Eighty-nine subjects undergoing coordinated coronary angiogram (Department of Cardiology of the Hospital Privado de Comunidad, Mar del Plata, Argentina) were included. Aortic valve disease, left ventricular (LV) outflow tract obstruction and/or arrhythmia were exclusion criteria. Prior to the study, a clinical evaluation enabled assessing the exposure to cardiovascular risk factors (18–23). All the included subjects gave their written informed consent. Data included in this work were not considered in prior publications. The protocol was approved by the Institutional Ethic Committee. The procedures agreed with the Declaration of Helsinki.

The following data were obtained: (1) invasive aortic BP (aoBP) and bBP (catheterization), (2) non-invasive bBP and aoBP, levels and waveforms assessed from oscillometric/plethysmographic brachial artery data (Mobil-O-Graph device, Model PWA, IEM GmbH, Stolberg, Germany).

Intra-arterial aoBP and bBP levels and waveforms were obtained with the subjects lying in supine position. Asepsis of the area, followed by cutaneous/subcutaneous injection of lidocaine was performed prior to the arterial (radial) access. Then, a 5 or 6 French introducer sheath was positioned in the arterial lumen and heparin was administered. After that, a 0.035-inch guide wire was placed in the ascending aorta and finally a 5 French pig tail catheter (Cordis, Miami, USA) was introduced. The catheter tip was always placed ∼4 cm away from the aortic valve. Once the correct positioning of the catheter was verified (fluoroscopy), the guide was removed and the catheter was washed with saline solution. Soft sedation was administered during the catheterization to minimize pain and discomfort.

To obtain intravascular (proximal ascending aorta or brachial) pressure, the fluid-filled catheter was connected to an external transducer (TruWave, PX260, Edwards, Dominican Republic), associated to a Mindray Mec 2012 system (Shenzhen Mindray Bio-Medical Electronics Co., China) which was synchronized with a x-ray device (Allura CV-20, Philips Healthcare, Netherlands). The external transducer was always kept at the heart level (mid-axillary line) and was calibrated in agreement with the system's inbuilt two-point calibration technique. The Allura CV-20 monitor allowed a display of the registered BP waves. Prior to any record or measurement the system was flushed with saline solution and the quality of the pressure signals was visually checked.

After obtaining aoBP data, the catheter was placed in the brachial artery (opposite to that of the limb of the vascular access), at the level in which the cuff for bBP measurement was positioned. Then, invasive intra-arterial bBP was measured and non-invasive (oscillometry-derived) bBP values were obtained immediately before or after the invasive recordings. After each bBP recording, the catheter was placed in the ascending aorta to check hemodynamic stability.

From invasive BP data, the processing systems enabled HR, systolic and diastolic BP values to be obtained.

After data collection, the catheter was withdrawn and the patient was sent to the recovery area.

Immediately before and/or after each invasive aortic or brachial record, bBP was non-invasively determined from a pneumatic cuff positioned in the arm opposite to that of the vascular access (oscillometry/plethysmography, Mobil-O-Graph device) (20, 24, 25). The system obtains bMBP (and HR) and after applying internal algorithms (manufactureŕs property) it gives systolic (bSBP) and bDBP (but not the bMBP), from which pulse pressure can be calculated (bPP, bPP = bSBP-bDBP).

From the data obtained, bMBP was quantified as follows (1, 15, 26, 27): (i)

bMBP_0.42= 0.42*bSBPosc + 0.58*bDBPosc

Using the equation proposed by Chemla et al. (11), and considering the different methods used to calculate MBP, aoSBP levels were obtained from invasive aoBP, invasive bBP and non-invasive bBP recordings. The non-invasive bBP data used to calculate aoBP were obtained simultaneously with the invasive aortic recordings. Then, as an example, aoSBP obtained from non-invasive bBP was named according to the approach used to quantify bMBP: (i) aoSBP_0.42, (ii) aoSBP_0.412, (iii) aoSBP_0.33, (iv) aoSBP_033 + 5, (v) aoSBP_0.33HR, (vi) aoSBP_SBP*DBP^0.5 and (vii) aoSBP_Osc_.

The studied subjects were distributed over a wide range of ages (37–85 y) (Table 1; Supplementary File S1, Supplementary Table S1). Invasive aoSBP and bSBP values were also distributed across a broad range: 7.8% and 4.5% were <100 mmHg, 53.9% and 48.4% were between 100 and 139 mmHg, 21.3% and 30.3% were between 140 and 159 mmHg, and 16.9% and 15.7% of the values were ≥160 mmHg. In turn, invasive aoDBP and bDBP values were <60 mmHg in 21.3% and 25.8% of cases; they were between 60 and 84 mmHg in 68.5% and 64.0%, and in 10.1% and 8.9% of cases the values were >85 mmHg. HR values were always within the expected (normal) range.

Related with the above, when analyzing the errors observed when considering different aoSBP levels it was found that (Table 3, Top): (i) all approaches underestimated aoSBP (error range: 22–51 mmHg) at high aoSBP levels (e.g., close to 220 mmHg, the maximum measured). The highest bias was obtained when using FF = 33% without correction for HR; (ii) at low invasive aoSBP levels (e.g., close to 88 mmHg, the lowest value measured) all approaches overestimated aoSBP (error range: 8–20 mmHg); (iii) for aoSBP values within 25th and 75th percentiles (116–151 mmHg), non-invasive approaches allowed reaching errors <10 mmHg (except for the method using a FF = 33%). The calculation of aoSBP using bMBPosc would enable to minimize errors when considering high invasive aoSBP levels (Table 3, Top); but the use of bMBP calculated using FF = 33% corrected for HR, FF = 42% or 41.2%, would result in acceptable bias.

Table 3 (Bottom), shows that different non-invasive approaches had different aoSBP ranges in which they “ensured” errors between (i) −5 and 5 mmHg (green), (ii) −10 and 10 mmHg (yellow) and/or (iii) −15 and 15 mmHg (orange). For instance, estimating aoSBP from bDBP and bMBPosc, allowed ensuring errors <10 mmHg when invasive aoSBP levels were between 115 and 183 mmHg, while the calculus of bMBP using an FF = 33%, would ensure reduced errors within an aoSBP range between 82 and 127 mmHg. In general terms, the remaining methods (FF = 33% corrected for HR, FF = 42% or 41.2%) resulted in errors <10 mmHg, within a pressure range of 100–110 (lower limit) and 155–165 mmHg (upper limit). In summary, the different approaches used to calculate aoSBP: (i) showed differences in “global” mean bias, (ii) over- and under-estimated aoSBP at low and high BP levels, respectively, and (iii) showed differences in the aoSBP range in which they would perform best as aoSBP estimators.

Chemla et al. developed their equation by comparing invasive (cathetersm) measurements in the ascending aorta with invasive measurements in the radial and brachial arteries. The validation of the Chemla et al. equation for cuff-based oscillometrically measured bBP values is still pending. Additionally, it remains to be assessed to what extent the method or equation used to obtain the bMBP levels has an influence on the validity of the Chemla et al. equation. In this work we applied, for the first time using invasive and non-invasive records, the method proposed by Chemla et al. and analyzed the obtained data with the aim of contributing to define to what extent the approach considered to determine the bMBP values to be used to calculate the aoSBP according to the method would impact on the accuracy and validity of the estimated data. The main contribution of this manuscript is the demonstration that the usefulness of the method recently proposed by Chemla et al. would be (i) highly dependent on the approach used to quantify bMBP, and (ii) on the aoBP levels considered. Our work highlights four issues.

First, the ability of Chemla et al. (11) approach to obtain aoSBP values close to those measured invasively depends on the way in which bMBP is obtained (measured or quantified) and on the actual aoSBP levels in the specific subject. Then, trying to generalize and define dichotomously whether the approach is “good or bad” without taking into account the above would be a mistake (and an over-simplification). The different approaches used to calculate aoSBP: (i) showed differences in “global” mean bias, (ii) over- and under-estimated aoSBP at low and high BP levels, respectively, and (iii) showed differences in the aoSBP range in which they would perform best as aoSBP estimators.

Second, calculating bMBP using a FF = 33% without HR adjustment (the most widespread way of calculating the MBP), would result in aoSBP values far from the invasive ones. Furthermore, that approach gave the highest mean error levels. Additionally, compared to other approaches, its best performance (lowest error) was observed at low aoSBP levels (Table 3, Bottom) which would be mainly observed in haemodynamic states or clinical situations in which assessing central haemodynamics could not be considered decisive (e.g., in terms of clinical decisions). On the other hand, and in the same line, at least in theory, aoSBP_033 could be considered useful to assess aoSBP in children and adolescents who have low aoSBP. However, it would not be useful in children/adolescents exposed to clinical conditions and/or risk factors (e.g., sedentary, overweight-obesity) in which aoSBP levels have been shown to be elevated (25, 31–33). This should be evaluated in future studies.

Third, the other methods used to calculate bMBP and/or bMBPosc, showed (quite) similarity in their ability to estimate aoSBP. Unfortunately, most brachial cuff-based methods (oscillometric devices) do not give bMBPosc, even though it is quantified as a prior step to the obtaining of bSBP and bDBP (the values actually given). In fact, most of the oscillometric devices do not show the researcher or clinician the bMBPosc (e.g., Mobil-O-Graph, Omron semi-automatic BP devices). These devices show on the display the HR and bSBP and bDBP values (calculated with the manufacturer's own internal algorithms). Then, the researcher and/or clinician can only quantify bMBP using equations such as those used in our manuscript. In other words, the systems “measure” the bMBPosc (as is widely known), but then use it to calculate bSBP and bDBP values, which are the values shown, and do not display the measured data. Therefore, the bMBPosc related approach while accurate would be difficult to apply and generalize in clinical practice.

Fourth, methods using a FF = 33% with HR correction and/or a FF close to 40% (42% or 41.2%) may be one step ahead of the rest when jointly considering three factors: (i) agreement with invasive aoSBP, (ii) aoSBP range within which they ensure the lowest errors (100–110 to 155–165 mmHg), and (iii) feasibility to be applied in clinical practice.

First, healthy subjects were not included in this work. This is a common feature of this kind of studies given the conditions required for the indication of invasive evaluations (e.g., suspected or known cardiovascular disease) (2). However, and in line with the above, the studied subjects would be representative of those whose accurate hemodynamic and/or cardiovascular assessment would be considered critical in clinical practice.

Second, the sample size (n = 89) exceeded the minimum recommended for studies aimed at analyzing the agreement (e.g., Blant-Altman test) between invasive and non-invasive BP measurements (17). In addition, despite, the sample size could be considered moderate, it enabled to detect statistical differences, thus achieving suitable statistical power (minimizing type 2 errors). Measurements in the brachial artery opposite to that of the vascular access limb and the need for additional recordings in the aorta considered in the study protocol, increased catheterization-time, which restricted the number of patients considered elective and/or who agreed to participate.

Third, although we are aware that differences between measured and estimated aoSBP could vary depending on covariates (34, 35) neither the sample size, nor its heterogeneity allowed to define subgroups (e.g., defined by age, sex and/or exposure to risk factors) and perform adequate statistical analyses. Further multicentre studies would be necessary to analyze the impact of covariates on the results.

Fourth, we used “fluid column” transducers instead of solid-state pressure sensors, which characteristically provide accurate BP waveforms (mainly due to their ability to detect the high-frequency components of the wave). In any case, fluid column transducers are not only the sensors used in our Hospital but they are widely used to measure aoSBP in clinical practice. Furthermore, in the ARTERY Society task force consensus statement on protocol standardization, Sharman et al. stated that while micromanometer-tipped catheters would be the sensors of choice, if carefully handled, fluid column catheters could be used to measure intra-arterial BP (17). Additionally, recently, in a systematic review and meta-analysis, fluid-filled and catheter-tipped transducers have shown similar mean bias in non-invasive aoSBP estimation (2). Taking into account the natural frequency and damping coefficient of our recording (catheter-tubing-external transducer) system, and although the methods and devices used are widely validated, the systolic and diastolic BP values obtained invasively could entail a small over- and under-estimation, respectively.

Fifth, an issue to consider is that regardless of the method used, non-invasively assessed bBP always has “inherent errors” (e.g., under- and over-estimation of bSBP and bDBP, respectively) (36). Then, the ability of bBP to accurately quantify aoBP (using the method of Chemla et al.) may depend (among other factors) on the approach and device used. Additionally, taking into account the inter-individual differences in BP amplification and in the brachial pulse waveform, the form factor that should be used to properly calculate bMBP may vary (37). In this regard, Schultz et al. showed that no universal form factor would achieve an accurate estimation of bMBP in all individuals (37). Thus, our results regarding the best approach to quantify bMBP to be used to estimate aoSBP must be analyzed in the context of the overall scenario, as there may be differences among individuals.