INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: capillary perfusion; intensive care; microsphere; reperfusion injury; side effect

For many years, the contamination of parenteral fluids and drugs by particulate matter (i.e., glass, synthetic polymers, different fibers, starch, talc, insect parts, etc.) has been recognized as a potential health hazard and associated with diverse adverse reactions, ranging from clinically occult pulmonary granulomas detected at routine autopsy examination (1) to local tissue infarction (6), severe pulmonary dysfunction, and death (9). Because of a lack of standardized methods for the detection of particulate matter contaminants (12, 13) and rather vague legal regulations (14, 15), particulate matter contamination frequently exceeds the legal limits (14, 16). Manufacturing standards as well as quality controls to limit particulate matter contamination vary greatly between different countries and pharmaceutical companies (17) and may be entirely unpredictable in an increasing burden of counterfeit drugs distributed in developing countries (22) and over the Internet. The possible deleterious effects of such contaminants have become all the more clinically relevant, as generic and counterfeit products are being increasingly used because of economic pressures on health resources. The legal framework on particulate matter contaminants has been vaguely limited by the United States Pharmacopeia (23) to less than 25 particles/ml larger than 10 µm. Guidelines as to the assessment of such contaminants do not exist, the most widely applied "golden" standard still being the observation of the glass vial against a "bright light source" (23).

Despite the unanimous recognition of particulate matter contamination as a potential hazard, definitive proof from controlled clinical studies cannot be obtained for obvious ethical reasons (24). The presumed pathomechanisms include the mechanical blockage of small-caliber arterioles and capillaries by large particles, exceeding in size the inner lumenal diameter of the microvessels, the activation of platelets and/or neutrophils with the subsequent generation of occlusive microthrombi, and indirect effects on vasomotor activity (reviewed in References 15, 24, and 25). Several animal models have been tested to simulate the clinical situation of particulate matter contamination, but all have failed to cause significant functional or morphological injuries beyond the formation of foreign body granulomas in the lungs and other organs (1, 5, 26, 27). In all these models, particles have been injected into healthy animals. In agreement with these observations, we could not elicit any deleterious effects on the nutritional capillary perfusion of striated muscle, when we injected particles from different parenteral antibiotic solutions intravenously into healthy hamsters (see below).

However, intravenous fluids and drugs are usually not administered to healthy subjects, but rather to patients who may have experienced trauma (or polytrauma), have undergone major surgery, or suffer from other conditions that adversely affect the microvascular blood supply of tissues and vital organs (i.e., shock, acute respiratory distress syndrome, sepsis, multiple organ failure). For this reason, we investigated in the present study whether particulate matter contaminants from different parenteral antibiotic solutions might adversely affect the nutritional blood supply in a microvascular bed that has been compromised by prior injuryinduced experimentally by exposure of muscle tissue to ischemia and reperfusion. The experimental design was such that a "gold standard" cefotaxime, Claforan, was compared in a blinded fashion with two generic cefotaxime products, proven to contain particulate matter contaminants.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Determination of Particle Contamination in Antibiotic Vials

Two separate techniques were used to detect particle contaminants: for small particles an HIAC/ROYCO light blockade particle counter for liquids was used (Pacific Scientific, Silver Springs, MD), calibrated with defined polystyrene spheres (Polysciences, Eppelheim, Germany), and coupled with a sensor operating on the light blockade principle. The equivalent spherical diameter of any detected particle was assessed as a mean particle count of three separate vials per batch, dissolved in demineralized, particle-free water. Larger particles were detected by a filter method, using membrane filters with a 19.6-mm i.d. filtration funnel. The filters were loaded with the solutes of five vials of the respective antibiotic batches, using a vacuum pump. Video prints were prepared with a dissection microscope (Stemi SV6; Zeiss, Oberkochen, Germany) at ×30 magnification and the number and longest linear diameters were assessed with an interactive image analysis system (IBAS; Kontron, Munich, Germany).

Animal Model

For intravital fluorescence microscopy, we used the dorsal skinfold chamber preparation in awake Syrian golden hamsters. This model permits the microscopic quantification of nutritional capillary perfusion in finely striated skin muscle (28). The experimental preparation used in this study is similar, except for minor modifications, to that described previously in detail (30). Using epi-illumination and a ×20 water immersion objective 10 sites of interest were selected per chamber, each containing several parallel capillaries in the foreground and one characteristic draining venule for orientation purposes (Figure 1). For visualization of the plasma column, fluorescein isothiocyanate-conjugated macromolecular dextran (FITC-dextran, M_r 150,000, 5 mg in 100 µl of saline; Sigma, St. Louis, MO) was administered intravenously immediately before the microscopy studies. A 4-h ischemia was induced by gently pressing the muscle against the coverslip with a silicone pad and an adjustable screw, just sufficient to empty the blood vessels (31). Subsequently, we intravenously injected 100 µl of a suspension of normal saline and the filtrates of three different 1-g vial of cefotaxime sodium antibiotics (vial A, Claforan [Hoechst, Frankfurt am Main, Germany], batch D607N22; vial B, Cefantral [Lupin, Mumbai, India], batch CTC002; vial C, Taxim [Alkem, Mumbai, India], batch 3045). For that purpose, the lyophilized antibiotics were dissolved by injection of 10 ml of sterile saline (0.9%; Braun, Melsungen, Germany) into the original vials. The solutions were filtered (Supor 200 membrane filter, 0.2-µm pore size; Gelman Sciences, Ann Arbor, MI) and the residual on the filters was resuspended in 10 ml of saline. After centrifugation for 15 min, the pellets were resuspended in 100 µl of normal saline that was subsequently injected intravenously as a bolus into the hamsters. The identical 10 microvascular sites of interest were visualized and recorded 10 min after particle injection. The experiments were performed in a blinded fashion in n = 9 hamsters per group and in n = 5 control hamsters. In another series of n = 3 hamsters (each group), the particle filtrates were injected into hamsters in which the muscle tissue had not been exposed to ischemia-reperfusion injury, and functional capillary density (FCD) was recorded in 10 sites of interest before and 10 min after particle injection. The antibiotic drugs were not injected along with the particles because of the potential confounding effect of the antibiotics in the experimental equation. We avoided the need to perform additional experiments controlling for potential differences between the three drugs per se by separating the antibiotics from the particles. In yet another series of n = 9 hamsters, we injected polystyrene microspheres (Polysciences, Eppelheim, Germany) with increasing diameters (1.5-20 µm in diameter) and quantified FCD in the muscle tissue at baseline and 10 min after each microsphere injection. The same experiments were also performed in nine hamsters in which the muscle tissue had been exposed to 4 h of ischemia followed by 2 h of reperfusion.

View larger version (167K): [in this window] [in a new window]   Figure 1.   Top: Representative image of fluorescently stained capillaries running parallel within striated hamster skin muscle contained within a skinfold chamber. Contrast enhancement was achieved by intravenous injection of FITC-labeled macromolecular dextran (FITC-dextran, Mr 150,000). Magnification bar: 50 µm. Bottom: FCD is assessed by Photoshop image analysis (32). The Pencil tool is selected from the toolbox and the capillaries depicted on the screen are manually redrawn in a layer superimposed over the image, using a distinctive color. Because the redrawn pencil line is automatically shown on the computer screen, it can be controlled at every step and necessary corrective measures can be taken. The color line is then selected with the Magic Wand tool and the Select Similar command, and the area covered by the pencil line is quantified by using the Histogram command in the Image menu. On the basis of the resolution and magnification of the image and the predefined width of the pencil line, the length of the pencil line is calculated and the length of the capillaries is expressed relative to the area covered by the image (given in cm/cm2).

Image Analysis

Images were imported at a resolution of 72 dots per inch (dpi), using the S-VHS port of a Macintosh computer (Power Macintosh G3) equipped with an inbuilt graphic capture board. Functional capillary density was quantified by one investigator (J.B.) using Photoshop-based image analysis (Photoshop software, version 5.0; Adobe, San Jose, CA) as previously described in detail (32). In contrast to the density of all fluorescently stained capillaries, the parameter FCD assesses only those capillaries that are perfused with red blood cells (30, 33). The technique of quantifying functional capillary density with this image analysis software yields reproducible results, with low interobserver variability (32).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Particulate Matter Contamination

The results for particle contamination of the three different antibiotic formulations are shown in Table 1. The number of small particles (2-10 µm in diameter) was 30 times higher in antibiotics B and C, as compared with antibiotic A. For larger particles (25-100 µm in diameter) the burden was 12 and 28 times higher in antibiotics B and C, respectively, as compared with antibiotic A. Representative video prints of membrane filters of the three different antibiotic solutions are shown in Figure 2.

View larger version (108K): [in this window] [in a new window]   Figure 2.   Video prints (×30) of three representative membrane filters (4-mm i.d. filtration funnel, corresponding to a 12.6-mm2 effective filtration area) loaded with the residuals contained in five 1-g vials of the antibiotics Claforan (A), Cefantral (B), or Taxim (C ). Magnification bar: 333 µm. Note the size and sharp configuration of some of the particle contaminants found in the antibiotic preparations intended for intravenous injection.

Functional Capillary Density in Nonischemic Muscle Tissue

Intravenous injection of normal saline or particle suspensions from any of the three antibiotic solutions did not reduce FCD in nonischemic muscle (Figure 3A).

View larger version (27K): [in this window] [in a new window]   Figure 3.   Functional capillary density (FCD) in striated muscle after injection of saline or particle filtrates from 1-g vials of different cefotaxime solutions. (A) No loss of functional capillary perfusion after filtrate injections was observed in normal muscle tissue. FCD is expressed as a percentage of baseline values in each observation chamber. Data for all three antibiotic filtrates are not significantly different from data in saline-injected hamsters (n = 3 hamsters per group, Wilcoxon test). (B) To assess the effect of the antibiotic filtrates on postischemic muscle tissue, FCD was measured in each animal after 4 h of ischemia and 2 h of reperfusion (postischemic value) and then again after injection of saline or the antibiotic filtrates. The data presented are FCD values after saline or filtrate injections, expressed as a percentage of the postreperfusion FCD value. Please note that injection of filtrates from antibiotics B and C resulted in significantly reduced capillary perfusion compared with saline-treated control hamsters. No such adverse effect was seen after injection of filtrates from antibiotic A. (Data represent means ± SEM of n = 9 hamsters per group, *p < 0.05, **p< 0.01 versus data from saline-treated hamsters, Wilcoxon signed rank test.)

Functional Capillary Density in Ischemic-reperfused Muscle Tissue

As a consequence of 4 h of pressure-induced ischemia and 2 h of reperfusion, we observed a reduction of FCD by approximately 30%. The loss of FCD was similar in the control group and the three treatment groups and no significant differences were observed. When the same sites of interest were observed 10 min after injection of saline, we saw a further reduction of FCD of another 14% (Figure 3B, open column). Similar changes were observed after injection of particle suspensions of antibiotic A (Figure 3B, light gray column). In contrast, a marked loss of FCD of the postischemic muscle tissue was observed after injection of particle suspensions from antibiotics B and C (Figure 3B, dark gray and solid columns); FCD was reduced by another 28 + 14% (SD) and 26 + 10%, respectively, versus the postischemic values before particle injection. These differences were significantly different from the changes observed after injection of antibiotic A (Figure 3B).

In agreement with previous studies (28, 30), we found that FCD showed a marked heterogeneity in different sites of interest within each muscle tissue under baseline conditions, and even more so after ischemia and reperfusion, with some sites showing a more pronounced loss of capillary perfusion than others. We had expected that the particle-induced loss of FCD should have been most pronounced in those sites of interest that had undergone a more severe capillary loss during ischemia and reperfusion. When the particle effects were analyzed separately for those microvascular sites of interest that had suffered from a mild (less than 30% of baseline) versus a severe (more than 30%) loss of FCD, we found that particles from antibiotic B had a more pronounced effect in the more extensively compromised capillary bed, whereas the adverse effect of particles from antibiotic C were just as pronounced in the severely as in the less severely compromised capillary beds (Figure 4). Particulate matter from antibiotic A resulted in a comparable loss of FCD as the saline control.

View larger version (35K): [in this window] [in a new window]   Figure 4.   FCD in striated muscle after ischemia and reperfusion. In each hamster, 10 different sites of interest were studied at baseline, after 4 h of ischemia and 2 h of reperfusion (postischemic value), and after injection of the antibiotic filtrates. Data were separately calculated for those sites of interest with mild ischemia-induced loss of FCD (< 30% postischemic loss of FCD when compared with baseline values; A) and those with severe ischemia-induced loss of FCD (> 30% postischemic loss of FCD when compared with baseline values; B). Left graphs: FCD as a percentage of baseline values. Right graphs: FCD as a percentage of the postischemic values in the respective groups. Particulate matter from antibiotic C resulted in a significant loss of FCD irrespective of the severity of microvascular compromise (A and B), whereas particles from antibiotic B resulted in a loss of FCD only in severely compromised muscle tissue (B). The filtrate of antibiotic A was not different from the saline control in either mildly or severely compromised microvascular beds. Data represent means ± SEM of n = 9 hamsters per group. *p < 0.05, **p < 0.01, Wilcoxon signed rank test.

Histological Evaluation

Histological evaluation of the tissue contained within the observation chamber demonstrated birefringent particles lodging within arterioles and capillaries within or immediately adjacent to the muscle tissue (Figure 5). Beside these particles, Figure 5 also demonstratesin agreement with previous studies (29, 30)a dense infiltration of the muscle tissue by inflammatory cells, predominantly polymorphonuclear neutrophils and mononuclear cells, in response to the postischemic reperfusion injury (Figure 5, arrows). Birefringent particles were observed in filtrate groups B and C, but not in filtrate groups A or the saline control.

View larger version (126K): [in this window] [in a new window]   Figure 5.   Histological examination of muscle tissue within the observation chamber reveals birefringent foreign material lodged within microvessels immediately adjacent to striated muscle tissue (from antibiotic C). Note the infiltration of tissue by neutrophils as a response to prior ischemia and reperfusion (29, 30). Magnification bar: 100 µm.

Injection of Polystyrene Beads

Injection of polystyrene beads of defined diameters resulted in a gradual loss of FCD in nonischemic muscle tissue that reached statistical significance only after injection of 10-µm or larger microspheres (Figure 6). Even the injection of 20-µm-diameter microspheres resulted in a loss of FCD of less than 30% of baseline values in nonischemic muscle tissue (Figure 6). The microsphere-induced loss of FCD was significantly more pronounced in postischemic muscle tissue than in nonischemic muscle tissue (Figure 6, gray and solid versus open circles). However, no differences were seen when the microsphere-induced capillary loss was compared in sites with mild (less than 30%; gray circles) or severe (more than 30%; solid circles) microvascular loss due to prior ischemia and reperfusion (Figure 6).

View larger version (32K): [in this window] [in a new window]   Figure 6.   FCD before and after injection of polystyrene beads of defined sizes into hamsters. Baseline values represent microcirculatory tissue perfusion before injection of beads. In postischemic muscle tissue, endothelial cell swelling had resulted in a reduction of FCD before the injection of beads was started. All data are expressed as a percentage of the FCD values before injection of the beads (100%). Shown are mean values ± SEM of n = 9 animals per group. Open circles: injection of beads into nonischemic muscle tissue. A significant drop in FCD occurs only with beads 10 µm or more in diameter (#p < 0.05 versus baseline, Wilcoxon signed rank test). When injected into mildly (< 30% postischemic loss of FCD before bead injection; gray circles) and severely (> 30% loss of FCD before bead injection; solid circles) compromised postischemic muscle tissue, even the smallest beads (1.5 µm) result in a significant drop in FCD from the baseline value (p < 0.05 for every bead size, not specifically indicated in the graph). For beads larger than 10 µm, the loss of FCD in postischemic tissue was significantly more pronounced than in nonischemic tissue (*p < 0.05, Wilcoxon signed rank test).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The pivotal finding in our present study is the demonstration that particulate matter contaminants from different parenteral antibiotic solutions jeopardizes nutritional capillary perfusion in a microvascular bed compromised by prior ischemia and reperfusion (Figures 3 and 4), but not in a physiologically perfused microvascular bed. These data provide the first experimental in vivo evidence that such contaminants in intravenous drugs and solutions may indeed cause adverse effects in patientsin particular in critically ill patients.

In 1949, von Glahn and Hall observed an association between the injection of cotton fibers in intravenous solutions and the development of pulmonary granulomas in six patients (1). A few years later, Jacques and Mariscal found that the number of cotton fiber granulomas in the lungs was linked to the amount of fluid administered parenterally to the patients in the days before death (2). Similar observations were later reported by Bruening in a study of 150 children who had received various amounts of parenteral fluids (3) and by Sarrut and Nezelof, who described 25 cases of pulmonary arterial lesions in autopsy specimens from premature infants who had received intravenous injections of large volumes of fluids (4). In all these cases, the granulomas had resulted in no apparent clinical manifestations.

Several reports have described harmful consequences in intravenous drug users due to particle contaminants, ranging from ophthalmologic complications (35, 36), tissue infarction (37, 38), and severe pulmonary distress (39) to lethal acute congestive heart failure (42, 43). Finally, several reports have pointed toward pulmonary complications due to particulate matter contaminants in cardioplegic solutions (44, 45). Of particular interest in the study by Walpot and colleagues was the observation that most microthrombi were associated with particles less than 2 µm in diameter (glass, latex, and polymers), which constitute the bulk of particulate matter contaminants in intravenous fluids (46). Drip phlebitis, a condition recognized to be due to particulate matter contaminants in intravenous fluids, was almost completely prevented by the use of in-line filters with pore sizes of 0.2 µm (47), 0.22 µm (48), 0.4 µm (49), or 0.45 µm (50). Likewise, a significant (70%) drop in coronary blood flow in rats after infusion of unfiltered cardioplegia solution was almost entirely prevented by 0.8-µm in-line filters, and this resulted in a significant improvement of postischemic recovery (51).

In the past, several animal models have been tested to demonstrate clinically significant adverse effects of particulate matter contaminants. In 1949, von Glahn injected cotton fibers intravenously into rabbits and observed the formation of pulmonary granulomas around fibers lodged within the pulmonary microcirculation (1). Wartmann and coworkers elicited pulmonary granulomas in rabbits by intravenous injection of fibers from filter paper (26) and Stehbens and Florey injected rabbits intravenously with particles between 0.2 and 0.5 µm in diameter and found occlusive microthrombi associated with platelets and neutrophils (27). Garvan and Gunner demonstrated that foreign body-type granulomas could be produced experimentally in the lungs of rabbits by administration of intravenous fluids that appeared to be particle free in daylight but that contained numerous microscopic particles (5). Nevertheless, all these studies were performed on healthy animals and none of these studies resulted in any pathologically relevant deleterious effects in the animals.

In our present study, we observed that particulate matter contaminations adversely affected the postischemic, not the normal, microcirculation (Figures 3 and 4). Because ischemia- reperfusion injury shares many microcirculatory manifestations with sepsis-induced microcirculatory compromise, we suggest that the postischemic microcirculation may representat least in parta relevant in vivo model for the intensive care patient in the process of developing multiple organ dysfunction syndrome (52, 53). By contrast, no loss of capillary perfusion was observed after injection of the same particles into hamsters with a normal muscle microcirculation (Figure 3). Histological studies demonstrated the presence of birefringent particles lodged in the arterioles that provide blood flow to the muscle (Figure 5). Moreover, we found no evidence of platelet or leukocyte aggregates within these vessels, suggesting a mechanical blockade rather than the stimulation of microthrombi as a cause for the loss of FCD. Whether indirect effects of the infused particles on arteriolar vasomotor activity may have contributed to the loss of capillary perfusion cannot be demonstrated because we focused on nutritional capillary perfusion and did not study precapillary vessel segments. At the capillary level, microvessel diameters are defined not by changes in vasomotor activity, but by the total vascular resistance of the microvascular network as a whole. In a study by Menger and coworkers of the same animal model in hamsters, capillary diameters were measured at 10 different sites within the striated muscle before and after ischemia and reperfusion (54). These authors found an inverse correlation between microvessel diameters and FCD, suggesting that capillaries dilate in response to the increased blood flow that now must pass through a reduced number of capillaries and/or that shunt vessels with larger diameters are preferentially perfused in compromised tissue beds (54). These findings help to explain that injection of polystyrene beads (Figure 6) or particles from antibiotic C (Figure 4) did not elicit a significantly more pronounced loss of FCD in mildly or severely compromised muscle tissue. These findings also suggest that the adverse effects of beads and/or particle contaminants on the compromised but not the normal microcirculation mayat least in partbe due to nonmechanical mechanisms such as increased adhesivity of the endothelial lining (reviewed in Reference 55), to the deposition of fibrinogen on the microvascular endothelium (56), or else to the activation of mononuclear cells by particulate matter as suggested by Dorris and coworkers (57) and more recently by Monn and Becker (58). Irrespective of the exact mechanism behind the observed phenomenon, the data in the present study suggest that particulate matter contaminants may adversely affect the nutritional blood supply that is already compromised by ischemia-reperfusion, a condition that is regularly encountered in clinical disorders such as trauma (or polytrauma), major surgery, sepsis, shock, and others. Although clinically irrelevant in an intact organism, under such conditions of compromised microvascular blood flow particle contaminants in intravenous fluids may tip the scale from recovery ad integrum to (multiple) organ failure and severe morbidity and mortality of the patients. The particle "dose" that we administered was from one vial of antibiotic solution. Even though this may seem considerable when injected into a hamster, which has about one-thousandth the weight of a human patient, the situation may be more comparable when taking into consideration that (1) that the microcirculation, particularly the size of corpuscular blood elements and the inner width of capillaries, are virtually identical in hamsters and humans, and (2) that patients in intensive care wards receive not only one but often literally hundreds of injections that are derived from such vials.

The data provided in the present study emphasize a dilemma with immediate clinical consequences. Drugs that are no longer protected by patent rights are free to be manufactured and distributed by any company and even though the drug may be the same, the formulation may differ and the overall quality and purity of the product may not be the same. Differences in drug quality may originate from the use of intermediates and solvents, manufacturing conditions (sterile filtration, sterile closed systems, etc.), reaction conditions, in-process controls, purification procedures, validation control procedures, and, finally, storage conditions. In times when medical care is increasingly governed by financial considerations, we hope that our present study raises the awareness for potential hazards induced by intravenous drugs of inferior quality. The presented animal model may help to further define what extent of particulate matter contamination is clinically tolerable and what is not.