A Novel Fluorescent Clinical Method to Rapidly Quantify Plasma Volume

In patients presenting with the potential of volume abnormalities, an accurate evaluation is critical for appropriate diagnosis and management. Unfortunately, volume assessment is challenging. Studies demonstrate that 10–20% of non-heart failure (HF) dyspneic patients are initially diagnosed with HF, i.e. they are incorrectly diagnosed with volume overload despite its absence [1, 2]. Similarly, in the 1,586 patient Breathing Not Properly study [3], adding bovine natriuretic peptide (BNP) to the history, physical exam, and chest X-ray improved the misdiagnosis rate, which remained high at 18%. These error rates occur because the clinical estimate of volume status is difficult and highly dependent on clinician experience and ability. These errors occur because the signs and symptoms usually considered suggestive of volume overload suffer from extremely poor inter-rater reliability. In a study of 68 emergency department patients, the kappa measurement of inter-rater agreement had an overall dismal performance. Although there was substantial agreement [4] for edema (κ = 0.67), agreement was only moderate for ascites (κ = 0.47) and crepitations (κ = 0.43), and was nonexistent for orthopnea (κ = 0.20), jugular venous distention (κ = 0.08), or the presence of a third heart sound (S3) (κ = 0.10) [5]. With such low inter-rater agreement, the practice of relying solely on clinical grounds for volume assessment is questionable.

Readily available rapid and accurate bedside techniques for measurement of a patient’s volume are limited. Bedside ultrasound can evaluate vena cava size, with consideration of respiratory variation, to aid in the assessment. However, the skill set to obtain these measures is not universally available. A number of hemodynamic measurements [6-9] that measure physiological surrogates of volume, e.g. pulse pressure [10, 11] and stroke volume [12, 13] variations – both indicators of fluid responsiveness, have been promoted. The currently available options for volume assessment include capillary refill, peripheral temperature, skin color, mean arterial pressure, urine output, creatinine, lactate, lung sounds, chest X ray, edema, weight changes, central venous oxygen saturation, and cardiac output monitoring [9, 14-17]. All of these parameters are surrogate markers of the adequacy of the circulating blood volume (BV) [14, 15]. Ultimately, the decision to administer or remove volume is almost always predicated on an assessment of volume status in the absence of a direct measurement.

Because of the need to accurately assess volume in real time, studies evaluating a new application of dye dilution technology to objectively determine its status offer promise [16-18]. Filtration assessment and surveillance technology (FAST) is able to measure the variables of plasma volume (PV) and glomerular filtration rate (GFR). These parameters are determined by dye dilution calculation after the injection of a pair of fluorescent molecules. In this application, a visible fluorescent injectate (VFI^TM), containing 12 mg of a large (150 kDa) carboxy-methylated dextran conjugated to a fluorescent rhodamine derivative, is the PV marker (PVM). It is simultaneously administered with 35 mg of a small (5-kDa) carboxy-methylated dextran conjugated to fluorescein that serves as the renal clearance marker (RCM). As the large molecule is retained in the intravascular space, its concentration reflects real-time PV, and since the small molecule is freely filtered across the glomerulus, its change represents the GFR.

As results for GFR measurements in this study population have been previously reported [19], the purpose of the present analysis is to assess the relationship of measured PV to that estimated by Nadler’s equation [20] and the serum stability of the measured PVM over time.

This was a phase 2b prospective, open label study (NIH, NCT03095391) observational study conducted at two sites: the University of Alabama at Birmingham and the Clinical Research Organization ICON (San Antonio, TX, USA). The study protocol and informed consent forms were approved by Western IRB in Puyallup, WA, USA. All patients received and completed an informed consent form prior to being enrolled in the study. Four cohorts (of 8 participants each) were enrolled between June 13, 2017, and August 30, 2017. All participants provided written informed consent, and the study adhered to the Declaration of Helsinki. Eligible participants were 18–75 years old, with body mass index ≥18 and ≤40, and nonsterile participants agreed to use medically acceptable methods of contraception.

Demographics, medical and surgical history (including concomitant medication use), height, weight, vital signs, and physical examination findings were documented. Laboratory tests included chemistry, hematology, hepatic function panel, follicular stimulating hormone (females only), creatinine phosphokinase, and HIV, hepatitis C and B serologies, as well as urine pregnancy tests and drug screens. 12-lead electrocardiograms were obtained. The estimated GFR (eGFR) was calculated using the CKD-EPI formula [21].

All studies were done in fasting patients in a clinical research unit. Subjects were provided a low-protein breakfast prior to the administration of the VFI and had free access to water throughout the study. All received a VFI that consisted of 12 mg of a 150-kDa carboxy-methylated dextran, conjugated to a rhodamine dye (PVM), and 35 mg of a 5-kDa carboxy-methylated dextran, conjugated to fluorescein (RCM), in a total volume of 3.0 mL. VFI was infused intravenously over 30 s. Subjects received a light, low-protein snack 5 h after dosing. All patients underwent PV determination within 21 days of screening.

Cohorts 1 and 2 consisted of healthy volunteers. Cohort 1 received a single VFI dose and measurement followed by a 350 mL 5% albumin in normal saline 30-min infusion beginning at 120 min, with plasma taken 5 min after the end of the infusion for determination of the change in volume. Cohort 2 resided overnight in the clinical research unit, receiving 2 VFI doses 24 h apart, to determine the variability in the determinations using a paired Student’s t test. They were fed a low-protein diet during their stay. Cohorts 3 and 4, representing chronic kidney disease (CKD) stage 3 (30≤ eGFR < 60 mL/min/1.73 m²) and 4 (15≤ eGFR < 30 mL/min/1.73 m²), respectively, received one VFI dose.

In all cohorts, 3.5 mL of blood was drawn prior to dosing, at 15, 30, 60, 120, 170, 310, 370, 480 min, and again at 12 and 24 h. While only the 15-min blood draw was used to determine PV, samples at the other routine time points were needed for determination of GFR by the FAST BioMedical technique (60 and 170 min) and iohexol technique (all 6 blood draws) as reported previously [19]. Additional blood draws varied by cohort in order to meet the specific experimental procedures prescribed by the protocol. In cohort 1, 350 mL of a 5% albumin solution in normal saline was administered over 30 min beginning at 130 min into the study, and additional blood samples were obtained at 165, 195, and 205 min to allow additional PV measurements after volume expansion. In cohort 2, the standard study was repeated a day later when a second-dose VFI was administered. The long-term pharmacokinetics of the PVM were determined in cohorts 2, 3, and 4, necessitating additional blood draws on days 1, 3, 7, 14, and 20 after the last VFI dose. All samples were collected in K2EDTA 3 mL tubes, from the opposite arm that VFI was administered. Samples underwent 1,500 g centrifugation for 15 min, after which 250 µL of plasma was pipetted and combined with a fluorescence-enhancing reagent (FAST BioMedical, Carmel, IN, USA) to total a volume of 2.250 mL. PVM and RCM dye concentration determinations were performed on a Turner Trilogy filter fluorimeter (Turner Designs, San Jose, CA, USA).

Pharmacokinetic parameters were calculated from the measured PVM and RCM plasma concentration data using noncompartmental methods (Phoenix WinNonlin, Version 6.3 or later, Pharsight Corporation, St. Louis, MO, USA) and actual sampling times. Using the 15-min time point, PV was determined by dividing the measured PVM concentration, in μg/mL, by the total dose given (12,000 μg). A measured hematocrit, taken prior to dosing on the same day of VFI dosing, was used to calculate BV by the following formula: BV = total PV/(1 – hematocrit) [22]. Nadler’s estimated BV was converted to PV by subtracting the volume occupied by the cells from the BV determination [21].

Subjects were followed for 21 (±1) days from last VFI injection and were assessed for adverse and serious adverse events at each encounter. Long-term monitoring was performed by blood sampling and pharmacokinetic analysis, with the median decay rate used to calculate PV > 1 h post-VFI injection. The correction used was determined by an equation: Adjusted PV = measured PV – ((measured PV × decay rate (0.0053)) × time).

Correlation analysis was done by comparing the measured PV to that estimated by Nadler’s formula for each cohort, using the Pearson product-moment correlation coefficient. Data are presented as Bland-Altman graphs and comparative graphs.

Table 1 shows the results in all 32 subjects studied under equilibrium conditions. Measured PV varied from a low of 2,115 mL to a high of 6,234 mL, with concomitant BV of 3,411 and 8,424 mL, respectively, in those 2 subjects. When adjusted for body weight, the ranges in PV and BV were 28.6–41.9 and 44.4–69.8 mL/kg, respectively. In healthy volunteers (cohorts 1 and 2), 6/16 (37.5%) were male, 8 (50%) Latino, 7 (43.7%) white, and 1 (6.2%) African-American, with a mean (± SD) weight of 79.7 ± 11.6 kg. Their mean plasma creatinine was 0.76 ± 0.17 mg/dL, and the mean eGFR was 99.6 ± 16 mL/min/1.73 m². The mean measured PV was 2,706 ± 370 mL and weight-adjusted PV was 36.2 ± 3.5 mL/kg, and for Nadler’s estimate the corresponding values were 2,672 ± 341 mL and 35.7 ± 2.7 mL/kg. The median measured PV was 2,640.3 mL versus that estimated by Nadler’s formula of 2,761.5 mL. In CKD patients (cohorts 3 and 4), 8 (50%) were male, 50% African-American, 7 white (43.7%), and 1 Hispanic (6.2%), with a mean (±SD) weight of 96 ± 18 kg. Their mean (±SD) plasma creatinine levels were 1.41 ± 0.30 and 2.75 ± 0.55 mg/dL, and mean (±SD) eGFRs were 50 ± 7.8 and 24 ± 3.3 mL/min/1.73 m² for cohorts 3 and 4, respectively. The mean measured PV was 3,368 ± 895 mL and weight-adjusted PV was 35.0 ± 4.5 mL/kg, and corresponding Nadler’s estimates were 3,448 ± 781 mL and 36.0 ± 3.5 mL/kg. The median measured total PV was 3,317 mL versus Nadler’s formula estimation of 3,448 mL. One patient in this group had a measured PV of 6,234 mL, making the standard deviation of the mean for this group very high. However, the weight-adjusted PV in mL/kg was similar to other subjects (Table 1).

Comparisons of the measured PV to Nadler’s PV estimating equation are shown in Figure 1, with an R² correlation coefficient of 0.85. If the highest PV patient is not included in the analysis, R² is 0.71 (suppl. Fig. 1; see www.karger.com/doi/10.1159/000496480 for all online suppl. material). A Bland-Altman analysis (Fig. 2) showed a symmetrical distribution around the mean. However, values as high as 555 mL above and 609 mL below Nadler’s equation estimated values were measured. This suggests that Nadler’s formula was up to 7.1 mL/kg (23.7%) above or 6.8 mL/kg (18.2%) below the FAST-measured PV.

Patients with consecutive-day PV measurements (cohort 2) had results that were stable (Fig. 3). The mean difference from day 1 to 2 was 98 ± 121 mL (p = 0.05), representing a mean change of 3.8%. In cohort 2–4 patients, followed for 21 days, the mean plasma half-life of the 150 kDa PVM was 95.3 ± 38.6, 99.1 ± 23.8, and 106.2 ± 22.5 h, respectively, suggesting that the presence of CKD did not alter metabolic elimination of the 150 kDa PVM.

Figure 4 shows the mean measured PV in mL/kg body weight for 24 patients (cohorts 2–4) over 6 h. Female and male subjects were separated into Figure 4a and b, respectively. Figure 4c shows the mean of all 24 patients over time. PVs were stable over time for individual patients and overall, but varied between patients. There was a tendency to see a rise in PV after the iohexol injection at 160 min, and the subjects received a small meal at 300 min. Subjects in cohort 1 were excluded due to the volume challenge administered and the subsequent alteration of PV during the 6 h.

Figure 5a and Table 2 show PV changes in response to a volume-expanding albumin infusion in cohort 1 beginning at 130 min after the start of the study. Five minutes after a 30-min infusion of 350 mL 5% albumin in normal saline, there was a mean increase of 355.8 ± 50.1 mL (range 294–427 mL), followed by a slow reduction thereafter. Figure 5b shows this weight-adjusted response in mL/kg body weight with the 350 mL volume challenge for individual patients.

The present study shows PV is accurately and reproducibly measured using a fluorescent 150 kDa PVM. We also described that the large inert PVM had a long half-life and could detect dynamic changes in PV induced by a volume challenge without re-dosing the patient for up to 6 h. Its use evaluating therapeutically induced changes, e.g. with diuretic administration or IV fluid bolus, will require additional study; however, the present study indicates it could be of significant benefit in a number of clinical situations where knowing PV and dynamic changes in PV are important.

Determination of PV and BV has been a clinical goal for over 70 years. Research studies using radioactive iodinated serum albumin (RISA) and chromium 51 erythrocyte (RBC) labeling have been used to measure PV and RBC volumes, respectively. These studies have identified many areas where measurement of PV and BV would be clinically useful including congestive HF [21, 23-26], sepsis [27], and surgical patients [28]. Research radioisotope measurements, using the indicator dilution technique have traditionally been considered the gold standard for BV and PV assessment [22, 29-32], but they are cumbersome and expensive, especially at the bedside. An FDA-approved BV measurement system (Daxor, BVA-100) has commercialized the RISA technique and utilized comparisons to the ideal BV weight method [33]. This technique uses I¹³¹ labeled albumin, requires multiple blood draws, is expensive, and only one BV determination is possible per injection of radioactive albumin.

The present fluorescent technique uses an inert carboxy-methylated 150-kDa dextran conjugated to a rhodamine-derived fluorescent dye as a PVM. The carboxy-methyl substitution markedly improves immunologic tolerance, as does the RCM [34]. In more than 89 subjects and 97 injections to date, there have been no serious adverse events described. Furthermore, the large size of the PVM eliminates vascular leakage resulting in a stable plasma concentration that allows for multiple determinations following a single injection for at least 6 h.

This technique is not without limitations. Its use in bleeding subjects may be limited, as the indicator dye will be lost with the extravasated blood and thus result in a falsely high PV calculation. Also, a high vascular permeability could result in leakage of the dextran thus limiting the time the marker could be used with accuracy. Additionally, determining whether or not a particular PV measurement is abnormal, for a particular patient, is not straightforward. To address this, the International Council on Standardization in Hematology have published a recommendation [35] on the interpretation of red cell mass and PV measurements and established prediction formulae based on body surface area for red cell mass. In general, 98–99% of measurements calculated using these formulae are within 25% of the predicted norm [31, 33]. One such formula is that described by Nadler et al. [20] using body surface area, where: total BV = (0.3669 × height (m)3) + (0.03219 × weight (kg)) + 0.6041 for males, and (0.3561 × height (m)3) + (0.03308 × weight (kg)) + 0.1833 for females.

Although current approaches to volume assessment [36-38] are entrenched in our daily practice, we believe a technique capable of rapid determination of PV (instead of surrogate markers of volume status and/or tissue perfusion) would represent a significant advance in the clinical management of critically ill patients. Patients with acute and/or chronic HF [39, 40], especially when complicated by renal insufficiency, frequently present difficult volume management problems. The coexistence of both heart and renal failure makes these patients particularly difficult to manage, and a small change in intravascular volume status can lead to a large change in overall clinical condition. A direct measurement of PV, used alone or in combination with surrogates of organ perfusion and/or measurements of blood flow, would guide fluid removal more precisely. In turn, this could potentially improve clinical outcomes, the ultimate test of any clinical technique. Given this technique also measures GFR accurately when compared to a 6 h iohexol measurement [19], its use in cardiorenal patients seems an ideal opportunity, where knowledge of both actual measured PV and GFR will assist in therapeutic decisions.

We report the results of a novel fluorescent dye technique, capable of repeatedly measuring PV easily and accurately. The technique correlates well with Nadler’s estimates of PV, but identifies outliers from the predicted value, and accurately tracks acute volume therapy. We believe this technology will prove valuable in the assessment and management of fluid therapy in hospitalized patients.

The study protocol and informed consent forms were approved by Western IRB in Puyallup, WA, USA. All patients received and completed an informed consent form prior to being enrolled in the study.

Bruce A. Molitoris is a co-founder, stockholder, and the Medical Director of FAST BioMedical. Erinn S. Reilly and Daniel Meier are employees of FAST BioMedical, and Ruben M. Sandoval is a paid consultant for FAST BioMedical. Patrick T. Murray, Erin Barreto, and Andrew D. Shaw are Scientific Advisory Board members, who have received compensation for consultancy. The investigators and authors did not receive any honoraria for their work on this study or authorship. FAST BioMedical, www.fastbiomedical.com, is a Delaware C-Corp. developing measured GFR and PV technology.

Funding was provided by an NIH SBIR Grant to D.M. and B.A.M. (R44 DK093274) and the National Center for Advancing Translational Sciences of the NIH under award UL1TR001417 to D.V.R. The product studied is under investigation and not for use at this time. NIH had no role in the design of the study, nor in the collection, analysis, or interpretation of the data or in manuscript preparation.

B.A.M., P.T.M., D.M., E.S.R., E.B., R.M.S., D.V.R., A.D.S. made substantial contributions to conception and design, or acquisition, analysis and or interpretation of data. B.A.M., A.D.S., A.G.G., D.M., E.S.R., D.V.R., and W.F.P. were involved in writing and or revising the manuscript. All authors gave final approval to the manuscript.

This study is registered at ClinicalTrials.gov (NCT03095391). Registration date: March 29, 2017.