Activating STAT3 mutations in CD8+ T-cells correlate to serological positivity in rheumatoid arthritis

Abstract

Objectives: Large granular lymphocyte (LGL) leukemia is a rare hematologic malignancy characterized by clonal expansion of cytotoxic T-cells frequent somatic activating STAT3 mutations. Based on the disease overlap between LGL leukemia rheumatoid arthritis (RA)a putative role for CD8+ T-cells in RA we hypothesized that STAT3 mutations may be detected in RA patient CD8+ T-cells correlate with clinical characteristics.

Methods: Blood samples, clinical parameters, and demographics were collected from 98 RA patients and 9 healthy controls (HCs). CD8+ cell DNA was isolated and analyzed via droplet digital (dd)PCR to detect STAT3 mutations common in LGL leukemia: Y640F, D661Y, and the S614 to G618 region. STAT3 data from 99 HCs from a public dataset supplemented our 9 HCs.

Results: RA patients had significantly increased presence of STAT3 mutations compared to controls (Y640F p=0.0005, D661Y p=0.0005). The majority of these were low variant allele frequency (VAF) (0.008-0.05%) mutations detected in a higher proportion of the RA population (31/98 Y640F, 17/98 D661Y) vs. HCs (0/108 Y640F, 0/108 D661Y). In addition, 3/98 RA patients had a STAT3 mutation at a VAF >5% compared to 0/108 controls. Serological markers, RF and anti-CCP positivity, were more frequently positive in RA patients with STAT3 mutation relative to those without (88% vs 59% RF, p=0.047; 92% vs 58% anti-CCP, p=0.031, respectively).

Conclusions: STAT3 activating mutations were detected in RA patient CD8+ cells and associated with seropositivity. Thus, STAT3 activating mutations may play a role in disease pathogenesis in a subset of RA patients.

Keywords: CD8-positive T-lymphocytes; JAK/STAT; anti-citrullinated protein antibodies; large granular lymphocytic leukemia; rheumatoid arthritis; rheumatoid factor; stat3.

Figure 1

Study overview: RA patients (n=150) were recruited for the study, and blood samples from HCs (n=9) were obtained. PBMCs were isolated from these samples. CD8+ T-cells were then isolated, DNA was extracted, and each sample was assayed using ddPCR for STAT3 Y640F and D661Y mutations as well as mutations in the S614-G618 region. After all processing and analysis was completed, 98 samples had sufficient material to determine mutation presence. Using a cutoff VAF of 0.05%, 82 were identified as WT and 16 as mutant STAT3. RA patient mutation status was then correlated with clinical parameters. Created with Biorender.com.

Figure 2

Sample ddPCR plots of STAT3 mutant detection in CD8+ T-cells: The STAT3 mutant region was pre-amplified from an input of 40,000 genomic copies. After the pre-amplification step, 40,000 amplified copies were used for the D661Y and Y640F mutation-specific ddPCR assays (A, B) and the S614-G618 region dropout assay (C). (A, B) The bottom right quadrant (green) shows droplets containing the WT product, the top left quadrant (blue) shows those with the mutant product, and the top right quadrant (orange) shows droplets with both WT and mutant DNA detected in the same droplet. (C) The S614-G618 assay was designed as a dropout assay where the mutant-detecting probe spans the entire region encoding for this range of amino acids, while an adjacent WT probe detects which droplets contain DNA. Green droplets have full intensity for each probe (WT), while blue droplets are those with diminished intensity indicating that a mutation exists in the probe-binding region.

Figure 3

External dataset confirms STAT3 mutation differences between RA and HCs: (A) Data from 99 HC samples (22) were added to our 9 HCs to give an expanded sample size of 108. The 98 RA samples were only from our current study and are the same as Table 2 (no RA samples were assessed in the Valori et al. study). Compared to RA samples, the combined HCs have statistically lower frequencies of Y640F and D661Y mutations (p=0.0005 and p=0.0005, respectively) by Pearson’s Chi-Squared test. (B–D) Logistic regression models to explain the presence of STAT3 mutations using age and RA status were built by binning mutation presence at a VAF cutoff of 0.008% (3 or more droplets detected). These models includes HC data from Valori, et al (22). RA patients were more likely to have (B) any mutation either Y640F or D661Y, (C) only Y640F, or (D) only D661Y mutations compared to HCs (p=1.74x10^-6, p=4.57x10^-5, p=0.005 respectively). Graphs (B–D) created in R.

Figure 4

STAT3 mutation correlates to seropositivity in RA patients: (A) RA patients with STAT3 mutations exhibited more RF positivity than WT patients (14/16, 88% vs 48/82, 59%, p=0.047 Pearson’s Chi-Squared). (B) RA patients with STAT3 mutations also had higher median serum RF values (152.5 U/mL vs 59.6 U/mL, p=0.099 Wilcoxon Test). (C) RA patients with STAT3 mutations exhibited more historical anti-CCP positivity than WT patients (12/13, 92% vs 46/80, 58% p=0.031 Pearson’s Chi-Squared). (D) RA patients with STAT3 mutations also had higher median anti-CCP values (173.5 U/mL vs 38.5 U/mL, p=0.339 Wilcoxon Test). (E) Following the same trend, the STAT3 mutant group exhibited slightly more current anti-CCP positivity (p=0.215) and (F) also had higher median anti-CCP values (305.7 U/mL vs 94.3 U/mL, p=0.177 Wilcoxon Test). (G) RA patients with STAT3 mutations exhibited more double positivity for both RF and anti-CCP than WT patients (11/13, 85% vs 39/80, 49% p=0.016 Pearson’s Chi-Squared). All graphs created in prism. Adjusted p-values for each statistical test are available in Supplementary Tables 1 - 3 . Star denotes statistical significance.