Recombinant tissue plasminogen activator (rt-PA) is a lytic medication widely used in the emergency department to treat acute thrombotic disorders such as ischemic stroke and myocardial infarction. It is known in the clinical use of this drug that it can be less effective in approximately 25% of individuals receiving such treatment. However, there are no data on the variation of lytic efficacy of rt-PA in decreasing individuals' clot size over time. In this study, in vitro lytic efficacy was determined by measuring the decrease in clot diameter after 30 minutes of drug exposure. The authors sought to explore whether there are individuals who do not respond to this lytic therapy and to estimate the rate of nonresponse.
Human whole blood clots were made from blood drawn from 22 adult volunteers. The only exclusion criterion was the use of aspirin within 72 hours of the blood draw. Blood clots were allowed to spontaneously form at room temperature and were then incubated at 37°C for 3 hours to ensure complete clot retraction. Sample clots from the same individuals were then exposed to human fresh-frozen plasma (hFFP) control or rt-PA in hFFP (rt-PA) at a concentration of 3.15 μg/mL. All clots were exposed at 37°C for 30 minutes, and clot diameter was measured as a function of time, using a microscopic imaging technique. The fractional clot loss (FCL), which is the percentage decrease in clot diameter at 30 minutes, was used as a measure of lytic efficacy.
Means with standard deviation (SD) FCL values were 8.6% (±3.0%) for control and 20.6% (±9.3%) for rt-PA–treated clots. The mean (±SD) difference in FCL values was 12.0% (±8.8%) and was significant (p < 0.05, paired t-test). Five of the 22 subjects (23%) were “rt-PA nonresponders,” in that their FCL (rt-PA) values fell within that of the FCL control values.
Overall, rt-PA does not produce clot lysis in vitro in clots from a substantial minority of the population, likely due to individual variations in clot composition and structure.
Eficacia Lítica Individual del Activador Tisular de Plasminógeno Recombinante en un Modelo de Coágulo Humano in Vitro: El porcentaje de “No Respuesta”
El activador tisular de plasminógeno recombinante (Recombinant tissue plasminogen activator o rt-PA) es una medicación lítica utilizada ampliamente en el servicio de urgencias (SU) para tratar enfermedades trombóticas agudas como el ictus isquémico o el infarto de miocardio. El uso clínico de este medicamento pone en evidencia que puede ser menos efectivo en aproximadamente un 25% de los individuos que lo reciben. Sin embargo, no hay datos sobre la variación de la eficacia lítica de rt-PA en la disminución del tamaño del coágulo en los individuos a lo largo del tiempo. Es este estudio, la eficacia lítica in vitro se determinó midiendo el descenso en el diámetro del coágulo tras 30 minutos de exposición al medicamento. Se exploró si hay individuos que no responden a esta terapia lítica, y se estimó el porcentaje de no respuesta.
Se hicieron coágulos sanguíneos humanos con sangre extraída de voluntarios tras la aprobación del protocolo de estudio por el comité ético de investigación. El único criterio de exclusión fue la toma de aspirina en las 72 horas previas a la extracción de sangre. Los coágulos de sangre de los 22 adultos voluntarios se formaron espontáneamente a temperatura ambiente y después se incubaron a 37°C durante 3 horas para asegurar la retracción completa del coágulo. Los coágulos de la muestra de un mismo individuo se expusieron posteriormente a plasma fresco humano congelado (hFFP) control, o a rt-PA en hFFP a la concentración de 3,15 μg/ml. Todos los coágulos se expusieron a 37°C durante 30 minutos. El diámetro del coágulo se midió en función del tiempo usando técnicas de imagen microscópicas. La fracción de coágulo perdida (fractional clot loss (FCL)), que es el porcentaje que disminuye el diámetro del coágulo en 30 minutos, se usó como medida de la eficacia lítica.
Las medias con desviación estándar (DE) de los valores de FCL fueron 8,6% (DE ±3,0%) para el grupo control y 20,6% (DE ±9,3%) para los coágulos tratados con rt-PA. La diferencia media en los valores de FCL fue del 12,0% (DE ±8,8%) y fue significativa (p < 0,05, t-test para datos apareados). Cinco de los 22 sujetos (23%), fueron “no respondedores a rt-PA,” ya que sus valores FCL (rt-PA) se encontraron dentro de sus valores de FCL control.
En general, el rt-PA no funciona para una minoría considerable de la población, probablemente debido a variaciones individuales en la composición y estructura del coágulo.
Tissue plasminogen activator (t-PA) is a 65-kDa protein that is representative of the class of enzymes known as plasminogen activators. Its mechanism of action is to convert the substrate plasminogen into plasmin in the in vivo setting of a thrombus. The plasmin then cleaves the fibrin mesh of the clot, yielding fibrin degradation products and lysing the clot. t-PA is manufactured recombinantly for clinical use and is denoted recombinant t-PA or rt-PA (alteplase).
Thrombotic disorders such as myocardial infarction (MI), ischemic stroke, thrombosis of dialysis grafts and central lines, and pulmonary embolus can all be treated with rt-PA. Although treatment is generally effective, there remain some individuals for whom there is limited efficacy. For example, in the thrombolytic treatment of acute MI, 25% of those treated did not achieve a thrombosis in myocardial infarction (TIMI) 3 flow within 90 minutes; achieving a TIMI 3 flow is associated with good clinical outcome. Similarly, when used to treat occluded central access catheters, 26% of patients treated with rt-PA did not have flow restored after 2 hours of therapy. When rt-PA has been used to treat acute ischemic stroke in patients suffering from a middle cerebral artery occlusion, 20% failed to show even partial recanalization.
There are many biochemical mechanisms that can contribute to a lack of rt-PA lytic efficacy in a given individual. Relative levels of α2-antiplasmin and plasminogen activator inhibitor 1 (PAI-1) have been shown to be weakly correlated with in vitro lytic resistance in a microtiter plate fibrin clot model. Substrate depletion due to lack of plasminogen in the clot can limit clot lysis.[11, 12] Other mechanisms, such as variation in the apolipoprotein phenotype, have been found to influence rt-PA–induced in vitro clot lysis in a recent study. In addition, the detailed clot composition such as relative amounts of platelets, fibrin, and red blood cells may contribute to lytic resistance.
Overall, there are many potential mechanisms that can contribute to lytic resistance. The clinical and mechanistic information suggest up to a quarter of individuals are nonresponsive to treatment with rt-PA, yet variation in the lytic efficacy of rt-PA between individuals, as determined by measuring clot size over time, has not been established. Decreasing clot size is the clinical goal of any lytic therapy.
In this study, the lytic efficacy of rt-PA was measured in human whole blood clots drawn from volunteers. Lytic efficacy was determined by measuring the total decrease in clot diameter after 30 minutes of drug exposure, using a microscopic imaging technique. In addition, as a control, clot lysis was measured in pooled human fresh-frozen plasma (hFFP) to eliminate plasma variability of fibrinolytic proteins[14, 15] and isolate the experimental clot response to rt-PA. Overall, we sought to explore whether there are individuals who do not respond to lytic therapy and to estimate the rate of nonresponse.
This was a laboratory study. Institutional review board approval was granted and written informed consent was obtained from the volunteer blood donors.
Study Setting and Population
Human whole blood was obtained by sterile venipuncture from 22 healthy volunteers recruited from emergency department staff. The only exclusion criterion was the use of aspirin within 72 hours of the blood draw. The mean (±SD) age of these volunteers was 36 (±7 years), ranging from 20 to 51 years of age. There were 17 men and five women; two were African American, one was Asian, and 19 were white.
Preparation of rt-PA and Human Plasma
The rt-PA was obtained from the manufacturer (Activase, Genentech, San Francisco, CA) as a lyophilized powder. Each vial was mixed with sterile water to a concentration of 1 mg/mL as per the manufacturer's instructions, aliquoted into 1.0-mL centrifuge tubes (Model 05-408-13, Fisher Scientific Research, Pittsburgh, PA), and stored at –80°C. The enzymatic activity of rt-PA is stable up to 7 years when stored in this fashion.[16, 17]
Human fresh-frozen plasma was procured from a blood bank in 250- to 300-mL units. Six units were thawed and mixed to ensure a representative average sample of hFFP for these experiments; this is similar to the approach used by others. The mixed units were then aliquoted into 50-mL polypropylene centrifuge tubes (Model 05-538-68, Fisher Scientific Research) and stored at –80°C. Aliquots of rt-PA and plasma were allowed to thaw for experiments; any excess was discarded following completion of each experiment.
Production of Blood Clots
Fresh whole blood samples of 1 to 2 mL were placed in sterile glass tubes (3 mL, VWR, Batavia, IL) and allowed to form clots in and around a small diameter (~600 μm) micropipette (Becton, Dickinson and Company, Franklin Lakes, NJ; 20λ) through which a segment of 7-0 silk suture (Ethicon Industries, Cornelia, GA) had been threaded. This is similar to clot production methods used in imaging studies by Winter et al. and Yu et al. The glass tubes containing the clots were placed in a 5-by-16 test tube rack, incubated for 3 hours at 37°C, and refrigerated at 5°C for 3 days ensuring maximal clot retraction, lytic resistance, and stability.[22-24] The refrigeration ensures preservation of the sample clots and likely preserves function of the coagulation factors during subsequent experiments at 37°C.[25, 26]
Before each experiment, the micropipette was removed to produce a cylindrical clot adherent to the suture. The clot was typically 5 to 8 μL in volume and approximately 300 μm in width. At this size the clots were similar in diameter to the intracerebral segments of the middle cerebral arteries (80 μm to 840 μm in diameter) or other cerebral vessels such as the recurrent artery of Heubner and its perforators (mean ± SD, 643 ± 237 μm in diameter).[27, 28]
Visualization of the Clot
For each experiment, the clot attached to the suture was placed in a clean micropipette (Drummond Scientific Company, Broomall, PA; 200 μL) and inserted into a U-shaped sample holder composed of hollow luer lock connectors and silicone tubing (Cole Parmer, Vernon Hills, IL; outer diameter 3.175 mm). The sample holder was placed in an acrylic water tank with a microscope, as shown in Figure 1.
Water in the tank was maintained at a temperature of 37°C (SD ± 1°C) during all experiments using a heating element (Hagen, Mansfield, MA; 50 W). The tank was placed over the objective of an inverting microscope (Olympus, Melville, NY) to visualize the clot. The field of view in the image was 340 × 260 μm (640 × 480 pixels). The entire apparatus was placed on top of a vibration isolation table (Newport, Irvine, CA; XL-G) for mechanical stability. Images were recorded at six frames/minute using a charge-coupled device (CCD) camera (Hitachi, Woodbury, NY; KP-M1A), and data were stored for later analysis on a computer. A complete description of the imaging apparatus is available elsewhere.[29-31]
Determination of Lytic Efficacy
Light intensity transmitted through the sample clot is reduced with increased clot thickness or clot density. The CCD camera records image light intensity I(x,z) at each pixel (x,z). By analyzing the light intensity in each pixel, the clot edges can be identified, thus enabling measurement of clot width.
An image from the CCD camera was stored on a desktop computer for each frame as a function of time. The average clot width (CW) was then calculated, using a computer program written in Matlab 6.5 R13 (Mathworks, Inc., Natwick, MA). First, the spatial gradient of the light intensities ∂I(x,z)/∂x was calculated for each row (fixed z) of pixels. The positions of the two clot–plasma interfaces were determined via an edge-detection routine that finds the values of x = (x1, x2) for each z such that ∂I/∂x(x1,2, z) is greater than or equal to Γ, where Γ is a constant. A Γ of 2.5 is sufficient to detect well-defined clot edges. The width W(z) of the clot at each z was then equal to |x1 – x2|. The mean CW for each image was calculated by averaging the width over all z values. The clot width data were subsequently corrected for the finite suture size used in these experiments. This yields the expression
where CWCorrected is the value of the total clot width adjusted for the finite suture width SW. The average suture width for these experiments was found to be 95 ± 15 μm, based on 252 samples of the 7-0 suture. The data were then normalized to the initial value of CWCorrected during the first minute (six frames; ), yielding the expression
Note that CWNORM(t) is equal to 1 at time t equal to zero, and becomes equal to zero if the clot is totally lysed, thus leaving only the suture contributing to the sample width. This parameter is averaged for all clots in a given treatment group. The fractional clot loss (FCL, expressed as%) is then defined as
where CWNORM(30 min) is the average normalized clot width over the final minute of treatment, as determined from Eq. . Note that a given sample clot is imaged six times per minute; therefore, the clot width over the final minute is averaged over six measurements.
Clots were exposed to either hFFP alone (control) or hFFP plus rt-PA (rt-PA) at a concentration of 3.15 μg/mL. To standardize treatment group assignment for a given clot, clots taken from even rows of the test tube rack were exposed to rt-PA, and clots from odd rows to hFFP. In all experiments, clots were taken sequentially from one end of the test tube rack to the other. The clots for each individual were either exposed to control or to rt-PA, and all individuals had clots exposed to both treatments. The FCL control value for a given subject was determined by averaging the measured FCL values from each clot from that subject exposed to hFFP alone. Similarly, the FCL rt-PA value for an individual was determined by averaging the FCL values for each clot from that subject that was exposed to rt-PA in hFFP. The median number of clots used in the control group for a given subject was 6 (interquartile range [IQR] = 5 to 11) and 5 (IQR = 3 to 8) for the rt-PA treatment group. A total of 354 clots were used in this study. The rt-PA concentration used here is comparable to those used in humans.[2, 30, 32]
Individual trials began by slowly injecting 1 mL of hFFP (control), or 1 mL of hFFP containing rt-PA, into the sample holder. At time T equal to zero, the solution was in contact with the clot. Removing the syringe exposed the ends of the sample holder to atmospheric pressure, and the clot surface to a static fluid column. Clots were exposed to a specific treatment regimen for 30 minutes; previous studies have shown that the majority of thrombolysis occurs within a 30-minute time period.[33, 34]
Mean values, standard deviations (SDs), and other statistical parameters were calculated using SPSS Statistics 18 software (IBM, Somers, NY). Student's t-test (paired) was used to compare between the two treatment groups, and a p-value of less than 0.05 was considered significant. The 95% confidence intervals (CIs) were also calculated for the FCL values. Lilliefor's test was used to test the normality of the data. Overall, we sought to explore whether there are individuals who do not respond to lytic therapy and to estimate the rate of nonresponse.
Figure 2 shows a typical clot exposed to rt-PA at a concentration of 3.15 μg/mL. The photomicrograph on the left shows the sample clot prior to any substantial lysis; the one on the right exhibits the clot after 30 minutes of clot lysis. Note that the clot diameter has decreased substantially as a result of rt-PA exposure.
Histograms for the FCL control and FCL rt-PA values are in Figure 3. The mean (±SD) FCL control group was 8.6% (±3%), and the range was 3.4% to 14%. The median was 8.0% (IQR = 6.5% to 11.4%) in control-treated clots. As a measure of within-individual variability, the coefficient of variation (CV) was calculated for each subject. For the control group, the mean CV was 0.7. Lilliefor's test was not significant for either the rt-PA or control groups; therefore, the data are reasonably described by a normal distribution. The mean ± 2 SD represents the lysis expected in the control condition for 95% of all clots. In this case, 95% of FCL values then should lie within the range of 2.6% to 14.6%. We defined an rt-PA nonresponder as an individual whose FCL rt-PA lies within this range.
For rt-PA–treated clots, mean (±SD) FCL was 20.6% (±9.3%) and ranged from –1.4% to 39.2%. The median was 20.3% (IQR = 14.2% to 29.0%). The mean CV for rt-PA–treated clots was 1.24. Overall, the mean (±SD) difference in FCL between the control and rt-PA–treated groups was 12.0% (±8.8%). The difference was significant (95% CI of the difference = 8.1% to 15.9%, p < 0.001). There were five individuals (23%) with FCL rt-PA falling within the control range.
These data show that the mean (±SD) FCL in human whole blood clots was 20.6% (±9.3%) when treated with rt-PA. This was significantly larger than the mean (±SD) FCL value of 8.6% (±3%) in control-treated clots. In addition, five of the 22 individuals in this study could be classified as rt-PA nonresponders; i.e., these individuals' FCL rt-PA fell within the range of FCL control values expected in this small population. One can then estimate the nonresponse rate as 23%, which corresponds well with that found in clinical studies of rt-PA.[4, 7]
Others have studied the lytic efficacy of rt-PA in both clinical and some in vitro models. Colucci et al. measured in vitro clot lysis using a radioactive fibrin assay in anticoagulated human clots. They found an average clot lysis of 46% after 3 hours of exposure to rt-PA at a concentration of 0.5 mg/mL. In addition, they found clot lysis of less than 5% in their model for the control-treated clots. The smallest amount of lysis they observed in the rt-PA–treated clots was 25%. Although this was not a primary aim of their study, this implies that all of their clots lysed to a greater degree than the control in response to rt-PA, implying a “nonresponder” rate of 0%. In addition, they found a weak correlation of whole blood clot lysis from rt-PA at a concentration of 0.5 μg/mL and the concentrations of serum PAI-1 and plasminogen. Trusen et al. measured lysis in human adult and newborn whole blood clots as a function of rt-PA concentration and time of lytic exposure. Clot lysis was determined by measuring the percentage of mass loss of clots before and after lytic exposure. They found clot lysis of only a few percent in their control-treated clots after 30 minutes of exposure. The rt-PA–treated clots exhibited about 10% lysis at an rt-PA concentration of 3 μg/mL, close to the concentration value used in this work.
The differences with the result presented here likely arise from the different clot models used in other work. In both studies discussed above,[35, 36] blood was drawn from volunteers and initially anticoagulated with sodium citrate. Clots were formed by then adding thrombin and calcium chloride. This clot preparation method is substantially different than that used here in which the clots are allowed to spontaneously form and retract without anticoagulation. Several studies have shown that the method of clot manufacture can significantly affect lysis results.[12, 37] For example, Niessen et al. found that thrombin-induced clots (TC) were substantially more resistant to rt-PA lysis than those allowed to form spontaneously (SC) in an in vivo rat stroke model. In their work, rat middle cerebral arteries were occluded using either SC or TC clots. A sample clot was introduced into the rat middle cerebral artery, followed by rt-PA administered at a dose of 10 mg/kg to induce in vivo clot lysis. Reperfusion was determined by magnetic resonance imaging, and the mean reperfusion time in SC-treated rats was 1.6 hours, compared with 4 to 5 hours in TC-treated rats. It is clear that the clot formation method substantially influences rt-PA lytic efficacy. An advantage to the clot model used in the current work is that no exogenous materials were used for either anticoagulation or for clot formation.
In addition, clot composition may significantly affect the lytic efficacy of a given therapy. As most thrombolytics, including rt-PA, act on the fibrin mesh to induce clot lysis, varying levels of fibrin could be expected to affect the efficacy of fibrinolytic medications. Silvain et al. found that fibrin content increased with ischemia time in clots obtained by a coronary artery aspiration technique. Leibeskind et al. found that higher RBC clot content correlated with increased incidence of the hyperdense middle cerebral artery sign on CT and/or MRI. Such variations in clot morphology could be expected to alter clot susceptibility to fibrinolytic medications. One can conclude that the “lytic resistance” found in the current study is likely multifactorial, depending on an individual's plasma and/or clot concentration of thrombotic and/or fibrinolytic enzymes[40, 41] and detailed clot composition.[38, 39, 42] Potential solutions to increasing lytic efficacy may involve the use of more than one thrombolytic medication, such as combined rt-PA and GPIIb–IIIa. This therapy is in clinical use for the treatment of MI and is being tested in clinical trials for acute ischemic stroke. The use of direct-acting lytics such as plasmin could increase lytic efficacy as well, since such medications would not be dependent on the amount of plasminogen substrate that is available. The experimental technique described here has been used to examine the effects of various lytic therapies such as combined ultrasound and rt-PA and could be used to elucidate the effects of clot composition by altering the clot manufacturing protocol.
A primary limitation is that this study used in vitro clots. As a result, they are likely different in structure and composition from those forming in the clinical realm. Overall, these clots are more likely to be similar to those occurring in deep venous thrombosis (DVT). In addition, there is no fluid flow or pressure gradient used in these lysis experiments. Although this model can be considered somewhat similar to the in vivo scenario of a totally occluded vessel, the clots in these experiments are not exposed to a pulsatile pressure as would occur in clots exposed to blood pressure in an actual vessel. Also, pooled plasma is used in exposing these clots to control and rt-PA. Given that pooled plasma may be different than a given donor's plasma in levels of the various proteins involved in both thrombogenesis and thrombolysis, this could result in differences in clot response. In addition, the surface of these sample clots are entirely exposed to the hFFP; this differs from that in vivo situation in that the typical occlusive thrombus is exposed to plasma only on its trailing surface.
In addition, detailed medical histories of the subjects were not done. Given that the only exclusion was the use of aspirin within 72 hours of the blood draw, it is possible that some of the subjects had medical conditions or medications that could affect thrombogenesis and/or thrombolysis. However, it was felt that reducing the biologic variability in the subjects would reduce the relevance and generalizability of the study findings.
These experiments likely are underestimating the amount of clot lysis that would occur in clinical occlusive disease, as exposure to a pressure gradient would “push” the rt-PA into the clot due to permeation and enhanced diffusion. This is justified in that the current model is conservative and likely underestimates treatment efficacy.
Recombinant tissue plasminogen activator did not produce clot lysis in vitro in clots from five of 22 individuals (23%). This proportion is similar to the recombinant tissue plasminogen activator failure rate seen in clinical trials, suggesting that clot composition may account for a substantial fraction of such treatment failure.
The authors thank the members of the Greater Cincinnati/ Northern Kentucky Stroke Team for their invaluable support.