Background High-frequency longitudinal monitoring may help identifying GVHD and GVT and intercepting these events before their occurrence. Methods We implemented a targeted multiplex microfluidics q-PCR based transcriptional assay, able to measure the fluctuations of 264 transcripts (TFA, Chaussabel Nat Rev, 2014) with high frequency. This transcripts panel is not tailored to GVHD-known biomarkers, rather it is representative of a large selection of whole-blood transcripts, identified via co-expression analyses and annotated for function. Samples consist of 50 ul whole blood, self-collected by patients through fingerstick, and stored in an RNA-preservative solution ( Mathew, BMC Res Notes, 2020). Twenty-two patients were followed weekly from engraftment until 149 days after allogeneic hematopoietic stem cell transplantation (allo-HCT) (19-149, median 72) with this method in a discovery cohort (DC). Additionally, 55 patients were studied in an independent validation cohort (VC) with the same transcripts panel, but using Ficoll gradient-separated blood samples collected at fixed post allo-HCT time-points (days +14/+21/+28 and +100) and live-frozen. Transcriptome results were correlated with the event of acute GVHD (aGVHD), graded according to Glucksberg modified criteria, and further sub-classified as “active” when clearly measurable in a GVHD clinical score (not “silenced” by therapeutic intervention). In both cohorts, samples were analyzed by longitudinal statistical analyses, performed with 3 different Linear Mixed Models (LMM): Random Intercept, Random Slope and both Random Intercept+Slope. Additionally, each post allo-HCT outcome was studied by Functional PCA (FPCA) for individual transcripts. Results A total of 19 samples pre- and 234 post allo-HCT were suitable for analyses in the DC, (median 11 samples/pt; 4-18) and 32 samples pre- and 165 post allo-HCT in the VC (median 4 samples/pt; 1-5). Overall, 13 out of 22 patients in the DC and 29/55 in the VC developed grade 1-3 aGVHD. Patients without any sign of GVHD in 149 days (“Never GVHD”; n=85 in DC, n=72 in VC) were compared with patients having GVHD at any time point (“GVHD”; n=149 in DC, n=93 in VC) and with the subset of “active GVHD” samples within each cohort. As expected, different gene sets significantly changed over time in GVHD patients and in the active GVHD subset in the DC (including whole fresh blood cells and plasma) with respect to the VC (including Ficoll-processed cells). IFN-, cytotoxic T cell- and erythroid cells-related transcripts were mostly significant in the DC GVHD patients, while B cells and monocytes transcripts were more represented in the VC GVHD selected PBMCs. Cytokines and chemokines were represented in both cohorts, probably reflecting cellular and plasmatic RNAs in the DC and cells' internal RNA reservoirs in the VC. Most strikingly, only one set of transcripts, all related to neutrophil activation was robustly validated in both cohorts. CEACAM6 (CD66c) and CEACAM8 (CD66b) linked to the expression of neutrophil adhesion molecules, the defensin DEFA4 and the neutrophil elastase ELANE distinguished longitudinally the post allo-HCT outcome of patients affected by aGVHD of any grade. The FPCA analysis confirmed a clear distinction of aGVHD patients by these 4 genes (Figure 1). We studied then if this signature could prognosticate aGVHD. In the DC GVHD cohort, the earliest timepoint (tp2) pre-GVHD onset was available for 8 patients (median 22.5 days; 22-23). The same timepoint was available for 8 Never GVHD patients. CEACAM8 was significantly higher in the GVHD vs the Never GVHD cohort at this timepoint, thus preceding GVHD. Differently, in the VC the early day +14 timepoint was available for all patients and preceded the GVHD onset in all cases. All 4 neutrophil transcripts in this cohort were significantly higher in the GVHD vs the Never GVHD group. Thus, neutrophil activation anticipates the onset of GVHD and may be considered for future studies as a candidate aGVHD prognostic biomarker, especially in early post allo-HCT stage. In conclusion our generic blood immune profiling assay revealed that neutrophil activation is a robust diagnostic marker, consistently identifying aGVHD patients in their post-transplant course, regardless of the sample source (whole blood or Ficoll-separated PBMCs, fresh or frozen/thawed). Moreover, CEACAM8 (CD66b) may likely represent a prognostic biomarker of aGVHD.
Read full abstract