A laboratory model to quantitate the resistance of collagen vascular grafts to biodegradation Joseph Megerman,* Esphiran Reddy,' Gilbert J. L'Italien, David F. Warnock, and William M. Abbott Vascular Research Laboratory, Surgical Services, Massachusetts General Hospital, and Department of Surgery, Harvard Medical School, Boston, Massachusetts 02114 and 'Department of Surgery, Faculty of Medicine, University of Natal, PO. Box 17039, Congella 4013, South Africa Recent reports have shown that despite extensive preclinical testing, vascular grafts of biological origin undergo severe biodegradation and aneurysm formation after two or more years of implantation in man. The purpose of this study was to develop a laboratory model to quantitate and correlate the stability of crosslinked collagen grafts in vitro and in vivo. This resistance to biodegradation was assessed by measuring changes in suture pullout force and sample weight in response to controlled digestion with bacterial collagenase, in 0.5-cm-long cylindrical graft segments (chemically processed bovine carotid artery and human umbilical cord vein) that were implanted in the rat subcutis for 2 to 12 weeks. Scar tissue was removed from the explants by brief
enzymatic digestion, a process that was inhibited when graft segments had become infected. Changes in dry weight were more consistent than were changes in wet weight; drying the graft segments had no effect on their degradation in vivo or in vitro. Intact cylindrical rings suffered somewhat less damage than did opened, flattened cylinders. Graft degradation increased markedly with implantation time, and was detected after only 3 weeks. We conclude that the rat subcutis model, when combined with controlled enzymatic digestion, first to remove scar tissue and then to challenge structural integrity, provides an accelerated assay by which to predict the stability of collagen vascular grafts.
IN TRODUCTION
Chemically crosslinked biologic grafts have been used for distal vascular reconstructions, in the absence of suitable autogenous vein, for over 30 years."6 However, many of them undergo continual biodegradation and become aneurysmal after 2 or more year~,ZI-'~ despite the expectation, from extensive animal implantation and in vitro st~dies,',2~-~,'~-'~ of a more favorable result. It is apparent that previous laboratory tests did not adequately predict this structural biodegradation. The questions are (a) whether such degrada*To whom correspondence should be addressed at Vascular Research Laboratory, ACC 458, Massachusetts General Hospital, Boston, MA 02114. Supported by grants from the Harvard University South Africa Fellowship Programs (E.R.) and the Fritz Straub Vascular Research Fund at the Massachusetts General Hospital. Journal of Biomedical Materials Research, Vol. 25,295-313 (1991) CCC 0021-9304/91/030295-19$04.00 0 1991 John Wiley 8.1 Sons, Inc.
MEGERMAN ET AL.
296
tion or, more accurately, its absence can ever be reliably predicted in the laboratory; (b) whether testing must include in v i m as well as in vitro studies; and if so, (c) whether results can be generated faster than the lengthy times required to observe graft deterioration in man. The stability of crosslinked collagen, the major structural component of biologically derived vascular prostheses, has frequently been assessed by physical or biomechanical measurement^:^ before and after enzymatic or chemical or after subcutaneous i m p l a n t a t i 0 n . 2 ~But ~ ~there ~~~ digestion in ~litro,’~~’”’~-’~ have been few attempts to correlate in vifro to in vivo changes. Stress relaxation during collagenase digestion of aldehyde-fixed collagen fibers was correlated with their weight loss after subcutaneous implantation in guinea pigs for 10 days:’ in vitro weight loss has been compared to the loss of radiolabeled collagen implanted subcutaneously in rats for up to 12 and in vitro diameter/compliance changes after collagenase digestion paralleled changes seen in biologic femoropopliteal implants, which were followed noninvasively in 12 patients for 2 years.I4 Using suture pullout force as a measure of mechanical integrity, it was observed that subcutaneous implantation in rats for only 8 weeks was sufficient to severely degrade similar grafts (Table I, Ref. 26), despite their demonstrated survival in some dogs and patients for over 5 years.15From this, we reasoned that a brief implant period might enhance the sensitivity of in vitro tests to inherent defects in biological prostheses. We, therefore, explored using controlled enzymatic digestion in vitro to quantitate the effects of subcutaneous implantation on the structural stability of two clinically relevant vascular grafts. The results suggest that this combined procedure may be useful in grading the potential stability of such prostheses and may, therefore, have application to quality control in commercial processes. METHODS
Three types of commercially produced biologic grafts were studied, two derived from bovine carotid arteries fixed with adipyl chloride and glutaraldehyde (Types A and B: Solco-P and Solco-M, Solco-Basel) and a third consisting of human umbilical vein fixed with glutaraldehyde alone (Type C: Biograft 11, Meadox Medical). Grafts were divided into ring segments approximately 0.5 cm long and their wet weight, after gentle blotting with soft paper to remove excess water, was measured on a Sartorius balance with a precision (least count) of 0.0001 g. Randomly selected segments served as controls while the rest were implanted in the rat subcutis for 3 to 12 weeks. Three-month-old female Lewis rats were anesthetized wth IM xylazine (3.6 mg/Kg) and ketamine (50 mg/Kg). Using sterile technique, two graft segments were placed in separate subcutaneous flank pouches, one on either side of the midline, through a common dorsal skin incision, and the skin was closed with interrupted 3/0 nylon sutures. At harvest, the rats were anesthetized and the segments easily retrieved through a direct incision over the
QUANTITATING COLLAGEN VASCULAR GRAFT STABILITY
297
implanted segment. Scar tissue was dissected from the segments under magnification to restore their original circular shape. The segments' resistance to degradation by collagenase was quantitated based on loss of dry weight and change in suture pullout force. Some segments from each graft were first dried to a constant weight in a desiccator, under vacuum, at room temperature. Prior to implantation, they were rehydrated in saline for 12 h at 8°C. During this time (2-3 days), paired hydrated samples were stored in a bacteriostatic solution of 0.02% sodium azide in saline at 8°C. Before use, these segments were thoroughly rinsed in normal saline; their dry weight was estimated from data on adjacent dried segments, as described below.
Collagenase treatment Explants and unimplanted control segments were simultaneously subjected to controlled enzymatic digestion, adapting a previous method.14After measuring wet and/or dry weights, the segments were placed in l mL of crude bacterial collagenase,* made up to give either a desired enzyme "concentration'; in Units/mL (U/mL), or "specific activity'; in Units/mg tissue (U/mg), depending on the study. The enzyme activity (U/mg of powder) specified by the manufacturer was assumed. The reaction mixture, in microfuge tubes, was incubated at 37°C for up to 96 h, after which the segments were rinsed thrice in ice cold saline, centrifuged at 10,OOOg for 10 min and the supernatants removed. Wet weight of the sediment was measured (if structural integrity permitted), followed by drying to a constant weight, as before. The weight of all explanted segments exceeded initial values, due to attached scar tissue. This tissue was not always solubilized by collagenase even after 96 h, as residual weights did not always drop to pre-implant levels. Possible explanations for this inconsistency were explored, as described below. A summary of these all studies, including sample groupings and experimental conditions, is given in Table 11.
1. Removal of scar fissue (Experiment 2) Sixteen Type B segments were explanted after 3 weeks, and maximum scar was dissected away without damaging the samples. They were then exposed to collagenase in two successive, brief intervals (average time: 1 h each; range: 15 to 120 min) at different concentrations (50 and 150 U/mg dry weight of sample just prior to digestion), followed by a third digestion with 50 U/mg dry weight for 96 h. All segments were weighed wet and dry before each exposure to with enzyme; comparisons with (estimated) dry weights prior to implantation yielded dry weight of explanted scar tissue. *Type la, #C9891, Sigma Chemical Co., St. Louis, MO, in a calcium-containing, 0.01 M TRIS buffered, lactated Ringer's solution (pH 7.4).
MEGERMAN ET AL.
298
2. Effect of drying (Experiment 3)
Segments of 2 Type B grafts were divided into two groups, one of which was dried to constant weight. From one graft, paired wet and dry segments were implanted for 3 and 6 weeks, one pair per rat; four pairs served as unimplanted controls. After explantation, all segments were weighed wet and, without drying, were placed in 180 U/mg scar tissue (estimated dry weight, see below) for 1 h. The residues were dried and weighed to determine the effect of this short digestion, and were then further digested by 1900 U/mL for 96 h. 3. Effect of sample configuration (Experiment 4)
A segment of tubular graft can be implanted in a cylindrical (intact ring) or rectangular (cut lengthwise and flattened) configuration. The ring rapidly becomes encased in scar tissue, which may partially shield it from its surroundings, while the rectangle exposes more surface area per unit volume to host enzymes and may thus suffer greater structural degradation by the host. Fourteen segments of a Type B graft, seven of which were cut and flattened, were studied in pairs. Four pairs were implanted for 3 weeks, without prior drying; three pairs were dried and served as unimplanted controls. All segments were exposed to 950 U/mL for 96 h; some were further digested with 50 U/mg dry weight for 30 min. 4. Effect of infection (Experiment 5) The scar tissue that encased several infected 3-week explants appeared to be less readily digested by collagenase. To determine if infection did, in fact, make the scar tissue more resistant to enzymatic degradation, 16 Type B segments were randomly assigned to two groups, one of which was intentionally contaminated with Pseudornonus ueruginosa.+The sterile and septic segments were implanted in separate rats; they were not paired within the same rat to avoid contamination of the sterile implants. After 2 weeks, all explanted segments were septic or sterile, as before. They were dried, weighed, and exposed twice to 200 U/mg current dry weight for 1 h, followed by 96 h in 100 U/mg in 1 mL.
Suture pullout force Two 4/0 silk sutures with tapered needle were passed through the wall of 5mm-long ring segments, 1 mm from each edge. The two sutures were +StrainATCC 27853, McFarland #1 (3
X
lo6 Colony Forming Units per segment).
QUANTITATING COLLAGEN VASCULAR GRAFT STABILITY
299
collinear and parallel to the axis of the ring. One suture was tied to a force transducer (Grass Instruments #FT03) and the other to a linear motion device. The segment was stretched at a rate of 0.3 mm/s until it tore at its weakest point, usually in a circumferential cleavage plane; occasionally the suture cut through the wall without causing a circumferential split. The force at which either event occurred was defined as the suture pullout force. Replicate measurements were obtained by dividing each ring segment lengthwise into two troughs ("half-rings"), and performing a single measurement on each. Since suture pullout measurements are destructive to the sample, they were not performed on segments being implanted. RESULTS
After 3 to 12 weeks in vivo, segments of graft Types A and B demonstrated a profound decrease in suture pullout force (SPF), but only after being digested with collagenase (Fig. 1).This contradicted a similar study of related grafts, wherein SPF decreased significantly without in vif YO digestion (Table I). This discrepancy, and the high variability we observed in SPF and weight loss results, were attributed to the inconsistent dissection of scar tissue from the explanted graft segments. More careful dissection alone did not reduce this variability, so we explored using limited enzymatic digestion to selectively remove scar tissue, before measuring SPF and weight loss. Changes in wet weight following enzymatic digestion were somewhat unreliable; for example, the wet weight of unimplanted (control) segments in-
' 600 0° n
i
5
500 W
8 CY
400 300 -
:200 / I-
1
g
100
I
Q
OC 1
I
0
3
7
12
WEEKS IMPLANTED
Figure 1. Suture pullout force vs. time of subcutaneous implantation in rats for two biological vascular grafts (Type A: circles, Type B: squares), ~) with collagenase for 96 hours (Experibefore (o,.) and after ( 0 ,digestion ment #1 in Table 11). Each point represents mean 2 standard deviation of measurements on two to four 0.5-cm ring segments.
MEGERMAN ET AL.
300
TABLE I Suture Pullout Force in Subcutaneous Implants of Biological Prostheses” Identification
Pre-implant
Expt. 1 (2-12 weeks) M1
Postimplant
Decrease
398 600 969
? k ?
143 (12) 127 (12) 169 (26)
128 f 9 5 ( 9 ) 288 f 141 (11) 609 f 205 (26)
68% 52% 37%
M10 M11 M12
364 269 268
? ? ?
91 (15) 48 (15) 56 (15)
94 f 62(12) 149 f 52 (14) 132 f 48 (15)
74 7% 45% 51%
GO G1 G2
724 2 156 (15) 767 ? 141 (15) 847 ? 153 (15)
134 f 71 ( 6) 550 ? 162 (15) 617 & 175 (15)
81% 28% 27%
M2 G2
Expt. 2 (8 weeks)
“Mean values 2 standard deviation (no. of samples) of maximum force (grams) maintained by 6-0 suture passed 3 mm from the end of 0.8 X 3.0 cm rectangular segments for 2 15 s. (Ref. 26).
variably increased after exposure to collagenase. Consequently, whenever possible, weight loss measurements were based on dry weights. When segments were not dried before implantation, their initial dry weights were estimated from measured wet weights, times an appropriate dry-to-wet weight ratio, derived from dried segments of the same graft. Because this ratio often varied considerably along the grafts’ lengths (Fig. 2), estimates were obtained by interpolation between measured values from closest adjacent dried segments, rather than from a pooled average. The dry weight of undried scar tissue was also estimated from its wet weight (explant minus pre-implant wet .19
0 c 6
LT I-
.17
I
13
I
0
5
10
15
20
25
SEGMENT NUMBER
Figure 2. Longitudinal variation of measured dry-to-wet weight ratio in 3 randomly chosen Type B grafts.
301
QUANTITATING COLLAGEN VASCULAR GRAFT STABILITY
weight of the graft segment) and a regression line (Fig. 3) derived from wet and dry weights of (a) isolated scar tissue measured in a separate study (Table I), and (b) graft-plus-scar explants, in 2 current studies, when corrected for pre-implant wet and dry weights of the graft segments. Figure 4 summarizes the weight changes experienced by ring segments of type B grafts, in 3 experiments. The excess weight of explants, compared to initial dry weights, was assumed to be scar tissue, most of which was removed by 1 to 2 h of digestion with collagenase. Weight loss to significantly below the weight of unimplanted controls did not occur, in most cases, without more extensive (96 h) digestion. The 6-week explants form a notable exception, since digestion for only 1 h caused significantly more degradation than occurred in the 3-week explants. Also noteworthy is the greater resistance to collagenase of scar tissue that formed around the deliberately infected segments; dry weight loss by infected and sterile explants of comparable size was 3.4 2 0.6 vs. 9.4 2 3.4 mg, respectively (mean rt SD, p c .001, unpaired t-test). The infected explants lost somewhat more weight during subsequent enzyme digests, but they ultimately weighed more than did the sterile segments (fractional weight loss: -0.6 2 6.0% vs. 12.2 2 3.7%, p c .OOl). Whether the residues contained any scar tissue or consisted of original implant alone could not be determined. Dehydration had no effect on dry weight loss due to enzyme digestion, p > .05, before or after implantation (Fig. 5). However, undried segments produced greater variability, probably due to errors in estimating initial dry weight. In what follows, results for dried and undried samples, within a
6o
50
1t I
I-
I
3
lo 0
1 0
/
J
50
100
150
200
250
300
350
WET WEIGHT (mg)
Figure 3. Dry weight vs. wet weight of scar tissue, derived from Type B ring segments that were implanted subcutaneously for 2 or 3 weeks (A, 0), and from isolated scar tissue associated with 8-week explants (+;Table I). The pre-implant dry weight of ring segments was estimated using wet weight multiplied by appropriate dry/wet weight ratios (see text); all other wet and dry weights were measured directly. Linear regression line is highly significant ( p < .0001).
MEGERMAN ET AL.
302 15 r
w
I
z -5
+
c3
a
"
I
I
/+
-10 L
1 2 CUMULATIVE ENZYMATIC DIGESTION (hrs)
0
98
Figure 4. Changes in dry weight of unimplanted control (o,O,V) and explanted (all others) ring segments of Type B grafts, after sequential diges3 (o,V), tion steps (mean values only). Segments were implanted for 2 (.,[?), or 6 (A)weeks. Samples intentionally infected with Pseudornonas:[?.For simplicity, abscissa gives average cumulative time of digestion, and dashed line was drawn between pooled average weight change for unimplanted controls. 50
r
40
t
1
WEEKS IMPLANTED Figure 5. Dry weight loss of ring segments that were ( 0 )or were not (0) dried prior to subcutaneous implantation. After being explanted, all segments were treated identically, being exposed to successive collagenase digestion steps of 1 and 96 h. Each point represents mean SD of four segments. In all cases, weight loss was referenced to pre-implant dry weight (estimated for nondried samples as detailed in text). +_
given experiment, have been combined. Configuration had a small but significant effect on dry weight loss after 3 weeks in vivo, followed by in vitro digestion for 96 h: 13.4 2 1.4% (n = 3) vs. 19.8 +- 3.9% (n = 3) for the cylindrical and rectangular samples, respectively ( p = C.05).
QUANTITATING COLLAGEN VASCULAR GRAFT STABILITY
303
Weight loss during enzymatic digestion depended on the relative amounts of tissue and enzyme. This relationship was best described in one of two ways. In most cases, a constant value of collagenase activity per unit weight of substrate yielded a range of enzyme concentrations (U/mL), to which weight loss (mg) per unit time was well correlated. With uniform concentration, however, weight loss per unit initial dry weight (mg/mg or %) was best correlated with a reduced enzyme concentration (U/mL/mg) or specific activity (U/mg in 1 mL). Figure 6 gives examples of these relationships. To compare results obtained after different digestion times, weight loss was plotted against "total enzyme action'; defined here as enzyme concentration X time. Figure 7 displays weight loss by segments of grafts B3, B5, and B7, during sequential digestion steps. The rate of digestion, proportional to the slope of weight loss vs. total enzyme action, decreased with each step, commensurate with the progressive loss of more easily digested scar tissue. To adjust for differences in enzyme concentration, activity, and/or exposure time, each graft segment's capacity to resist biodegradation was expressed as
/
01
0
0
I
I
I
I
I
1
2
3
4
5
I
6
7
CONCENTRATION 6o
r
0
50
100
150
200
250
SPECIFIC ACTIVITY
Figure 6. Dry weight loss by graft segments as a function of enzyme concentration (Units/mL, x W3)or specific activity (Units/mg dry weight, in 1 mL). Top: subcutaneous explants of Type B grafts following a 1 h exposure to collagenase (symbols as in Fig. 4). Bottom: unimplanted Type A graft after collagenase digestion for 96 h.
MEGERMAN ET AL.
304 20
-
STEP 1
n
m 15
I 1
E
W
v) v)
9
10
STEP 3
I-
I
52
s =
STEP 2
0
0
10
5
//
100
150
200
250
TOTAL ENZYME ACTION
Figure 7. Dry weight loss vs. total enzyme action, expressed as the product of concentration and time (1,000 Units/mL x h), for successive enzyme digestion steps. Steps 1 and 2 correspond to the 1-h exposures (average) to collagenase; the 96-h exposure is uniformly identified as step 3. Data points show mean 2 SD of x and y coordinates; regression lines based on mean values are drawn to include the origin, and exclude "septic" group for step 1.Symbols are as given in Figure 4.
weight loss divided by the cumulative sum of total enzyme action. This parameter, for all grafts tested, is plotted as a function of implantation time in Figure 8. Just 3 weeks in vivo had a significant effect on the susceptibility of graft segments to in vitro degradation. This result correlates well with the
/
/@
40 20 -
L
I
3 6 9 12 WEEKS IMPLANTED Figure 8. Dry weight loss, referenced to pre-implant dry weight and enzyme concentration (% loss per Unit-hour in 1 mL), vs. time of implantation, in Type A (0), C (0),and B (all others) graft segments after completing all digestion steps. Paired wet and dry segments have been combined (see Fig. 6 ) ; data points represent mean values for each experiment, including infected samples (0).
0
QUANTITATING COLLAGEN VASCULAR GRAFT STABILITY
305
previously observed decrease in suture pullout force following collagenase digestion of the explanted segments (Fig. 1). In all 2- and 3-week explants, in d u o weight loss after 1 h was less than the estimated weight of scar tissue. Assuming this weight loss reflects the digestion of scar tissue alone, and none of the graft, one can estimate how much enzyme will remove this scar tissue entirely. Figure 9 shows the fractional loss of scar tissue vs. enzyme specific activity, referenced to the dry weight of explanted scar tissue. Omitting the infected segments, these coarsely estimated values are surprisingly well correlated (u = 0.89, p c .02); they suggest that, in 1 h, approximately 10,000 collagenase U/mg will solubilize all of the scar tissue in these closely dissected 3-week explants, which weighed up to 25 mg (dry weight). Even without removing all scar tissue, however, a 1-h digest significantly reduced their mechanical strength compared to that of unimplanted controls, as measured by suture pullout force (Fig. 10). This result seemed independent of graft type.
DISCUSSION
Biologically derived grafts will continue to be an attractive conduit for arterial bypass, when no vein graft is available, for a number of reasons. These grafts tend to have a less thrombogenic surface, more desirably compliant wall, and a lower rate of infection, when compared to currently available synthetic materials.',35Unfortunately, many of these biologic grafts tend to form aneurysms, which often are detected only after several years.'-" Nevertheless, many have not failed, and continue to function well even after
0 0
2
4
6
8 1 0 1 2 1 4
SPECIFIC ACTIVITY
x 1o - ~
Figure 9. Estimated fractional loss of scar tissue (dry weight) in 2- and 3-week explants of Type B segments vs. enzyme specific activity (Unithours/mg dried scar in 1 mL; mean ? SD for each experiment; infected samples: 0).
MEGERMAN ET AL.
306
0
P96
DIGESTION TIME (hours) Figure 10. Fractional change in suture pullout force, following enzymatic degradation of unimplanted controls (0) and 3-week explants ( 0 )for up to 96 h. Statistical significance after 1 h ( p < .03, explant results compared to unity) and 96 h ( p c .005) was determined using 2-tailed t-tests. Points indicate mean & S.E.M of measurements of all graft types.
10 year^.^^,^^ Are these successes due to the chance lack of defects in grafts produced by an inherently sound chemical process, or to their use in patients who, also by chance, are unable to degrade any foreign biologic implant, however poorly processed? Two observations suggest that patient variability is less important than is consistency of biologic graft processing: (a) water content can vary considerably along the length of these biological grafts (Fig. 2); and (b) a composite ilioperoneal bypass fashioned from four Type Alike grafts in tandem failed within 2 years, but with marked dilatation in only one of the four segments.26At a minimum, therefore, tests of quality control must be improved. The effective screening of biologic vascular prostheses requires an assay that permits frequent sampling to account for biological variability, and rapidly measures the graft's resistance to biodegradation over a projected period of several years, while consuming only a small fraction of the graft. In all respects, a test based on chronic intra-arterial implants is unacceptable: clinically relevant sizes (3-6 mm I.D.) require the use of large mammals, a costly and difficult option; detecting structural failure in the animal implant may take as long as in man; and such testing consumes too much of the graft. Therefore, biologic grafts must be evaluated by alternative means. Ideally, grafts should be tested nondestructively in vitro, using means that do not preclude their subsequent use in patients. Some in vitro tests of small graft samples, which measure elasticity, heat shrink temperature, and resistance to chemical or enzymatic degradation, have been devised to characterize collagen stability in crosslinking experiment^.'^^'^-^^"^ It is not clear, however, that in vitro testing adequately represents in v i m degradation due to the
QUANTITATING COLLAGEN VASCULAR GRAFT STABILITY
307
host’s biological responses, including inflammation and possible immunological attack. Therefore, numerous in vivo models have been used to evaluate collagen bioprostheses, including the rat In most cases, including the study summarized in Table I, weight loss and mechanical measurements were easily performed, and assumed to be independent of scar tissue explanted together with the retrieved segments. Our own results, however, were critically sensitive to the amount of attached scar tissue, especially when using small segments of graft. In all cases, the segments weighed more when explanted than before they were implanted, and despite all attempts to surgically remove this scar tissue, their mechanical strength (by suture pullout force) was unchanged before exposure to collagenase. This strengthening effect of host-generated (scar) tissue on biomaterials, which has previously been observed with cellulose spongesz5and vascular anastomoses?*could make weight loss and mechanical changes following the simple removal of subcutaneous implants too unreliable for use in graft evaluation or quality control. We, therefore, explored using limited collagenase digestion to selectively remove scar tissue from the explants. This objective requires that scar and graft collagen have very different susceptibilities to in vitro degradation by collagenase, which is probably the case, judging from the decreasing rate of weight loss in successive digestion steps (Fig. 7). It cannot be assumed, however, that a l-h digestion removes scar tissue alone, nor that it removes it entirely. This was particularly evident in 6-week explants of graft B5 (Table 11, Fig. 4),where digestion with only 180 U/mg scar tissue (908 k 430 U/mL) for 1 h was sufficient to reduce weights to below pre-implant levels. Whether this reflected less scar tissue attached to the explants (subjectively, they seemed to be more easily dissected from their subcutaneous pouches), or increased degradation during the longer implant period is not known. In contrast, scar tissue attached to the deliberately infected %-weekexplants was much more resistant to a greater concentration of collagenase (200 U/mg total explant; 6009 2 584 U/mL); this seems consistent with the increased resistance to degradation of type 111 collagen associated with rheumatoid and inflamed synovial membranes39and inflammation due to glutaraldehydetreated sponges.4o Our results suggest loss of enzyme effectiveness during a lengthy digestion of scar tissue. Explants treated with up to 950 U/mL for 96 h, often weighed more than before implantation, while, their further exposure to only 50 U/mg dry weight (444 f 68 U/mL) for 30 min produced a significant additional weight loss. A similar loss of potency was observed when a single aliquot of collagenase was used to digest successive segments of fresh bovine artery (data not shown). To avoid this problem, separate aliquots of collagenase should be used to digest scar tissue vs. the graft segment. From Figure 9, 10,000 U/mg dried scar tissue for 1 h, or approximately 1,500 U/mg wet scar tissue will effectively remove the host tissue from graft segments explanted after 3 weeks, enabling their further evaluation by in nitro tests. Corresponding figures for longer term implants may be lower (scar collagen
Description
Preliminary comparisons
Expt
1
c2
c1 0 3
0 3
W,D W
W W W,D
-
W,D
-
W,D
-
950 U/mL (96) 950 U/mL (96)
950 U/mL (96) 950 U/mL (96)
950 U/mL (96) 950 U/mL (96)
0 3,7
82
W W
3800 U/mL (96) 950 U/mL (96)
(W, some D) (W, some D)
Step 1
037 12
~
B1
Wt,, 950-3800 U/mL (96) 3800 U/mL (96) 950 U/mL (96)
~~
(W, some D) (W, some D) (W, some D)
Wto
0 3 7,12
~
50 U/D, (.5)
-
50 U/D1 (.5)
-
50 U/Dl (.5)
-
Step 2
Step 3
Concentration of Bacterial Collagenase
A
Graft
Weeks in vivo
TABLE I1 Summary of Experimental Conditions Studied”
86 B7
Configuration
Infection
5
W,D W
W,D W
180 U/Wo (1) 180 U/D,,, (1) 180 U/D,,, (1)
-
-
W,D
-
W,D
200 U/Do (1) 200 U/D, (1)
950 U/mL (96) 950 U/mL (96)
50 U/Do (96)
-
W W,D
50 U/Do (
enzymatic digestion, a process that was inhibited when graft segments had become infected. Changes in dry weight were more consistent than were changes in wet weight; drying the graft segments had no effect on their degradation in vivo or in vitro. Intact cylindrical rings suffered somewhat less damage than did opened, flattened cylinders. Graft degradation increased markedly with implantation time, and was detected after only 3 weeks. We conclude that the rat subcutis model, when combined with controlled enzymatic digestion, first to remove scar tissue and then to challenge structural integrity, provides an accelerated assay by which to predict the stability of collagen vascular grafts.
IN TRODUCTION
Chemically crosslinked biologic grafts have been used for distal vascular reconstructions, in the absence of suitable autogenous vein, for over 30 years."6 However, many of them undergo continual biodegradation and become aneurysmal after 2 or more year~,ZI-'~ despite the expectation, from extensive animal implantation and in vitro st~dies,',2~-~,'~-'~ of a more favorable result. It is apparent that previous laboratory tests did not adequately predict this structural biodegradation. The questions are (a) whether such degrada*To whom correspondence should be addressed at Vascular Research Laboratory, ACC 458, Massachusetts General Hospital, Boston, MA 02114. Supported by grants from the Harvard University South Africa Fellowship Programs (E.R.) and the Fritz Straub Vascular Research Fund at the Massachusetts General Hospital. Journal of Biomedical Materials Research, Vol. 25,295-313 (1991) CCC 0021-9304/91/030295-19$04.00 0 1991 John Wiley 8.1 Sons, Inc.
MEGERMAN ET AL.
296
tion or, more accurately, its absence can ever be reliably predicted in the laboratory; (b) whether testing must include in v i m as well as in vitro studies; and if so, (c) whether results can be generated faster than the lengthy times required to observe graft deterioration in man. The stability of crosslinked collagen, the major structural component of biologically derived vascular prostheses, has frequently been assessed by physical or biomechanical measurement^:^ before and after enzymatic or chemical or after subcutaneous i m p l a n t a t i 0 n . 2 ~But ~ ~there ~~~ digestion in ~litro,’~~’”’~-’~ have been few attempts to correlate in vifro to in vivo changes. Stress relaxation during collagenase digestion of aldehyde-fixed collagen fibers was correlated with their weight loss after subcutaneous implantation in guinea pigs for 10 days:’ in vitro weight loss has been compared to the loss of radiolabeled collagen implanted subcutaneously in rats for up to 12 and in vitro diameter/compliance changes after collagenase digestion paralleled changes seen in biologic femoropopliteal implants, which were followed noninvasively in 12 patients for 2 years.I4 Using suture pullout force as a measure of mechanical integrity, it was observed that subcutaneous implantation in rats for only 8 weeks was sufficient to severely degrade similar grafts (Table I, Ref. 26), despite their demonstrated survival in some dogs and patients for over 5 years.15From this, we reasoned that a brief implant period might enhance the sensitivity of in vitro tests to inherent defects in biological prostheses. We, therefore, explored using controlled enzymatic digestion in vitro to quantitate the effects of subcutaneous implantation on the structural stability of two clinically relevant vascular grafts. The results suggest that this combined procedure may be useful in grading the potential stability of such prostheses and may, therefore, have application to quality control in commercial processes. METHODS
Three types of commercially produced biologic grafts were studied, two derived from bovine carotid arteries fixed with adipyl chloride and glutaraldehyde (Types A and B: Solco-P and Solco-M, Solco-Basel) and a third consisting of human umbilical vein fixed with glutaraldehyde alone (Type C: Biograft 11, Meadox Medical). Grafts were divided into ring segments approximately 0.5 cm long and their wet weight, after gentle blotting with soft paper to remove excess water, was measured on a Sartorius balance with a precision (least count) of 0.0001 g. Randomly selected segments served as controls while the rest were implanted in the rat subcutis for 3 to 12 weeks. Three-month-old female Lewis rats were anesthetized wth IM xylazine (3.6 mg/Kg) and ketamine (50 mg/Kg). Using sterile technique, two graft segments were placed in separate subcutaneous flank pouches, one on either side of the midline, through a common dorsal skin incision, and the skin was closed with interrupted 3/0 nylon sutures. At harvest, the rats were anesthetized and the segments easily retrieved through a direct incision over the
QUANTITATING COLLAGEN VASCULAR GRAFT STABILITY
297
implanted segment. Scar tissue was dissected from the segments under magnification to restore their original circular shape. The segments' resistance to degradation by collagenase was quantitated based on loss of dry weight and change in suture pullout force. Some segments from each graft were first dried to a constant weight in a desiccator, under vacuum, at room temperature. Prior to implantation, they were rehydrated in saline for 12 h at 8°C. During this time (2-3 days), paired hydrated samples were stored in a bacteriostatic solution of 0.02% sodium azide in saline at 8°C. Before use, these segments were thoroughly rinsed in normal saline; their dry weight was estimated from data on adjacent dried segments, as described below.
Collagenase treatment Explants and unimplanted control segments were simultaneously subjected to controlled enzymatic digestion, adapting a previous method.14After measuring wet and/or dry weights, the segments were placed in l mL of crude bacterial collagenase,* made up to give either a desired enzyme "concentration'; in Units/mL (U/mL), or "specific activity'; in Units/mg tissue (U/mg), depending on the study. The enzyme activity (U/mg of powder) specified by the manufacturer was assumed. The reaction mixture, in microfuge tubes, was incubated at 37°C for up to 96 h, after which the segments were rinsed thrice in ice cold saline, centrifuged at 10,OOOg for 10 min and the supernatants removed. Wet weight of the sediment was measured (if structural integrity permitted), followed by drying to a constant weight, as before. The weight of all explanted segments exceeded initial values, due to attached scar tissue. This tissue was not always solubilized by collagenase even after 96 h, as residual weights did not always drop to pre-implant levels. Possible explanations for this inconsistency were explored, as described below. A summary of these all studies, including sample groupings and experimental conditions, is given in Table 11.
1. Removal of scar fissue (Experiment 2) Sixteen Type B segments were explanted after 3 weeks, and maximum scar was dissected away without damaging the samples. They were then exposed to collagenase in two successive, brief intervals (average time: 1 h each; range: 15 to 120 min) at different concentrations (50 and 150 U/mg dry weight of sample just prior to digestion), followed by a third digestion with 50 U/mg dry weight for 96 h. All segments were weighed wet and dry before each exposure to with enzyme; comparisons with (estimated) dry weights prior to implantation yielded dry weight of explanted scar tissue. *Type la, #C9891, Sigma Chemical Co., St. Louis, MO, in a calcium-containing, 0.01 M TRIS buffered, lactated Ringer's solution (pH 7.4).
MEGERMAN ET AL.
298
2. Effect of drying (Experiment 3)
Segments of 2 Type B grafts were divided into two groups, one of which was dried to constant weight. From one graft, paired wet and dry segments were implanted for 3 and 6 weeks, one pair per rat; four pairs served as unimplanted controls. After explantation, all segments were weighed wet and, without drying, were placed in 180 U/mg scar tissue (estimated dry weight, see below) for 1 h. The residues were dried and weighed to determine the effect of this short digestion, and were then further digested by 1900 U/mL for 96 h. 3. Effect of sample configuration (Experiment 4)
A segment of tubular graft can be implanted in a cylindrical (intact ring) or rectangular (cut lengthwise and flattened) configuration. The ring rapidly becomes encased in scar tissue, which may partially shield it from its surroundings, while the rectangle exposes more surface area per unit volume to host enzymes and may thus suffer greater structural degradation by the host. Fourteen segments of a Type B graft, seven of which were cut and flattened, were studied in pairs. Four pairs were implanted for 3 weeks, without prior drying; three pairs were dried and served as unimplanted controls. All segments were exposed to 950 U/mL for 96 h; some were further digested with 50 U/mg dry weight for 30 min. 4. Effect of infection (Experiment 5) The scar tissue that encased several infected 3-week explants appeared to be less readily digested by collagenase. To determine if infection did, in fact, make the scar tissue more resistant to enzymatic degradation, 16 Type B segments were randomly assigned to two groups, one of which was intentionally contaminated with Pseudornonus ueruginosa.+The sterile and septic segments were implanted in separate rats; they were not paired within the same rat to avoid contamination of the sterile implants. After 2 weeks, all explanted segments were septic or sterile, as before. They were dried, weighed, and exposed twice to 200 U/mg current dry weight for 1 h, followed by 96 h in 100 U/mg in 1 mL.
Suture pullout force Two 4/0 silk sutures with tapered needle were passed through the wall of 5mm-long ring segments, 1 mm from each edge. The two sutures were +StrainATCC 27853, McFarland #1 (3
X
lo6 Colony Forming Units per segment).
QUANTITATING COLLAGEN VASCULAR GRAFT STABILITY
299
collinear and parallel to the axis of the ring. One suture was tied to a force transducer (Grass Instruments #FT03) and the other to a linear motion device. The segment was stretched at a rate of 0.3 mm/s until it tore at its weakest point, usually in a circumferential cleavage plane; occasionally the suture cut through the wall without causing a circumferential split. The force at which either event occurred was defined as the suture pullout force. Replicate measurements were obtained by dividing each ring segment lengthwise into two troughs ("half-rings"), and performing a single measurement on each. Since suture pullout measurements are destructive to the sample, they were not performed on segments being implanted. RESULTS
After 3 to 12 weeks in vivo, segments of graft Types A and B demonstrated a profound decrease in suture pullout force (SPF), but only after being digested with collagenase (Fig. 1).This contradicted a similar study of related grafts, wherein SPF decreased significantly without in vif YO digestion (Table I). This discrepancy, and the high variability we observed in SPF and weight loss results, were attributed to the inconsistent dissection of scar tissue from the explanted graft segments. More careful dissection alone did not reduce this variability, so we explored using limited enzymatic digestion to selectively remove scar tissue, before measuring SPF and weight loss. Changes in wet weight following enzymatic digestion were somewhat unreliable; for example, the wet weight of unimplanted (control) segments in-
' 600 0° n
i
5
500 W
8 CY
400 300 -
:200 / I-
1
g
100
I
Q
OC 1
I
0
3
7
12
WEEKS IMPLANTED
Figure 1. Suture pullout force vs. time of subcutaneous implantation in rats for two biological vascular grafts (Type A: circles, Type B: squares), ~) with collagenase for 96 hours (Experibefore (o,.) and after ( 0 ,digestion ment #1 in Table 11). Each point represents mean 2 standard deviation of measurements on two to four 0.5-cm ring segments.
MEGERMAN ET AL.
300
TABLE I Suture Pullout Force in Subcutaneous Implants of Biological Prostheses” Identification
Pre-implant
Expt. 1 (2-12 weeks) M1
Postimplant
Decrease
398 600 969
? k ?
143 (12) 127 (12) 169 (26)
128 f 9 5 ( 9 ) 288 f 141 (11) 609 f 205 (26)
68% 52% 37%
M10 M11 M12
364 269 268
? ? ?
91 (15) 48 (15) 56 (15)
94 f 62(12) 149 f 52 (14) 132 f 48 (15)
74 7% 45% 51%
GO G1 G2
724 2 156 (15) 767 ? 141 (15) 847 ? 153 (15)
134 f 71 ( 6) 550 ? 162 (15) 617 & 175 (15)
81% 28% 27%
M2 G2
Expt. 2 (8 weeks)
“Mean values 2 standard deviation (no. of samples) of maximum force (grams) maintained by 6-0 suture passed 3 mm from the end of 0.8 X 3.0 cm rectangular segments for 2 15 s. (Ref. 26).
variably increased after exposure to collagenase. Consequently, whenever possible, weight loss measurements were based on dry weights. When segments were not dried before implantation, their initial dry weights were estimated from measured wet weights, times an appropriate dry-to-wet weight ratio, derived from dried segments of the same graft. Because this ratio often varied considerably along the grafts’ lengths (Fig. 2), estimates were obtained by interpolation between measured values from closest adjacent dried segments, rather than from a pooled average. The dry weight of undried scar tissue was also estimated from its wet weight (explant minus pre-implant wet .19
0 c 6
LT I-
.17
I
13
I
0
5
10
15
20
25
SEGMENT NUMBER
Figure 2. Longitudinal variation of measured dry-to-wet weight ratio in 3 randomly chosen Type B grafts.
301
QUANTITATING COLLAGEN VASCULAR GRAFT STABILITY
weight of the graft segment) and a regression line (Fig. 3) derived from wet and dry weights of (a) isolated scar tissue measured in a separate study (Table I), and (b) graft-plus-scar explants, in 2 current studies, when corrected for pre-implant wet and dry weights of the graft segments. Figure 4 summarizes the weight changes experienced by ring segments of type B grafts, in 3 experiments. The excess weight of explants, compared to initial dry weights, was assumed to be scar tissue, most of which was removed by 1 to 2 h of digestion with collagenase. Weight loss to significantly below the weight of unimplanted controls did not occur, in most cases, without more extensive (96 h) digestion. The 6-week explants form a notable exception, since digestion for only 1 h caused significantly more degradation than occurred in the 3-week explants. Also noteworthy is the greater resistance to collagenase of scar tissue that formed around the deliberately infected segments; dry weight loss by infected and sterile explants of comparable size was 3.4 2 0.6 vs. 9.4 2 3.4 mg, respectively (mean rt SD, p c .001, unpaired t-test). The infected explants lost somewhat more weight during subsequent enzyme digests, but they ultimately weighed more than did the sterile segments (fractional weight loss: -0.6 2 6.0% vs. 12.2 2 3.7%, p c .OOl). Whether the residues contained any scar tissue or consisted of original implant alone could not be determined. Dehydration had no effect on dry weight loss due to enzyme digestion, p > .05, before or after implantation (Fig. 5). However, undried segments produced greater variability, probably due to errors in estimating initial dry weight. In what follows, results for dried and undried samples, within a
6o
50
1t I
I-
I
3
lo 0
1 0
/
J
50
100
150
200
250
300
350
WET WEIGHT (mg)
Figure 3. Dry weight vs. wet weight of scar tissue, derived from Type B ring segments that were implanted subcutaneously for 2 or 3 weeks (A, 0), and from isolated scar tissue associated with 8-week explants (+;Table I). The pre-implant dry weight of ring segments was estimated using wet weight multiplied by appropriate dry/wet weight ratios (see text); all other wet and dry weights were measured directly. Linear regression line is highly significant ( p < .0001).
MEGERMAN ET AL.
302 15 r
w
I
z -5
+
c3
a
"
I
I
/+
-10 L
1 2 CUMULATIVE ENZYMATIC DIGESTION (hrs)
0
98
Figure 4. Changes in dry weight of unimplanted control (o,O,V) and explanted (all others) ring segments of Type B grafts, after sequential diges3 (o,V), tion steps (mean values only). Segments were implanted for 2 (.,[?), or 6 (A)weeks. Samples intentionally infected with Pseudornonas:[?.For simplicity, abscissa gives average cumulative time of digestion, and dashed line was drawn between pooled average weight change for unimplanted controls. 50
r
40
t
1
WEEKS IMPLANTED Figure 5. Dry weight loss of ring segments that were ( 0 )or were not (0) dried prior to subcutaneous implantation. After being explanted, all segments were treated identically, being exposed to successive collagenase digestion steps of 1 and 96 h. Each point represents mean SD of four segments. In all cases, weight loss was referenced to pre-implant dry weight (estimated for nondried samples as detailed in text). +_
given experiment, have been combined. Configuration had a small but significant effect on dry weight loss after 3 weeks in vivo, followed by in vitro digestion for 96 h: 13.4 2 1.4% (n = 3) vs. 19.8 +- 3.9% (n = 3) for the cylindrical and rectangular samples, respectively ( p = C.05).
QUANTITATING COLLAGEN VASCULAR GRAFT STABILITY
303
Weight loss during enzymatic digestion depended on the relative amounts of tissue and enzyme. This relationship was best described in one of two ways. In most cases, a constant value of collagenase activity per unit weight of substrate yielded a range of enzyme concentrations (U/mL), to which weight loss (mg) per unit time was well correlated. With uniform concentration, however, weight loss per unit initial dry weight (mg/mg or %) was best correlated with a reduced enzyme concentration (U/mL/mg) or specific activity (U/mg in 1 mL). Figure 6 gives examples of these relationships. To compare results obtained after different digestion times, weight loss was plotted against "total enzyme action'; defined here as enzyme concentration X time. Figure 7 displays weight loss by segments of grafts B3, B5, and B7, during sequential digestion steps. The rate of digestion, proportional to the slope of weight loss vs. total enzyme action, decreased with each step, commensurate with the progressive loss of more easily digested scar tissue. To adjust for differences in enzyme concentration, activity, and/or exposure time, each graft segment's capacity to resist biodegradation was expressed as
/
01
0
0
I
I
I
I
I
1
2
3
4
5
I
6
7
CONCENTRATION 6o
r
0
50
100
150
200
250
SPECIFIC ACTIVITY
Figure 6. Dry weight loss by graft segments as a function of enzyme concentration (Units/mL, x W3)or specific activity (Units/mg dry weight, in 1 mL). Top: subcutaneous explants of Type B grafts following a 1 h exposure to collagenase (symbols as in Fig. 4). Bottom: unimplanted Type A graft after collagenase digestion for 96 h.
MEGERMAN ET AL.
304 20
-
STEP 1
n
m 15
I 1
E
W
v) v)
9
10
STEP 3
I-
I
52
s =
STEP 2
0
0
10
5
//
100
150
200
250
TOTAL ENZYME ACTION
Figure 7. Dry weight loss vs. total enzyme action, expressed as the product of concentration and time (1,000 Units/mL x h), for successive enzyme digestion steps. Steps 1 and 2 correspond to the 1-h exposures (average) to collagenase; the 96-h exposure is uniformly identified as step 3. Data points show mean 2 SD of x and y coordinates; regression lines based on mean values are drawn to include the origin, and exclude "septic" group for step 1.Symbols are as given in Figure 4.
weight loss divided by the cumulative sum of total enzyme action. This parameter, for all grafts tested, is plotted as a function of implantation time in Figure 8. Just 3 weeks in vivo had a significant effect on the susceptibility of graft segments to in vitro degradation. This result correlates well with the
/
/@
40 20 -
L
I
3 6 9 12 WEEKS IMPLANTED Figure 8. Dry weight loss, referenced to pre-implant dry weight and enzyme concentration (% loss per Unit-hour in 1 mL), vs. time of implantation, in Type A (0), C (0),and B (all others) graft segments after completing all digestion steps. Paired wet and dry segments have been combined (see Fig. 6 ) ; data points represent mean values for each experiment, including infected samples (0).
0
QUANTITATING COLLAGEN VASCULAR GRAFT STABILITY
305
previously observed decrease in suture pullout force following collagenase digestion of the explanted segments (Fig. 1). In all 2- and 3-week explants, in d u o weight loss after 1 h was less than the estimated weight of scar tissue. Assuming this weight loss reflects the digestion of scar tissue alone, and none of the graft, one can estimate how much enzyme will remove this scar tissue entirely. Figure 9 shows the fractional loss of scar tissue vs. enzyme specific activity, referenced to the dry weight of explanted scar tissue. Omitting the infected segments, these coarsely estimated values are surprisingly well correlated (u = 0.89, p c .02); they suggest that, in 1 h, approximately 10,000 collagenase U/mg will solubilize all of the scar tissue in these closely dissected 3-week explants, which weighed up to 25 mg (dry weight). Even without removing all scar tissue, however, a 1-h digest significantly reduced their mechanical strength compared to that of unimplanted controls, as measured by suture pullout force (Fig. 10). This result seemed independent of graft type.
DISCUSSION
Biologically derived grafts will continue to be an attractive conduit for arterial bypass, when no vein graft is available, for a number of reasons. These grafts tend to have a less thrombogenic surface, more desirably compliant wall, and a lower rate of infection, when compared to currently available synthetic materials.',35Unfortunately, many of these biologic grafts tend to form aneurysms, which often are detected only after several years.'-" Nevertheless, many have not failed, and continue to function well even after
0 0
2
4
6
8 1 0 1 2 1 4
SPECIFIC ACTIVITY
x 1o - ~
Figure 9. Estimated fractional loss of scar tissue (dry weight) in 2- and 3-week explants of Type B segments vs. enzyme specific activity (Unithours/mg dried scar in 1 mL; mean ? SD for each experiment; infected samples: 0).
MEGERMAN ET AL.
306
0
P96
DIGESTION TIME (hours) Figure 10. Fractional change in suture pullout force, following enzymatic degradation of unimplanted controls (0) and 3-week explants ( 0 )for up to 96 h. Statistical significance after 1 h ( p < .03, explant results compared to unity) and 96 h ( p c .005) was determined using 2-tailed t-tests. Points indicate mean & S.E.M of measurements of all graft types.
10 year^.^^,^^ Are these successes due to the chance lack of defects in grafts produced by an inherently sound chemical process, or to their use in patients who, also by chance, are unable to degrade any foreign biologic implant, however poorly processed? Two observations suggest that patient variability is less important than is consistency of biologic graft processing: (a) water content can vary considerably along the length of these biological grafts (Fig. 2); and (b) a composite ilioperoneal bypass fashioned from four Type Alike grafts in tandem failed within 2 years, but with marked dilatation in only one of the four segments.26At a minimum, therefore, tests of quality control must be improved. The effective screening of biologic vascular prostheses requires an assay that permits frequent sampling to account for biological variability, and rapidly measures the graft's resistance to biodegradation over a projected period of several years, while consuming only a small fraction of the graft. In all respects, a test based on chronic intra-arterial implants is unacceptable: clinically relevant sizes (3-6 mm I.D.) require the use of large mammals, a costly and difficult option; detecting structural failure in the animal implant may take as long as in man; and such testing consumes too much of the graft. Therefore, biologic grafts must be evaluated by alternative means. Ideally, grafts should be tested nondestructively in vitro, using means that do not preclude their subsequent use in patients. Some in vitro tests of small graft samples, which measure elasticity, heat shrink temperature, and resistance to chemical or enzymatic degradation, have been devised to characterize collagen stability in crosslinking experiment^.'^^'^-^^"^ It is not clear, however, that in vitro testing adequately represents in v i m degradation due to the
QUANTITATING COLLAGEN VASCULAR GRAFT STABILITY
307
host’s biological responses, including inflammation and possible immunological attack. Therefore, numerous in vivo models have been used to evaluate collagen bioprostheses, including the rat In most cases, including the study summarized in Table I, weight loss and mechanical measurements were easily performed, and assumed to be independent of scar tissue explanted together with the retrieved segments. Our own results, however, were critically sensitive to the amount of attached scar tissue, especially when using small segments of graft. In all cases, the segments weighed more when explanted than before they were implanted, and despite all attempts to surgically remove this scar tissue, their mechanical strength (by suture pullout force) was unchanged before exposure to collagenase. This strengthening effect of host-generated (scar) tissue on biomaterials, which has previously been observed with cellulose spongesz5and vascular anastomoses?*could make weight loss and mechanical changes following the simple removal of subcutaneous implants too unreliable for use in graft evaluation or quality control. We, therefore, explored using limited collagenase digestion to selectively remove scar tissue from the explants. This objective requires that scar and graft collagen have very different susceptibilities to in vitro degradation by collagenase, which is probably the case, judging from the decreasing rate of weight loss in successive digestion steps (Fig. 7). It cannot be assumed, however, that a l-h digestion removes scar tissue alone, nor that it removes it entirely. This was particularly evident in 6-week explants of graft B5 (Table 11, Fig. 4),where digestion with only 180 U/mg scar tissue (908 k 430 U/mL) for 1 h was sufficient to reduce weights to below pre-implant levels. Whether this reflected less scar tissue attached to the explants (subjectively, they seemed to be more easily dissected from their subcutaneous pouches), or increased degradation during the longer implant period is not known. In contrast, scar tissue attached to the deliberately infected %-weekexplants was much more resistant to a greater concentration of collagenase (200 U/mg total explant; 6009 2 584 U/mL); this seems consistent with the increased resistance to degradation of type 111 collagen associated with rheumatoid and inflamed synovial membranes39and inflammation due to glutaraldehydetreated sponges.4o Our results suggest loss of enzyme effectiveness during a lengthy digestion of scar tissue. Explants treated with up to 950 U/mL for 96 h, often weighed more than before implantation, while, their further exposure to only 50 U/mg dry weight (444 f 68 U/mL) for 30 min produced a significant additional weight loss. A similar loss of potency was observed when a single aliquot of collagenase was used to digest successive segments of fresh bovine artery (data not shown). To avoid this problem, separate aliquots of collagenase should be used to digest scar tissue vs. the graft segment. From Figure 9, 10,000 U/mg dried scar tissue for 1 h, or approximately 1,500 U/mg wet scar tissue will effectively remove the host tissue from graft segments explanted after 3 weeks, enabling their further evaluation by in nitro tests. Corresponding figures for longer term implants may be lower (scar collagen
Description
Preliminary comparisons
Expt
1
c2
c1 0 3
0 3
W,D W
W W W,D
-
W,D
-
W,D
-
950 U/mL (96) 950 U/mL (96)
950 U/mL (96) 950 U/mL (96)
950 U/mL (96) 950 U/mL (96)
0 3,7
82
W W
3800 U/mL (96) 950 U/mL (96)
(W, some D) (W, some D)
Step 1
037 12
~
B1
Wt,, 950-3800 U/mL (96) 3800 U/mL (96) 950 U/mL (96)
~~
(W, some D) (W, some D) (W, some D)
Wto
0 3 7,12
~
50 U/D, (.5)
-
50 U/D1 (.5)
-
50 U/Dl (.5)
-
Step 2
Step 3
Concentration of Bacterial Collagenase
A
Graft
Weeks in vivo
TABLE I1 Summary of Experimental Conditions Studied”
86 B7
Configuration
Infection
5
W,D W
W,D W
180 U/Wo (1) 180 U/D,,, (1) 180 U/D,,, (1)
-
-
W,D
-
W,D
200 U/Do (1) 200 U/D, (1)
950 U/mL (96) 950 U/mL (96)
50 U/Do (96)
-
W W,D
50 U/Do (
