Patient’s selection and classification
This study was recruited IHD patient, which were admitted for coronary artery bypass grafting (CABG) in South City and Aga Khan University Hospital of Karachi. The samples were collected from both hospitals after obtaining written informed consent from the patients. The ethical approval boards of South City, Aga Khan University Hospital and Institutional Review Board (IRB) of ICCBS approved the current study. All methods were performed in accordance with the relevant guideline and regulations approved by the committee. The assessment of systolic and diastolic function with left ventricular (LV) morphology was conducted by 2D Doppler echocardiography according to the guideline of ASE 2009.
Patients were classified into different categories according to the ejection fraction (EF < 45% and EF > 45%) and diastolic dysfunction by E/e values (> 15, 9–15 and < 8) and grades (0–1, 1A and 2) as depicted in Table 1. It was observed that the corrected p-values are significant for few patients` basic characteristics. Hence an unsupervised learning algorithm was applied on the concentrations of metals with reference to the age, weight, BMI and blood pressure (both systole and diastole). The results in the form of PCA score plot are shown in Supplementary Figure S1. No significant separation trend/grouping was observed in all plots. Dysfunction of diastole13,14 were categorized with different grades include impaired relaxation of LV with or without increased in filling pressure indicated by grade I or IA, moderate increased in LV filling pressure is linked with the pseudo-normalization of LV i.e. grade II and increased in filling pressure markedly is the restrictive LV filling denoted as grade III. Patients with malignancy, constrictive pericarditis and infiltrative, established pulmonary disease, renal insufficiency, moderate to severe valvular disease, hypertrophic cardiomyopathies and metabolic bone diseases were omitted from the study It is possible that the screened biochemical parameters are influenced by the cardiac dysfunction either systole or diastole. Therefore, we tested this hypothesis and results are incorporated in Supplementary Table 1. It was clearly observed from the table results that the biochemical parameters of the recruited patients are not statistically significant different with reference to the ejection fraction values and grades.
Healthy control subjects (n = 55) were selected based on the following criteria; subjects were recruited randomly from the community if they had no symptoms suggestive of IHD, no past history of IHD or any proven past myocardial infarction or coronary intervention and without any symptoms of functional class I.
Sample collection and processing
BD vacutainer tube based on gel (Cat # 367381), was used to transfer 4 ml of blood, which was collected from the subject under investigation. After standing the BD vacutainer tube for 10–15 min, serum samples were separated by using centrifugation at 2000 rpm for 10 min. After centrifugation, aliquots of serum were transferred into locking Eppendorf tube. Serum samples were kept in freezer at − 80 °C until sample analysis.
Reagents and standards
During the experimental work, filtered and extremely pure deionized water was obtained from the water filtration and purification system (Thermo scientific, MA USA). Analytical grade reagent (AR, ACS), 70% concentrated HNO3 (RCI Labscan Ltd, Bangkok, Thailand) was used for analysis after purification by NanoPure Acid purification system (Nanonex, USA). Trace metal grade ≥ 30% concentrated H2O2 was acquired from Merck KGaA company (Darmstadt, Germany). The tuning solution with 1 µg/L concentration of Mg, Li, Tl, Y, Co and Ce in 2% HNO3, multi-element calibration solution 2A (Part Number: 8500-6940) with confirmed concentration of 10 µg/mL of each element and, internal standard of 100 µg/mL (Tb, Rh, Ge, Lu, Ir, Bi, and Sc) were bought from Agilent Technologies (Santa Clara, CA, USA). The optimization of ICP-MS parameter was carried out by using tuning solution, before the start of analysis. All cleaned glassware and polypropylene bottles were immersed in 10% (v/v) HNO3 reagent for overnight. After that each apparatus was washed with ultra-pure water three times and kept in laminar-flow hood (Airstream ESCO, Singapore) to dry. The air contamination was prevented by performing experiment in clean hood and working table.
Preparation of standard solutions
Matrix solution prepared as 5% nitric acid was used to prepare calibration standards for 16 elements. The 100 µg/L solution of internal standard was made in matrix solution, from the prepared stock solution. Sixteen points calibration curves were prepared in the range of 0.0076–1,000 µg/L of each metal using 5% HNO3 as matrix. A blank was also prepared with same matrix and no metal added. The measurement of sensitivity was done by checking the slope of regression equation. The standard solution was used to validate the current study by calculating correlation coefficients, LOD and LOQ.
Preparation of the standard reference material
The developed method was validated by measuring precision and accuracy of the result obtained from trace metal Seronorm serum L-1 (Sero, Billingstad, Norway). The complete procedure for preparation of certified reference material (CRM) was followed as given in the protocol of manufacturer. For preparation, 3 mL of sterile deionized water was added to dissolve the content of vial by rolling for 30 min so that all the content is mixed completely. The content was transferred to screw cap plastic tube and diluted with sterile deionized water. Each trace and ultra-trace element were analysed in triplicate in diluted CRM samples by ICP-MS.
Sample preparation for ICP-MS
The digestion of serum samples were carried out in pressure sealed microwave system with 64MG5-T64 rotor (Anton Paar GmbH, Austria). It was equipped with high performance pressure released system. The program with Multiwave ECO software (version 1.51), was setup in the microwave system. The single time useable screw cap (13–425, Wheaton 15 × 45 mm) and polytetrafloroethylene (PTFE) lip seal tube were used as standard apparatus. For digestion, serum sample equivalent to 50 µL aliquot was added in the MG5 vials (Anton Paar, Hungary), where 50 µL of ≥ 30% H2O2 and 150 µL of 70% HNO3 were mixed, kept in laminar fuming hood for 15 min so that fumes were evolve from the vials. Then, vials were sealed with the PTFE lip and screw cap. After that sealed vials containing serum samples were placed and digested in two steps by adjusting same parameters in Anton Paar microwave system with same ramp (10.0 min), hold (30), fan (1), power (850), stir rate (medium) but temperature set 90 °C in step 1 and 150 °C in step 2. All MG5 vials were kept in laminar hood and wait until all samples were cooled at room temperature, as digestion process completed. The vials were sealed with septum so that pressure of gas released by penetrating steal pin in septum then, the resultant samples were took into 15 mL of autosampler polypropylene tubes and deionized water were used to dilute samples upto 3 mL. All the samples were analysed in triplicate for every element. Matrix correction was done by using internal standard solution, to check the accuracy of ICP-MS results.
Inductive coupled plasma–mass spectrometry (ICP-MS) analysis
The quantification of chosen element was carried by Agilent 7,700 × ICP-MS system (Santa Clara, CA USA). The decontamination was removed from ICP-MS after each sample analysis by using washing solution contains 0.1% HCl, 2% HNO3 and ultra-pure water. The workstation used for the operation of ICP-MS data is Mass Hunter software. ICP-MS parameter are given in the Supplementary Table 2.
The chemometric analysis was done in multiple steps by Mass Profiler Professional (MPP), which were bought Agilent technologies (Santa Clara, CA, USA) with complete licenced. The absolute abundance was adjusted with 10,000 counts for filtration process. The normalization of data was carried out by external scalar to allocate the values for each sample particularly scale up and down. The selection of baseline i.e. Z transform, is important to treat all components equally, whatever their intensity.
Statistical analyses were performed in-between healthy and ischemic heart disease patient samples; characterized by systolic dysfunction by ejection fraction < 45% and > 45%, and diastolic dysfunction by grades and E/eʹ. Unpaired t test was used for statistical analysis of two groups while ANOVA was employed for more than two groups’ analyses. p-value was calculated by Asymptotic computation method and Benjamini–Hochberg FDR was used for multiple test corrections. All variables with p values < 0.05 and fold change > 1.5 will be considered as significant variables throughout the manuscript.
Multivariate data analyses including unsupervised principal component analysis (PCA) (shows an overview and outlier behaviour) and supervised partial least squares discriminant analysis (PLS-DA) and orthogonal partial least squares discriminant analysis (OPLSDA) were performed on processed data by SIMCA MKS Umetrics AB (version 14.1) software.
Method validation and data quality assessment
Linear calibration curve was obtained in the concentration range of 0.0076–1,000 µg/L for every element. The least-square regression method was used to plot the data of counts per second (cps) against the measured concentrations. In the validity of the quantification method was evaluated by performing various parameters such as correlation coefficients (R2), limit of quantification (LOQ) and limit of detection (LOD). The excellent linear relationship was obtained from the calibration curve (n = 3) with correlation coefficients (R2) between 0.993 and 1.000. LOQ and LOD was calculated using equation LOD = 3.3σ/S and LOQ = 10σ/S, where (σ) standard deviation of residual a regression line and (S) slope for each element. LOD and LOQ were found for all selected element in the range of 0.002–9.551 µg L−1 and 0.006–28.941 µg L−1 respectively. Supplementary Table 3 compiled the regression equation, R2, LOD and LOQ of each element.
To check the precision and accuracy for the developed method, Seronorm trace elements serum L-1 was used as reference standard material (CRM) for trace metal analysis in bio samples. It was found that all observed values agree with the certified values with nonsignificant variance between them. The % recovery of selected element was varied between 81.879 and 113.779, which are given in Supplementary Table 4.
For the assessment and validation of any analytical technique, it is necessary to perform spike recovery procedure. The analyte detection was monitored by checking the variation between the diluent used to prepare the sample and standard solution. The spike recovery test was also performed in real serum samples of two concentration levels for each element, to check the reliability and precisions (RSD, %) of our method. Mostly, the precision was found below 10%. The following relationship used to measure the precision of the method, Precision (RSD, %) = [standard deviation (SD)/CM] 100, where CM for measured concentration and SD means standard deviation. For serum samples, the coefficient of variation (% RSD) was obtained from 0.052 to 7.580 as shown in Supplementary Table 5. Thus, our method, is sensitive, accurate with good precision and can be employed for the routine metals analysis in biological samples.
Compliance with ethical standards
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.