Disodium cromoglycate stabilizes mast cell degranulation while reducing the number of hemoglobin-induced microvascular leaks in rat mesentery

Maxwell I. Ginsburg, Ann L. Baldwin

This article has a correction. Please see:


Blood substitutes, such as diaspirin cross-linked Hb (DBBF-Hb), have been considered for use during blood transfusions. Unfortunately, bolus injection of modified Hb has been shown to rapidly increase the leakage of microvessels to plasma albumin. This effect may result from production of excess reactive oxygen species (ROS) and could be linked to the observed increase in degranulated mast cells (DMC). Disodium cromoglycate (cromolyn) stabilizes mast cells and therefore might minimize the venular permeability in the rat mesentery. In 10 anesthetized Sprague-Dawley rats, the mesenteric preparation was continuously suffused with cromolyn while the microvasculature was filled with DBBF-Hb solution (10 mg/ml) for 10 min. Six animals received cromolyn pretreatment [two intravascular injections over 30 min (experiment A)] and four animals received pretreatment with 2% HEPES-buffered saline (HBS)-BSA (experiment B). Two more animals were pretreated with HBS-BSA without DBBF-Hb infusion but with cromolyn suffusion (experiment C). Another set of experiments was performed on five animals without cromolyn suffusion or any pretreatment but with DBBF-Hb infusion (experiment D). All groups then received a 1-min perfusion of FITC-albumin, fixation for 60 min, and microscopic examination. Experiments A and B demonstrated a significant reduction in the number of venular leaks and DMC compared with experiment D, but not in the area of venular leaks. These results suggest mast cell degranulation is not a major contributor to microvascular leakage induced by DBBF-Hb.

  • blood substitutes
  • cromolyn
  • venular leaks
  • diaspirin cross-linked hemoglobin

in the last decade, the desire for a more efficient and convenient blood substitute for use in transfusions and major surgery has resulted in the development of synthetic oxygen transporters, such as diaspirin cross-linked hemoglobin (DBBF-Hb). Previous studies have indicated that DBBF-Hb alone is not a safe and effective blood substitute because it increases microvascular permeability and mast cell degranulation (5). Autoxidation of Hb is the primary oxidative reaction that initiates the oxidative cascade and is an important concern when considering the use of chemically modified blood substitutes (3, 15, 16, 20). D'Agnillo and Alayash (14) described how DBBF-Hb has been shown to autoxidize faster than native Hb. When autoxidized, Hb produces superoxide, which in turn reacts with Hb, oxidizing oxyhemoglobin to methemoglobin or reducing methemoglobin to oxyhemoglobin (27). The superoxide can also be spontaneously converted to hydrogen peroxide via superoxide dismutase (19). The hydrogen peroxide can then react with DBBF-Hb to produce ferryl hemoglobin (HbFe4+). HbFe+4 is a highly reactive oxygen species (ROS) that can induce lipid peroxidation and oxidize other biological molecules (1, 13). These physiological adverse effects are detrimental for several reasons. For example, ROS activate mast cell degranulation (21), which contributes to the vascular permeability by increasing the amount of released histamine (16, 28, 29). Mast cells are potent mediators that cause most of the symptoms observed during hypersensitivity reactions (22). With an increase in venular leakage, the blood substitute itself can leave the circulation and the patient may be deficient regarding oxygen transport. Also, the increase in leakage will alter the kinetics of delivered hormones and enzymes to appropriate tissues, all of which are critical, especially during hemorrhagic shock when a transfusion would be necessary (2, 9, 18). In other words, there is sufficient evidence to indicate that modified Hb can be unstable in vivo.

Previously, we have shown that injection of the mesenteric microcirculation with DBBF-Hb causes venular leakage to FITC-albumin and also produces mast cell degranulation. However, it is not clear whether leak formation is exacerbated by the mast cell degranulation or whether these two responses are independent of each other. This study addresses the following question: if the mast cells are stabilized, do the leaks still form by some mast cell-independent mechanism? No experiments have previously been performed to determine the role that mast cell degranulation plays in Hb-induced microvascular leakage.

It is well documented that disodium cromoglycate (cromolyn) works well as a mast cell stabilizer. Cromolyn has been widely used as a treatment for asthma, most likely because of its ability to prevent the release of the mediators of anaphylaxis from mast cells (24) and because no cromolyn-induced toxicity has been observed in studies using the prescribed therapeutic routes (10). Little information is known about the mechanism by which cromolyn works; nevertheless, previous studies, as well as our own, indicate that cromolyn does in fact reduce the number of degranulated mast cells (DMC) (10, 2224, 29). Crossman et al. (12) demonstrated that cromolyn may function by modifying tachykinin actions, thus reducing the effects of substance P. Substance P is a neuropeptide that can activate mast cells (25). This mechanism is further supported by the findings of Yamawaki et al. (29), who determined that cromolyn inhibits the response of microvascular leakage to substance P aerosols. Previous studies suggest that cromolyn acts as a tachykinin antagonist in the airways and reduces vascular permeability (25, 29).

For this reason, cromolyn was used in experiments on the rat mesentery to determine whether mast cell stabilization reduces the microvascular leakage produced by DBBF-Hb. As Baldwin et al. (9) and Valeski and Baldwin (25) have previously determined, the rat mesentery provides an excellent preparation to quantify the degree of vascular leakage invoked by particular mediators because the tissue is very thin and the microvascular networks are almost two-dimensional, thus facilitating image acquisition and analysis (57, 9, 26).


Preexperimental treatment of rats. The research described in this study adheres to the American Physiological Society's “Guiding Principles in the Care and Use of Animals,” and the protocol has been approved by Institutional Animal Care and Use Committee. The subjects of the experiment were male Sprague-Dawley rats from Harlan Teklad (Madison, WI) weighing between 440 and 485 g. Extreme care was taken in the ways in which the rats were housed and tended, as explained in Baldwin et al. (9).

Hb solution. DBBF-Hb was obtained from the Walter Reed Army Institute of Research (Washington, DC) as a generous gift. It is human Hb cross-linked by bis(3,5-dibromosalicyl)fumerate (DBBF-Hb). Its half-life is ∼3–24 h in the circulation.

Cannulation and injection of rat mesenteric circulation. Fourteen Sprague-Dawley rats were preanesthetized with 1 mg/kg body wt of the following mixture: ketamine hydrochloride (5 ml of 100 mg/ml), acepromazine maleate (2 ml of 10 mg/50 ml), and xylazine (8 ml of 20 mg/ml). After the preanesthetic, an intraperitoneal injection of pentabarbitone sodium (30 mg/kg) was given before the first incision. To maintain respiration, a tracheostomy was preformed. The ventilator was set for 100 strokes/min. The abdomen was then opened, and the mesenteric windows were carefully wrapped in gauze. Rats in experiments A–C (see Table 1, cromolyn drip) received a suffusion of cromolyn, and therefore the gauze was presoaked in warm (37° C) cromolyn solution [5 g cromolyn in 1 liter HEPES-buffered saline with 0.5% bovine serum albumin (HBS-BSA)]. The cromolyn-soaked wrap was then placed on a platform of wet gauze level with the abdomen. The warm cromolyn-HBS-BSA solution was continuously dripped on the gauze-wrapped mesentery throughout the experiment to ensure adequate cromolyn suffusion. Rats in experiment D (Table 1) did not receive the cromolyn drip and were continuously suffused with HBS solution.

View this table:
Table 1.

Data recorded for each treatment

Once the portal vein was exposed, rats in experiment A received a warm cromolyn-HBS-BSA injection (0.1 ml/100 g body wt) in the portal vein in the direction of blood flow using a 28.5-gauge needle. This procedure was repeated 30 min later. Cromolyn was injected before the mesentery was manipulated because mast cells are exquisitely sensitive to the physical disturbances that are likely to occur during surgery. Rats in experiments B and C were administered two warm 2% HBS-BSA injections (0.1 ml/100 g body wt) instead of the cromolyn injection into the portal vein.

To expose the superior mesenteric artery, surgical forceps were used for blunt tissue dissection. Once the artery was free of adipose and connective tissue, two 4-0 surgical sutures were tightly secured around the artery with sufficient space between each to allow perfusion of ∼6–10 mesenteric windows. We define a mesenteric window as the mesenteric tissue lying between two adjacent pairs of feeding arterioles/collecting venules. These ligatures allow for perfusion of only a select number of windows and ensure proper clearance and pressure of fluid within the feeding arterioles and venules. In Fig. 1, only one mesenteric window is shown for clarity. The cannula (catheter that would later be placed into the mesenteric artery for the injection of appropriate reagents) was positioned just upstream of the downstream ligature, and a clamp was placed just upstream of the cannula. This clamp helped to prevent backflow of blood into the cannula, or catheter, that would later be placed into the mesenteric artery for the perfusion of appropriate reagents. Another 4-0 suture was loosely tied around the site of cannulation to help secure the cannula when in place.

Fig. 1.

Diagram to illustrate the mesenteric window preparation. Although this preparation is similar to the experimental design of Baldwin and Thurston (6), we altered the position of the ligations and clamps to ensure a more efficient circulation of the appropriate reagents. The collecting venules (which connect the intestine to the mesenteric artery) separate the individual mesenteric windows. One full window is centered in the diagram. The circulation of the windows was effectively isolated from the rest of the mesenteric circulation by ligating the mesenteric artery at the proximal and distal ends of the chosen windows. The artery was then cannulated near the most distal collecting venule. While cannulating, a clamp was placed on the artery to prevent the backflow of blood. This clamp was removed after the cannula was secured. The portal vein was then cut to allow the microvasculature of the windows to be perfused free of blood via the cannula.

After the cannula was inserted retrograde into the artery and secured in place with the suture, the clamp was removed. A small incision was made in the portal vein to allow the microvasculature to be flushed clear of blood. First, the blood was cleared by injecting 5 ml of warm heparin flush (22 mg heparin/4 ml deionized water at pH 7.4) through the cannula. In all cases, the heparin flush could be seen entering the feeding arteriole, followed by the microvascular network, and finally flowing through the collecting venule into the mesenteric vein. Next, in 15 of the animals (groups A, B, and D), 5 ml of warm DBBF-Hb solution (2 mg/ml in HBS-BSA) was injected. The portal vein was then clamped, and the DBBF-Hb solution was allowed to incubate within the mesenteric microvasculature over a 10-min interval. During the 10-min interval, another 1 ml of DBBF-Hb solution was injected every 3 min, removing the portal vein clamp for each extra 1 ml added. Therefore, a total of 8 ml of DBBF-Hb solution was injected into the rat mesentery over 10 min. Two of the animals that received the HBS-BSA injections did not receive the DBBF-Hb infusion (group C). Thus the effects of HBS-BSA injections and cromolyn suffusion alone on microvascular leakage could be determined.

The portal vein clamp was removed once more and DBBF-Hb was replaced by warm FITC-albumin solution (5 mg FITC in 10 ml HBS-BSA). The portal vein was reclamped, and the FITC-albumin was allowed to incubate in the microvasculature for 1 min. Next, the clamp was removed, 3 ml of cold (4° C) fixative (3% formaldehyde in HBS) were perfused via the cannula, and the portal vein was then reclamped. An extra 2 ml of fixative were added while the vein was still clamped to maintain pressure inside the mesenteric microcirculation (∼40 mmHg). At this time, the animal was promptly terminated with 1 ml of Beuthanasia into the heart. The cromolyn drip (or HBS drip) was removed, and the gauze was soaked in the fixative and covered in ice. The fixation was allowed to continue for 60 min at 4° C. Every 10–15 min, the portal vein clamp was removed, an extra 1 ml of fixative was added, and the clamp was replaced.

The mesenteric tissue was carefully excised, placed in a small vial containing fixative, and refrigerated for at least 1 h. The mesenteric windows were then carefully separated and individually mounted between a glass slide and a glass coverslip using aqueous mounting medium (Vectashield, Vector Laboratories; Burlingame, CA). Baldwin et al. (9) have previously described a similar procedure.

Assessment of venular leakage. Vascular leakage was assessed by measuring the number and area of regions with extravascular FITC-albumin. Each slide was examined microscopically using a ×10 objective (numerical aperture 0.6) on a Zeiss Axioplan light microscope. The light source was a 100-W Hg lamp for epifluorescence and a halogen lamp for transmitted illumination. A video camera (Optronix 750D) was mounted at the camera port of the microscope. Ten to twenty images were recorded per slide in 5-s intervals to obtain the appropriate sites where vascularization was present. Videotaped images were later analyzed using an analog-to-digital converter and appropriate software (NIH Image) to measure the length and diameter of each venule, number of leaks per venule, and area of each leak. Data were pooled within each group [i.e., cromolyn pretreatment with DBBF-Hb (group A); HBS-BSA pretreatment with DBBF-Hb (group B); HBS-BSA pretreatment alone (group C); and no pretreatment followed by DBBF-Hb (group D)] and the following values were calculated: 1) the average number of leaks per length of venule, and 2) the average area of leaks per micrometer of a venule.

Assessment of DMC. After the leak data were recorded and analyzed, the largest three windows from each animal preparation were selected for analysis of DMC. The coverslip was removed, and the mounting medium was carefully removed with deionized water. An aqueous 1% toluidine blue solution was suffused on the preparation for 20 s and washed off with deionized water. The slides were then placed under the microscope and viewed using a ×20 objective (field area 1.13 mm2). DMC were counted in each field as the slide was scanned systematically in columns up and down, from right to left. The possibility that the location of the leaks was directly correlated with the position of the DMC was examined. Two tissue slides were randomly selected from each experimental preparation. With the coverslips removed, the tissues were video recorded using fluorescence microscopy to locate microvascular leaks and systematically scanning the slides in vertical columns from left to right. The slide was then returned to its original position on the microscope stage. Next, without moving the slide, a drop of 0.1% aqueous toluidine blue was placed on the slide for 10 s and then absorbed with tissue paper. The slide was then scanned as before and the number of DMC in each frame was counted.

Statistical analysis for animal experiments. Each parameter was compared among groups A, B, and D using one-way ANOVA. Group C was not used in this analysis because it only contained two animals. The low leak numbers and areas obtained in these two animals were sufficient to prove that the injection process, per se, was not increasing microvascular leakage. An additional group from a previous study (9) was used in this analysis for further comparisons. This group entailed HBS-BSA perfusion without any pretreatment or cromolyn suffusion. If a significant difference was found between groups, pairs of groups were compared using Student's t-test with a P value < 0.05 indicating statistical significance. All values are presented as means ± SE; n used in these studies is the number of venules per group and that used for mast cell degranulation was the number of mesenteric fields of view examined per group, both of these quantities being the units of variation.


Distribution of leaks. A previous study (9) showed that microscopic examination of the HBS-BSA perfusion experiments (n = 374) by epifluorescence revealed very few leaky sites. In the present study, the HBS-BSA injection pretreatment with cromolyn suffusion (n = 156, experiment C) produced similar results. However, the DBBF-Hb perfusion treatment revealed an abundance of microvascular leaks, which occurred mainly in the venules (n = 221, experiment D) as opposed to the arterioles and capillaries. This result was similar to that demonstrated in a previous study (9). Significantly fewer leaks were seen in venules of preparations that were suffused with cromolyn and pretreated with either cromolyn (n = 490, experiment A) or HBS-BSA injections (n = 330, experiment B).

Quantification of leaks. Figure 2A illustrates the average number of leaks per micrometer of venule that resulted from each treatment. The mean number of leaks per micrometer of venule length for each treatment was as follows: experiment A, 0.50 ± 0.14 × 10–3 leaks/μm venule; experiment B, 0.67 ± 0.15 × 10–3 leaks/μm venule; experiment C, 0.09 ± 0.06 × 10–3 leaks/μm venule; and experiment D, 1.52 ± 0.37 × 10–3 leaks/μm venule. Suffusion with cromlyn, with or without cromolyn injection pretreatment (groups A and B), significantly reduced the number of leaks per micrometer of venule produced by DBBF-Hb (group D).

Fig. 2.

A: average number of mesenteric leaks per venule length (in leaks/μm venule length) for each of the 4 different treatments. B: average mesenteric leak area per venule length (in μm2/μm) for each of the 4 different treatments. C: average number of degranulated mast cells (DMC) per microscopic field of view for each of the 4 treatments. Error bars indicate SE. In all cases, significant differences are as follows: *treatments significantly lower than the diaspirin cross-linked hemoglobin (DBBF-Hb) alone. Bar A, DBBF-Hb with cromolyn injection and cromolyn suffusion; bar B, DBBF-Hb with HEPES-buffered saline-bovine serum albumin (HBS-BSA) injection and cromolyn suffusion; bar C, HBS-BSA with cromolyn suffusion; bar D, DBBF-Hb alone.

As illustrated in Fig. 2B, the mean area of leaks per micrometer of venule length for each treatment was as follows: experiment A, 0.100 ± 0.040 μm2/μm venule; experiment B, 0.114 ± 0.032 μm2/μm venule; experiment C, 0.008 ± 0.006 μm2/μm venule; and experiment D, 0.119 ± 0.034 μm2/μm venule. The value for experiment A did not differ significantly from that found in experiments B or D. On the other hand, experiment C produced very low leak areas (no statistics, n = 2 for group C). Thus although cromolyn suffusion (with or without cromolyn injection) significantly reduced the number of leaks per micrometer of venule produced by DBBF-Hb, it did not reduce the average leak area per micrometer of venule. This observation suggests that the individual leaks must be larger on average with the cromolyn suffusion than with DBBF-Hb injection alone. Therefore, either cromolyn inhibits the formation of the smaller leaks or the leaks are aggregating into larger leaks, thus reducing the number of leaks without changing the total area. The average individual leak areas for the treatments including DBBF-Hb were as follows (means ± SD): experiment A, 160 ± 238 μm2 (percentage of leaks <50 μm2 = 45%); experiment B, 212 ± 307 μm2 (percentage of leaks <50 μm2 = 35%); and experiment D, 70 ± 75 μm2 (percentage of leaks <50 μm2 = 70%). Thus treatment with cromolyn (groups A and B) significantly increased the individual leak area of preparations perfused with DBBF-Hb (group D).

Mast cell degranulation. DMC were easily distinguished from intact mast cells because of the fashion in which the mast cell granules are expelled into the surrounding tissue, as demonstrated by toulidine blue. The mast cells were best observed at the ×20 optical view with a 1.13-mm2 field of view. Figure 3A illustrates how the cromolyn treatment (injection and suffusion) effectively stabilized the mast cells during perfusion with DBBF-Hb. The intact mast cells were typical of the preparation as a whole. Figure 3B shows an example of the minimal mast cell degranulation that was observed when saline injection and cromolyn suffusion accompanied DBBF-Hb perfusion. One DMC can be seen (arrow). The mean number of DMC for the treatments was as follows: experiment A, 1.07 ± 0.12 (n = 458); experiment B, 1.68 ± 0.23 (n = 212); experiment C, 0.17 ± 0.08 (n = 90); and experiment D, 4.40 ± 0.40 (n = 340). There was a significant difference between the DBBF-Hb preparations treated with cromolyn (groups A and B) and the DBBF-Hb preparations with no cromolyn (group D) (Fig. 2C). In addition, group A, which was treated with cromolyn injection and suffusion before DBBF-Hb infusion, showed significantly less mast cell degranulation than group B (saline injection and cromolyn suffusion).

Fig. 3.

A: light micrograph showing mesentery stained with 1% aqueous toluidine blue to reveal intact mast cells. This field of view demonstrates intact mast cells as a result of the cromolyn injection and suffusion. Scale bar = 25 μm. B: light micrograph showing mesentery stained with 1% aqueous toluidine blue to reveal DMC. In this field of view, from a HBS-BSA treatment with cromolyn suffusion, there is clear evidence of a mast cell releasing its granules (arrow). Scale bar = 25 μm.

It was hypothesized that the fields with a high number of DMC would correlate with the fields with microvascular leaks. However, no such correlation was observed. Because the granules are expelled away from the mast cell, it was thought that venules further removed from the DMC might also be affected. Therefore, areas four times larger than the field of view were considered by grouping four consecutive fields of view together and overlapping two fields within each larger group (i.e., 1, 2, 3, and 4 = group 1; 3, 4, 5, and 6 = group 2). Again, no correlation was observed between the location of the DMC and the leak sites, even for the larger areas (R2 = 0.0528). Al-Naemi and Baldwin (4) also determined that there is probably not a direct link between mast cell degranulation and leak formation induced by applying exogenous histamine. Similar to the present study, their unpublished data indicated that the leak sites did not coincide with the exact sites of mast cell degranulation.


This study has demonstrated that the suffusion with cromoglycate sodium salt significantly reduces the numbers of DMC and microvascular leaks caused by the synthetic blood substitute DBBF-Hb. In fact, the cromolyn suffusion and preinjection, in conjunction with the administration of DBBF-Hb, proved extremely effective in significantly reducing the number of DMC below that of a previous HBS-BSA control [1.28 ± 0.20, n = 152 (9)] and in minimizing the number of leaks (similar to the HBS-BSA control). If the cromolyn preinjection was substituted by a saline injection (group B), the leak number was significantly greater than that of the HBS-BSA control, although it was still only a fraction of the value obtained for DBBF-Hb alone (group D). In addition, in these animals, the number of DMC per field of view, although small, was significantly larger than for the group with cromolyn preinjection (group A). Thus it appears that preinjection with cromolyn, in addition to suffusion with cromolyn, further enhances mast cell stability. However, despite cromolyn's mast cell-stabilizing capabilities, some mesenteric leaks still formed in response to DBBF-Hb, and their total area per unit venule length was not significantly lower than that for group D (DBBF-Hb alone). In fact, the mean area of individual leaks was increased in groups A and B compared with group D. The HBS-BSA treatments [group C and the previous HBS-BSA control (9)] both produced lower leak areas compared with any of the DBBF-Hb treatments [statistical significance for HBS-BSA control (0.02 ± 0.02 μm2/μm venule) versus DBBF-Hb treatments in groups A, B, and D]. To the best of our knowledge, this study is the first attempt to demonstrate the connection between cromolyn's ability to stabilize mast cells and its effect on increased microvascular permeability produced by a modified Hb.

Our observation that cromolyn has the ability to reduce the number of DMC is consistent with other studies (10, 2224, 29). By reducing mast cell degranulation, the release of mast cell mediators in the surrounding tissues will be reduced. There is ample evidence to support the theory that histamine, a mast cell mediator, causes tremendous increases in vascular permeability (28). Therefore, it was not surprising that treatment with cromolyn reduced the number of leaks produced by DBBF-Hb. However, the fact that cromolyn did not reduce the mean leak area per micrometer of venule was surprising and indicates that there is a property of DBBF-Hb that induces microvascular leaks independently of the inflammatory mediators released by mast cells. One possible suggestion for this mechanism involves the formation of ROS by Hb. Perhaps ROS are reacting directly with the venules, causing large interendothelial cell gaps and therefore increasing the permeability. It is known that a bolus injection of DBBF-Hb causes formation of ROS in intestinal tissue (8) and that reaction of ROS with microvessels increases their permeability to macromolecules (11). Leak formation by ROS would not require mast cell degranulation, but the ROS formed by oxidation of the modified Hb could diffuse out of the venules and into the tissue and then cause mast cell degranulation, which could exacerbate the leak formation.

The lack of a correlation between exact sites of mast cell degranulation and microvascular leakage suggests that the products released by mast cells in response to Hb do not necessarily exacerbate leakage in nearby venules. One possibility to explain this surprising finding is that the high permeability sites in the microvasculature are dynamic. In other words, leak formation will only occur, in response to mast cell mediators, at sites in which the endothelial cells are in a susceptible phase of their cycle. Because the endothelial actin cytoskeleton and junctional protein strands are continually altering, it is plausible that the smaller leaks can close up as new ones form; thus the leak sites will change over time. Possibly, ROS could diffuse out of the leaky venules and cause nearby mast cells to degranulate and meanwhile the leak could reseal. This would result in mast cell degranulation with apparently no leak formation. This idea is consistent with the hypothesis suggested by the data that mast cell mediators produce a population of smaller leaks than does DBBF-Hb alone. It is also possible that mast cells could have reduced susceptibility to degranulation under certain conditions or that some other mediator may be involved between production of ROS and degranulation of mast cells. If so, ROS could produce a leak in a venule and diffuse into the adjacent tissue without causing nearby mast cells to degranulate either because they are in a “robust” condition or because they have not been stimulated by an intermediate component. Such an intermediate mediator could be substance P released from peripheral nerves. This would be consistent with the idea that cromolyn acts as a mast cell stabilizer by reacting with substance P.

To summarize, it appears that DBBF-Hb induces venular leak formation by a process that does not require mast cell degranulation but that this process can also cause mast cells to degranulate. The lack of a direct correlation between sites of mast cell degranulation and sites of vascular leakage suggests that the endothelial cells in the same venule vary in their response to a given degree of inflammatory stimulus.


The authors give special thanks to Dr. Ed Valeski for the guidance and expertise on the surgical procedures and to Ela Cudillo for aiding in data collection and analysis.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract