding patients on day 0 and day 29 and examined the effect of HDI therapy on Type I IFN signaling. Paired samples from pre-treated HDI PBMCs were compared to lymphocytes from their corresponding post-HDI treated PBMCs in both melanoma responding and non-responding patients. Two-sided Wilcoxon-Mann-Whitney paired analysis demonstrated that there was a statistically significant increase in STAT1 activation after HDI treatment in the responding group, and the response was equally observed in lymphocytes overall and in CD8 T cells (adjusted p-values 0.039, respectively). CD4 T cells were on the cusp of significance (adjusted p-value, 0.052) and B cells showed no significant differences in STAT1 activation (adjusted p-value, 0.30) (Figure 1B). In contrast, there were no significant differences in the response levels within any lymphocyte subset of non-responders from pre- to post-HDI treatment (Figure 1C). To determine whether the overall response was due to the basal levels of pSTAT1, we analyzed changes in pSTAT1 in PBLs in unstimulated cells from before to after HDI treatment. We found no significant differences in the basal expression of pSTAT1 from pre to post HDI treatment in both the responding and non-responding patients (Figure 2A-B).

It was observed that disease-free and overall survival was longer amongst patients with a clinical response at day 29 compared with non-responders, although the results did not reach statistical significance [12]. We investigated the association of long-term clinical outcome with IFN-α induced pSTAT1 levels in HDI treated melanoma patient lymphocytes from Day 0 to Day 29. Paired-analysis was used to compare patients who showed no evidence of disease (NED) or who developed subsequent metastatic disease (MET) at the time of clinical follow-up. NED patients demonstrated a significant increase in the activation of STAT1 from Day 0 to Day 29 in lymphocytes and both CD4 and CD8 T cells (adjusted p-values, 0.04 [Figure 3A, Additional File 2 Figure S2]). In contrast, lymphocytes from MET patients did not show a consistent increase in the induction of pSTAT1 from IFN-α stimulation from pre- to post-HDI therapy (Figure 3B).

Since STAT1 activation correlated with long-term clinical outcome, we further examined pSTAT1 responses in lymphocytes with disease-free and overall survival. A ratio for each patient's lymphocytes was determined by dividing the fold change in pSTAT1 post-treatment by the fold change in pSTAT1 pre-treatment. A median of these ratios (1.25) was generated using all patients in the study (n = 14) and patients were segregated according to whether they fell within a range of ± 0.1 around the median. Patients whose pSTAT1 ratios fell within this range had better disease-free and overall survival (Figure 4A-B, respectively) as compared to patients who had minimal or negative signaling changes (median <1.15), and interestingly, also patients who had larger increases from pre- to post- HDI treatment (median >1.35).