Data on plasma tumour and metabolism related proteins’ potential in differentiation of HFpEF-PH from PAH and in prognosis of left heart failure patients with pulmonary hypertension

The data in the current paper constitutes supplementary material to our article entitled “Plasma tumour and metabolism related biomarkers AMBP, LPL and Glyoxalase I differentiate heart failure with preserved ejection fraction with pulmonary hypertension from pulmonary arterial hypertension” Ahmed et al. (2021). The study investigated 69 plasma tumour- and metabolism related proteins in healthy controls (n = 20) and in 115 patients of whom 48 had pulmonary arterial hypertension (PAH; n = 48) and 67 with left heart failure with pulmonary hypertension (LHF-PH) [heart failure with- preserved ejection fraction-PH (HFpEF-PH; n = 31) and reduced ejection fraction-PH (HFrEF-PH; n = 36)]. The haemodynamic data were obtained with right heart catheterization, and clinical data from medical records. The present article describe the plasma levels of tumour- and metabolism related proteins, analyzed with proximity extension assay, along with their uni- and multivariable diagnostic and prognostic potential. High sRAGE levels univariably emerged as a negative prognostic marker in LHF-PH.


Value of the Data
• The data outline the potential of tumour-and metabolism related proteins in the diagnostic differentiation of heart failure with preserved ejection fraction with pulmonary hypertension (HFpEF-PH) from pulmonary arterial hypertension (PAH). The data also describe the potential of such proteins in predicting transplantation-free survival in patients with left heart failure and pulmonary hypertension (LHF-PH). • These data may aid clinical professionals working in the field of pulmonary hypertension in decision making to avoid or initiate PAH-specific therapy early to improve survival and quality of life in patients with PAH. • The present data provide cardiologists/pulmonologists and researchers working in the field of pulmonary hypertension and/or heart failure a basis for further investigations aimed at facilitating the diagnostic differentiation between PAH and HFpEF-PH. • Using these data, clinicians and researchers can obtain further insights into the potential utility of proteomics in diagnosis and prognosis of LHF-PH and PAH.

The plasma levels of tumour-and metabolism related proteins
The plasma levels -expressed in linear normalized protein expression (NPX) scale as arbitrary units (AU) -in the study group, as well as heart failure specific classification of tumourand metabolism related proteins are described in ( Tables 1 and 2 ).

Tumour and metabolism related proteins in assessing prognosis in HFpEF-PH and/or LHF-PH
To assess the proteins' crude prognostic performance in identifying events in HFpEF-PH or LHF-PH and to define their optimal plasma thresholds for Kaplan-Meier analyses, ROC analyses were conducted for all 36 proteins comparing death or transplantation (events, n = 53) vs eventfree survival (non-events, n = 14) ( Table 3 ). Endocan had the largest AUC in predicting death or transplantation, followed by sRAGE, IGF1R, FGF-23 and IGFBP7 ( Table 3 ).

Survival and Cox regression analyses
Patients with LHF-PH with plasma levels > threshold values for endocan, sRAGE, IGF1R, FGF-23 and IGFBP7 had a lower probability of event-free survival compared to levels ≤threshold values ( Fig. 1 B-F, log-rank p < 0.05). In the univariable Cox regression models, IGF1R was the strongest predictor of events per unit increase, followed by sRAGE, IGF1R, endocan and FGF-23 ( p < 0.05). However, after adjustment for age, sex, atrial fibrillation, and systemic hypertension in multivariable models, the prespecified biomarkers were no longer significantly associated with events ( p > 0.05), although plasma sRAGE (HR 1.017, 95% CI 0.99-1.036; p = 0.07) displayed such propensity ( Table 4 ).

Table 3
Receiver operating characteristic analysis illustrating the diagnostic and prognostic potential of tumour and metabolism related proteins. The proteins are sorted according to largest area under the roc curve (AUC). In the diagnostic approach, the proteins' levels that significantly differed heart failure with preserved ejection fraction with pulmonary hypertension (HFpEF-PH) from pulmonary arterial hypertension (PAH) and controls were selected for receiver operating characteristic analysis (ROC). As for the prognostic approach, all 36 proteins that differed left heart failure with PH (LHF-PH) and/or HFpEF-PH from controls were included. The use of HFpEF-PH or the LHF-PH group was according to proteins' classifications. Abbreviations: 5 -NT, 5 -nucleotidase; AMBP, protein AMBP (alpha-1-microglobulin/bikunin precursor); CA9, carbonic anhydrase 9; CI, confidence interval; FABP4, fatty acid-binding protein 4; FGF-21 and 23, fibroblast growth factor 21 and 23; IGF1R, insulin-like growth factor 1 receptor; IGFBP2, 3 and 7, insulin-like growth factor-binding protein 2, 3 and 7; LOX-1, lectin-like oxidized LDL receptor 1; LPL, lipoprotein lipase; LYPD3, Ly6/PLAUR domain-containing protein 3; PCSK9, proprotein convertase subtilisin/kexin type 9; PON-3, paraoxonase-3; RARRES2, retinoic acid receptor responder protein 2; S100A11, protein S100A11; SCGB3A2, secretoglobin family 3A member 2; sRAGE, soluble receptor for advanced glycation end products; TFF3, trefoil factor 3; WFDC2, WAP four-disulfide core domain protein 2; VSIG2, V-set and immunoglobulin domain-containing protein 2. ¤ PAH ( n = 48) and HFpEF-PH (31) patients; event = HFpEF-PH diagnosis. # LHF-PH ( n = 67) and/or HFpEF patients ( n = 31); event = death or heart transplantation. * The five proteins with highest prognostic AUC. Cut-off was defined either with closest top left or Youden's method AU, arbitrary units. Univariable and multivariable Cox regression analyses were conducted for the five proteins that rendered the largest crude area under the roc curve in differentiating event occurrence from event-free survival in left heart failure with pulmonary hypertension ( n = 67) or heart failure with preserved ejection fraction with pulmonary hypertension ( n = 31) patients, according to proteins' classifications. Five predictors were included in each multivariable model due to the limited number of events ( n = 53) and adjustments were done for age, sex and the two most common comorbidities i.e., atrial fibrillation and systemic hypertension. Abbreviations: AU, arbitrary units; CI, confidence interval; FGF-23, fibroblast growth factor 23; HR, hazard ratio; IGF1R, insulin-like growth factor 1 receptor; IGFBP-7, insulin-like growth factor-binding protein 7; sRAGE, soluble receptor for advanced glycation end products. * p < 0.05 Event = death or heart transplantation.

Table 5
Spearman's correlation analyses of potential prognostic or diagnostic biomarker candidates with NT-proBNP and hemodynamic parameters.

Study population
The study was based on Lund Cardio Pulmonary Registry (LCPR) a unique, single centre prospective cohort in Region Skåne's biobank, initiated by Göran Rådegran in 2011. LCPR contains blood samples of healthy controls, as well as haemodynamic-and clinical data of patients evaluated for pulmonary hypertension (PH) and heart transplantation in a PH-expert centre. The haemodynamic and plasma samples were collected between 2011 and 2017 at Skåne University hospital during the routine clinical assessments. The blood samples were stored at Region Skåne's biobank, Lund, Sweden. Transplantation free survival of patients with LHF-PH were followed-up until 2020.
In the present study, a total of 135 participants were included, comprising healthy controls ( n = 20) devoid from cardiovascular disease, as well as patients with pulmonary arterial hypertension (PAH) and LHF-PH ( n = 67); [HFpEF-PH, ( n = 31) and HFrEF-PH ( n = 36)]) [1] . A detailed description of the study population has been described elsewhere [1] . The plasma samples were analysed with proximity extension assay, a multiplex immunoassay based on the use of protein-specific pairs of oligonucleotide-linked antibodies and microfluidic qPCR for detection and quantification, conferring high specificity and sensitivity. The proteins were analysed using the Proseek Multiplex Cardiovascular II, III and Oncology II 96-plex immunoassays (Olink, Proteomics, Uppsala, Sweden). Per default, the proteins' levels are expressed in arbitrary log2 NPX scale, which is a relative quantification scale corresponding to the inverted Ct-value [1] . Due to non-normally distributed log2 scale values and for interpretation purposes, the conventional log2 values were linearized using the following formula: linear NPX (AU) = 2 (log2 NPX) . The proteins were analysed using the Proseek Multiplex Cardiovascular II, III and Oncology II 96-plex immunoassays (Olink, Proteomics, Uppsala, Sweden) [4] .

Validation of the analytical performance
The multiplex analyses of the 96-pelx immunoassays (Cardiovascular II, III and Oncology II) were performed, each by using a separate plate. For quality control, internal and external controls were used. In each well of the three plates, four internal controls were added for quality control of immunoreaction, extension, amplification, and detection. Additionally, external controls, comprising inter-plate controls and negative controls were added, each in three separate wells in every plate. The median of the inter-plate controls was used to monitor and normalize potential variation between runs and plates, whereas the negative controls, consisting of buffer with no antigens, were used to monitor potential background noise and to establish the proteins background levels (calculate the limit of detection). Validation of the analytical performance has been conducted for each panel's measuring range, specificity, sensitivity, precision, reproducibility and scalability (Olink proteomics, Uppsala, Sweden) [4] .
Although detection range and standardized curves "calibration curves" are available for almost all proteins -established during the validation of the analytical performance -these can only estimate the expected measuring range of the immunoassays and cannot be used to convert arbitrary NPX to absolute concentrations. Thus, detection range and standardized curves or (calibration curves), specifically for the current analyses, are not available. Nevertheless, standardized curves are not a requisite for interpretation as the proteins are measured in a relative quantification scale. Panel and protein specific validation data for Cardiovascular II, III and Oncology II panels can be downloaded at www.olink.com/downloads (Olink proteomics, Uppsala, Sweden) [4] . capillary PH (DPG < 7 mmHg and/or PVR ≤3 WU) or combined post-capillary and pre-capillary PH (DPG ≥7 and/or PVR > 3 WU) [6] . Left ventricular systolic and diastolic dysfunction were diagnosed with echocardiography and/or magnetic resonance imaging. The World Health Organisation functional class, 6 min walking distance, and other clinical parameters including mixed venous oxygen saturation, arterial oxygen blood saturation and arteriovenous oxygen difference were measured or performed in conjunction with the haemodynamic assessments. Patients' demographics were retrieved from the LCPR file. The estimated glomerular filtration rate was creatinine-based and calculated with the revised Lund-Malmö formula [7] .

Statistical analyses
Statistical tests were performed with R version 4.0.2, (Foundation for Statistical Computing, Vienna, Austria) and GraphPad Prism version 8.4.1 for Windows, (GraphPad Software, San Diego, California USA. Normality was assessed visually with histograms, and data is presented median (inter quartile range), unless stated otherwise. Due to the dominance of non-normally distributed data, non-parametric tests were employed. Kruskal Wallis tests with a following multiple comparison post-hoc analyses in addition to Mann-Whitney U tests were used to assess the differences in protein's levels between groups. To correct for mass significance, and to define significance threshold, false discovery rate (FDR) was used. Statistically significant results were either defined as p-values less than the achieved thresholds, or when FDR were not applicable, as p < 0.05. Area under the ROC curve (AUC) was used to choose the best prognostic proteins (highest AUC). Youden's index or closest distance to top left were used to define the ideal thresholds for potential prognostic proteins. Kaplan-Meier method was used to estimate eventfree survival in relation to potential prognostic biomarkers during the overall follow-up time, and the log-rank test to compare transplantation-free survival between groups. Transplantation-free survival was defined as survival free from death or transplantation. Censoring was applied to the end-of-study follow-up time. Assumptions of proportional hazards were graphically examined, and Cox proportional hazards regressions were used to estimate the univariable and the multivariable hazard ratios (HR) of potential prognostic biomarkers. Due to the limited number of events, five independent variables were used to adjust for age, sex and most common comorbidities in the Cox regression analyses. Spearman's rank correlation was used to associate NT-proBNP and haemodynamics with the plasma levels of potential prognostic biomarkers.

Eligible proteins for prognostic analyses
Briefly, the criteria applied to identify proteins eligible for diagnostic analyses were also applied to identify proteins eligible for prognostic analyses. The criteria are described more in detail elsewhere [1] .

Ethics Statements
All participants signed a written informed consent upon enrolment. The study has been performed in accordance with the declaration of Helsinki and Istanbul and was approved by the local ethical board in Lund, Sweden

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Dr. Salaheldin Ahmed and Dr. Abdulla Ahmed report personal lecture fees from Janssen-Cilag AB. Dr. Rådegran reports personal lecture fees from Actelion Pharmaceuticals Sweden AB, Bayer Health Care, GlaxoSmithKline, Janssen-Cilag AB and Nordic Infucare outside the submitted work.
Dr. Rådegran reports unrestricted research grants from ALF, a non-interventional investigatorinitiated study research grant from Janssen-Cilag AB during the conduct of the study.
Dr. Rådegran is, and has been primary-, or co-, investigator in; clinical PAH trials for Actelion Pharmaceuticals Sweden AB, Bayer, GlaxoSmithKline, Pfizer, Janssen-Cilag AB and United Therapeutics, and in clinical heart trans-plantation immuno-suppression trials for Novartis.
The relationship with the organisations described above played no role in the collection, analysis or interpretation of the data, had no right to restrict the publishing of the manuscript, and did not impose a competing interest.