The general characteristics of study group (n = 26) at baseline visit (and comparison with control healthy group) are presented in previous manuscript [5]. PAH group median SUVRV/SUVLV ratio was 1.02 [0.42–1.21] and control group was 0.16 [0.13–0.25], p < 0.001. At baseline visit patients with PAH had low BNP level (78.8 pg/ml [46–282]) and mean distance in 6-min walking test was 382 ± 102.3 m. Most of the enrolled patients were in WHO Class III and mPAP was 48.9 ± 18.7 mmHg.

In current study, we observed PAH group for 48 months and verified previously published results considering prognostic role of SUVRV/SUVLV. In total, 19 patients (73%) had experienced CEP. Mean time to clinical worsening was 25.2 ± 16.1 months. ROC analysis revealed that in extended observation cut-off value of SUVRV/SUVLV parameter was 0.54 (AUC: 0.86 (0.69–0.98), p < 0.001), suggesting that even subtle shift in cardiac glucose metabolism in PAH may result in worse prognosis (median SUVRV/SUVLV in control healthy group was 0.16 [0.13–0.25]). PAH patients with SUVRV/SUVLV higher than 0.54 had significantly worse prognosis in long term (log-rank test, p = 0.0007, Fig. 3).

Kaplan–Meier curves presenting deterioration-free survival in patients with pulmonary arterial hypertension based on SUVRV/SUVLV ratio in 48 months of observation, log-rank test, p = 0.0007. °, complete events; + , censored events; LV Left ventricle, RV Right ventricle, SUV Standardized uptake value

There were eight deaths in total, 30% of baseline number of patients – all of them were related with rapid PAH progression. Analysis of small subgroup of PAH patients who died revealed that they had significantly higher WHO class at baseline visit (2.75 ± 0.46 vs 2.0 ± 0.50, p = 0.01), BNP concentrations (456 ± 280 vs 164 ± 47 pg/ml, p = 0.006) and lower 6MWT distances (312 ± 90 vs 425 ± 83 m, p = 0.01). No other significant changes in hemodynamic or imaging parameters were observed. However, seven patients who died (88%) had either a SUVRV/SUVLV ratio > 1 or a RVEF < 40% at baseline, what confirms our previous results about the prognostic role of PET/MRI patients’ assessments [5].

After 24 months from baseline visits, we performed first structured follow-up (FU-1) visits. Twenty PAH patients remained in the study group (4 deaths, 2 patients did not agree to participate in the follow-up visits). Patients’ characteristics and the comparison of functional, PET/MRI/RHC parameters obtained at all three visits (baseline, FU-1, and FU-2) are presented in Table 1. At FU-1, we observed significant change of MRI-derived RVEF (45.1 ± 9.6 to 52.4 ± 12.9%, p = 0.01), and improvement in hemodynamic parameters obtained from RHC e.g., mPAP (50.5 ± 18.3 to 42.8 ± 18.6 mmHg, p = 0.03) and PVR (8.9 ± 5.7 to 7.15 ± 4.2WU, p = 0.04). Mean change of SUVRV/SUVLV (follow-up scans to baseline) was − 0.20 ± 0.74 confirming that most of the patients had decreased glucose metabolism in RV cardiomyocytes. Importantly, there was no significant change in RV mass/BSA parameter. Median FDG uptake of this group, presented as SUVRV/SUVLV tended to decrease from 0.94 [0.48–1.38] to 0.62 [0.41–1.16], p = 0.19, but 7 patients (35%) still had SUVRV/SUVLV ratio higher than 1. Interestingly, patients who had improvement in SUVRV/SUVLV (lower follow-up value than baseline, n = 12) had significantly higher baseline mPAP (56 ± 20.5 vs 42.4 ± 10.9 mmHg, p = 0.04) but improved cardiac index at FU-1 (from 2.47 ± 0.45 l/min/m2 to 2.98 ± 0.39, p = 0.007). Follow-up SUVRV/SUVLV ratio significantly correlated with follow-up RV hemodynamic parameters confirming strong relationship between RV function and metabolic alterations (Table 2).

Furthermore, at FU-1 visit it occurred that 16 patients (61%) had experienced clinical end-point during 24-months follow-up (4 deaths, 12 had clinical symptoms of PAH progression). CEP patients presented lower 6MWT distance (p = 0.01), higher SUVRV/SUVLV ratio (p = 0.005) and mPAP (p = 0.0002) and lower RVEF (p = 0.002), what is consistent with our previous results [5].

All CEP + patients had PAH therapy escalation between baseline and FU-1 visits—twelve patients started parenteral PGI (treprostinil or epoprostenol), one patient oral PGI analogue (treprostinil), two patients had added second-line drug – macitentan and one – inhaled iloprost. Five patients, who eventually died in that time, had PGI treatment initiated during hospitalizations preceding death, thus only eight patients on PGI therapy continued the study.

We observed the study group for another 24 months (48 months in total since baseline visit). Between FU-1 and FU-2, 10 patients had CEP (three patients for the first time in the study) including four deaths.

It occurred that second assessments of SUVRV/SUVLV ratio and RVEF done at FU-1 visits had also prognostic significance. ROC analysis revealed that both SUVRV/SUVLV and RVEF had similar prediction for second CEP (AUC: 0.8 (0.60–0.96) vs 0.93 (0.81–0.97), p = 0.24, respectively). Patients with second SUVRV/SUVLV > 1 had significantly worse prognosis after FU-1 visits (log-rank test, p = 0.006, Fig. 4).

Kaplan–Meier curves presenting deterioration-free survival in patients with pulmonary arterial hypertension based on SUVRV/SUVLV ratio after follow-up PET/MRI scans, log-rank test, p = 0.006. °, complete events; + , censored events; LV Left ventricle, RV Right ventricle, SUV Standardized uptake value

At FU-2 visits a significant improvement we observed only in RHC parameters (mPAP, PVR, CI) in the group of sixteen patients who were present at FU-2 (Table 1).

Out of twelve PAH patients who had improvement in SUVRV/SUVLV ratio (mean improvement − 0.59 ± 0.39; 33.86 ± 25% of baseline value) between baseline and FU-1 visits, eight had PAH therapy escalation to PGI (initiation of parenteral PGI in seven patients, oral PGI in one patient). Subgroup of patients who eventually received PGI treatment had baseline higher SUVRV/SUVLV ratio (1.09 [0.78–2.1] vs 0.52 [0.29–1.32], p = 0.03), mPAP (58.42 ± 15.96 vs 42.3 ± 15.5 mmHg, p = 0.01), and lower RVEF (37.82 ± 9.05 vs 48.82 ± 9.11%, p = 0.009).