Abstract
A prospective, randomized clinical trial was conducted concurrently at eighteen medical centers in order to compare the safety and efficacy of two types of graft material for the treatment of fractures of long bones: autogenous bone graft obtained from the iliac crest, and a composite material composed of purified bovine collagen, a biphasic calcium-phosphate ceramic, and autogenous marrow. Two hundred and thirteen patients (249 fractures) were followed for a minimum of twenty-four months to monitor healing and the occurrence of complications.We observed no significant differences between the two treatment groups with respect to rates of union (p = 0.94, power = 88 per cent) or functional measures (use of analgesics, pain with activities of daily living, and impairment in activities of daily living; p > 0.10). The prevalence of complications did not differ between the treatment groups except for the rate of infection, which was higher in the patients who were managed with an autogenous graft. Twelve patients who were managed with a synthetic graft had a positive antibody titer to bovine collagen; seven of them agreed to have intradermal challenge with bovine collagen. One patient had a positive skin response to the challenge but had no complications with regard to healing of the fracture.We concluded that, for traumatic defects of long bones that necessitate grafting, use of the composite graft material appears to be justified on the grounds of safety, efficacy, and elimination of the increased operative time and risk involved in obtaining an autogenous graft from the iliac crest.
The superiority of autogenous bone as a graft material for the filling of traumatic defects or the repair of segmental losses of bone is widely recognized. However, autogenous bone graft has certain limitations. If the need for a graft is determined intraoperatively, the patient may be draped and positioned unsuitably for obtaining the graft. Occasionally, the autogenous bone available for grafting is of poor quality (osteoporotic) or not enough is available. Moreover, obtaining an autogenous graft may entail serious morbidity at the donor site and an increase in the operative time of twenty minutes or more9,13,15. In order to circumvent these problems, we sought an effective material to use as a bone-graft substitute.
The graft material evaluated in this study as an alternative to autogenous bone graft is a composite material consisting of type-I collagen, a biphasic ceramic (hydroxyapatite and tricalcium phosphate), and marrow obtained from the patient by aspiration, usually from the iliac crest or, alternatively, from the site of the fracture. The material has been characterized in detail, and the rationale underlying its development has been reported previously3,10,11. Preclinical trials of the material demonstrated its effectiveness in bridging segmental defects of long bones in rats and dogs1,5,6,8,11,14.
The current study was performed in order to evaluate the efficacy and safety of the collagen-calcium phosphate ceramic graft material, as compared with those of autogenous graft, for the treatment of acute fractures of long bones in humans. The design of the study was intended to fulfill the requirements of the Food and Drug Administration for approval of the collagen-ceramic graft material for clinical use.
†The following investigators and institutions participated in the study: Wayne Akeson, M.D., University of California at San Diego, San Diego, California; Alfred Behrens, M.D., St. Paul-Ramsey Medical Center, St. Paul, Minnesota; Bruce Browner, M.D., Hermann Hospital, Houston, Texas; Allan Bucknell, M.D., Brooke Army Medical Center, Fort Sam Houston, Texas; Ramon Gustilo, M.D., and David Templeman, M.D., Hennepin County Medical Center, Minneapolis, Minnesota; James Hughes, M.D., University of Mississippi Medical Center, Jackson, Mississippi; John D. Lubahn, M.D., Hamot Medical Center, Erie, Pennsylvania; D. Allan MacKenzie, M.D., and Milton Mudge, M.D., San Bernardino County Medical Center, San Bernardino, California; Theodore Papademetriou, M.D., Erie County Medical Center, Erie, Pennsylvania; Ramond Pierce, M.D., William Wishard Memorial Hospital, Indianapolis, Indiana; Thomas A. Russell, M.D., Regional Medical Center, Memphis, Tennessee; David Seligson, M.D., and Stephen Henry, M.D., Humana Hospital, University of Louisville, Louisville, Kentucky; Harold Shuster, M.D., Germantown Hospital, Philadelphia, Pennsylvania; John Siliski, M.D., Massachusetts General Hospital, Boston, Massachusetts; and Phillip Spiegel, M.D., and David Helfet, M.D., Tampa General Hospital, Tampa, Florida.
‡Department of Orthopaedic Surgery, University of California, Davis Medical Center, 2230 Stockton Boulevard, Sacramento, California 95817.
§University of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Boulevard, Dallas, Texas 75235.
¶The Hospital for Special Surgery, 535 East 70th Street, New York, N.Y. 10021.
Graft Material
The collagen-calcium phosphate ceramic graft material1 (Collagraft bone-graft matrix; Zimmer, Warsaw, Indiana) is supplied to the surgeon as a sterile kit containing two separate components: five grams of ceramic and five milliliters of collagen (a suspension of sixty-five milligrams per milliliter). The contents of the kit are mixed aseptically at the time of the operation with two milliliters of the patient's own bone marrow, aspirated from the iliac crest or from the site of the fracture. Typically, one kit is sufficient for repair of a fracture of the forearm, and two kits are sufficient for repair of a fracture of the tibial shaft. For large femoral defects, as many as four kits are used.
The collagen component of the graft material (Collagen Corporation, Palo Alto, California) is a highly purified bovine dermal collagen consisting mostly of type-I collagen (more than 95 per cent) with a small amount of type-III collagen (less than 5 per cent). The ceramic component (Zimmer) is a biphasic mixture of 65 per cent hydroxyapatite and 35 per cent ß-tricalcium phosphate. The ceramic particles are formed, with use of a sintering process, into irregularly shaped granules of 0.5 to one millimeter in diameter.
The autogenous bone graft was obtained from the ilium, typically from the posterior iliac crest. Although cancellous autogenous graft was specified in the protocol, the addition of morselized cortical bone by the surgeon was not precluded. Inclusion of the thin cortical shell of the ilium is unlikely to have affected the performance of the graft.
Design of the Study
The study was designed as a prospective, randomized clinical trial. The independent variable of interest was the graft material—autogenous bone graft or collagen-calcium phosphate ceramic. The dependent variables of interest were union of the fracture; the functional outcome as measured on the basis of the use of analgesics, pain with activities of daily living, and impairment in activities of daily living; and the rate of complications. Power calculations (based on assumptions of a 5 per cent rate of failure for the autogenous grafts, no more than a 20 per cent rate of failure for the experimental grafts, a 15 per cent loss of patients to follow-up, an alpha of 0.05, and a power of 90 per cent) indicated that an initial sample of 300 fractures was required; in order to achieve this sample size within a reasonable time-period, the study was carried out concurrently at multiple centers. The study was approved by the Institutional Review Board at each center, and informed consent was obtained from each subject.
Selection of Patients and Assignment of Treatment
The criteria for inclusion in the study were an age between eighteen and seventy years and a traumatic fracture of the ulna, radius, humerus, tibia, or femur for which the planned treatment included both bone-grafting and either internal or external fixation. The criteria for bone-grafting were not formally defined prospectively; the need for a graft was based on the judgment of the operating surgeon. The criteria for exclusion were an interval of more than thirty days after the injury; osteomyelitis in the involved limb; concurrent use of corticosteroids or immunosuppressive agents; a pathological fracture; pre-existing severe vascular disease; pregnancy; coexisting degenerative bone disease; a weight of more than 250 pounds (113 kilograms) because these patients were considered to have a high risk of failure of the fixation; an expectation of non-compliance because of alcoholism or mental illness; unavailability for the full two-year follow-up because of a lack of a local address; a fracture necessitating more than thirty cubic centimeters of graft material; and an open fracture with a type-IIIB or type-IIIC wound, according to the classification of Gustilo and Anderson, unless the wound was reclassified to a lower grade (on the basis of a more thorough assessment during débridement in the operating room) before grafting and within thirty days after the injury. Both the surgeon and the patient were blinded with respect to the assignment of the treatment until the criteria for inclusion were determined to have been met and the patient had signed an informed-consent document and had been formally enrolled in the study.
Each center used a computer-generated randomization schedule incorporated in two sets of consecutively numbered sealed envelopes: one set for weight-bearing bones (the tibia and femur) and the other for non-weight-bearing bones (the humerus, radius, and ulna). Although the randomization process required classification of the fractures according to whether they involved weight-bearing or non-weight-bearing bones, the treatment was intended to be randomized by patient. Thus, if one patient had multiple fractures, all of the fractures were treated with the same type of graft material. Early in the study, a separate randomization envelope was mistakenly opened for fractures of the upper and lower extremities for several patients, eight of whom were managed with different graft materials in the upper and lower extremities. Because the randomization was blinded, there is no reason to believe that this deviation from the protocol introduced any bias into the study, and these patients were not excluded (except from demographic analyses, for which they would have been in both groups).
Documentation and Follow-up Examinations
Records were kept for each patient independent of the patient's regular hospital charts. At the time of enrollment, a profile of the patient was recorded, and at the time of the operation, the details of the procedure were recorded. The patients were subsequently evaluated in the clinic at one and one-half, three, six, twelve, eighteen, and twenty-four months after the operation. The patients received modest financial compensation for each follow-up visit after six months; this was considered essential to minimize the loss of subjects to follow-up. Annual follow-up was continued after twenty-four months to detect any long-term complications.
The clinical assessment included physical examination, two orthogonal radiographs of the site of the fracture, assessment of pain, and assessment of function in activities of daily living. Functional impairment was scored by the physician with use of a list of eleven typical activities of daily living involving use of the upper extremities (washing the face and the hair, brushing the teeth, tying the shoes, putting on a belt, wringing out a washcloth, throwing a ball, making a bed, reaching a high shelf, pushing open an emergency exit or a revolving door, tending to personal hygiene, and lifting twenty pounds [nine kilograms]) and a list of ten activities involving use of the lower extremities (walking, running, arising from a chair, stair-climbing, putting on pants while standing, getting in and out of an automobile, arising from lying on the floor, sexual function, picking up items from the floor, and lifting twenty pounds [nine kilograms]). If the patient could carry out all activities with the injured limb, impairment was rated as none; if the patient was unable to perform one, two, or three activities, impairment was considered slight; if the patient was unable to perform four to seven activities, impairment was rated as moderate; and if the patient was unable to perform eight activities or more, impairment was rated as severe. Any complications, reoperations, or failures of the hardware were recorded. The surgeon's evaluation of radiographic healing was conducted with use of a 16-point ordinal scale; however, for the purposes of this report the radiographic analysis is simplified to a determination of healed or not healed.
Because the surgeon might have been biased in the radiographic assessment by clinical knowledge of the fracture, the radiographs also were reviewed independently by a radiologist who was blinded with respect to the treatment and the clinical outcome. The radiographs for each patient were presented in random order without indication of the time-interval since the operation, and the radiologist assessed them according to whether they showed the fracture to be healed or not healed. It should be noted that, although the radiologist was blinded with respect to the treatment, the type of treatment could be determined from many radiographs because the radiopaque ceramic particles of the collagen-ceramic graft were visible.
Serum samples were obtained from all patients preoperatively and postoperatively at one and one-half, three, and twelve months to screen for antibodies to bovine collagen. Patients who demonstrated such antibodies also were evaluated for antibodies to type-I and type-III human collagen. In addition to serological testing, skin-challenge testing was performed after the twelve-month visit for patients who had a positive antibody titer to bovine collagen and for a matched control group of patients who had been managed with a synthetic graft but did not have such a titer. The patients who had a positive titer and the controls were matched for age, gender, location of the fracture, and volume of the graft material used.
Statistical Analysis
Because of the discrete nature of the follow-up variables, non-parametric statistical methods were used. The Wilcoxon rank-sum, Fisher exact, chi-square, and Mantel-Haenszel tests were used when appropriate. A p value of 0.05 or less was accepted as demonstrating significance.
Study Population
Eighteen centers participated concurrently in the trial. Three hundred and twenty-five patients (374 fractures) were randomized to treatment between September 1986 and December 1989. Of these 325 patients, twenty-two (twenty-two fractures) did not complete enrollment for one of several reasons: an intraoperative decision not to perform bone-grafting (five patients who had been scheduled to receive a collagen-ceramic graft and six who had been scheduled to receive an autogenous graft), a decision to treat the fracture non-operatively (one patient in each group), a soft-tissue problem that delayed the procedure for more than thirty days (two patients in each group), a fever that delayed the procedure for more than thirty days (one patient in each group), a positive pregnancy test (one patient who had been scheduled to receive an autogenous graft), death before the procedure could be done (one patient who had been scheduled to receive a collagen-ceramic graft), and failure of the surgeon to obtain consent from the patient (one patient who had been scheduled to receive a collagen-ceramic graft). This left 303 patients (352 fractures) in the study. It was found that the protocol had been violated for six of these patients (seven fractures), and therefore the six patients were excluded from the study. These six included one patient who was quadriplegic and one who was paraplegic before the index injury, one who had a fracture that was treated with both a synthetic graft and an autogenous graft, one who removed the fixation device and bore weight on the injured extremity soon after the operation, and two who had a type-IIIB or type-IIIC open fracture wound7. This left 297 patients (345 fractures) in the study. For the analyses of outcome, all fractures were considered independently.
Excluding the eight patients who received one type of graft for a fracture of an upper extremity and the other type for a fracture of a lower extremity, there were 199 men (69 per cent) and ninety women (31 per cent). The average age of the men who were managed with a collagen-ceramic graft was thirty-seven years, compared with thirty-two years for those who were managed with an autogenous graft; this difference was significant (p = 0.009). The average age of the women who were managed with a collagen-ceramic graft was thirty-eight years, compared with forty-six years for those who were managed with an autogenous graft; this difference also was significant (p = 0.018). The average age of the patients did not differ substantially between the groups (37.0 years for the patients who received a collagen-ceramic graft, compared with 36.5 years for those who received an autogenous graft).
During the course of the study, there was a gradual attrition in the number of patients who were enrolled, so that for each subsequent follow-up interval the sample size was slightly smaller. The reasons for the attrition (reported as the number of fractures) included death (five fractures that had been treated with an autogenous graft and five that had been treated with a collagen-ceramic graft), revision after an unsuccessful outcome (fifteen fractures that had been treated with an autogenous graft and six that had been treated with a collagen-ceramic graft), loss of the patient to follow-up (fifteen fractures that had been treated with an autogenous graft and fourteen that had been treated with a collagen-ceramic graft), a missed visit to the clinic but not a loss to follow-up (eleven fractures that had been treated with an autogenous graft and fourteen that had been treated with a collagen-ceramic graft), and voluntary withdrawal from the study (six fractures that had been treated with an autogenous graft and five that had been treated with a collagen-ceramic graft). By the time of the twenty-four-month evaluation, 213 patients (249 fractures [117 that had been treated with an autogenous graft and 132 that had been treated with a collagen-ceramic graft]) remained in the study. This resulted in a statistical power of 88 per cent for the parameters specified in the original calculation of sample size. Cumulative attrition of fractures from the original complement of 345 included fifty-two that had been treated with an autogenous graft and forty-four that had been treated with a collagen-ceramic graft. Variations in the denominators describing the results reflect the attrition just mentioned as well as occasional failure of the surgeon to record a datum or failure of the patient to answer a question.
Characteristics of the Fractures
The characteristics of the original 345 fractures (location, type, whether they were metaphyseal or diaphyseal, whether they were open or closed, and degree of comminution) were not significantly different between the two groups. Of the 176 fractures that were treated with a collagen-ceramic graft, thirty-seven were located in the femur; nine, in the humerus; thirty, in the radius; sixty-five, in the tibia; and thirty-five, in the ulna. Of the 169 fractures that were treated with an autogenous graft, thirty-five were located in the femur; seventeen, in the humerus; thirty, in the radius; sixty-one, in the tibia; and twenty-six, in the ulna. Of 171 fractures that were treated with a collagen-ceramic graft (five fractures were not classified), ninety-eight (57 per cent) were strictly diaphyseal, compared with 103 (64 per cent) of 161 that were treated with an autogenous graft (eight fractures were not classified); the remaining fractures in each group were metaphyseal, with or without a diaphyseal extension. Fifty-five (31 per cent) of the 176 fractures that were treated with a collagen-ceramic graft were open, as were forty-seven (28 per cent) of the 169 that were treated with an autogenous graft.
The volume of the graft ranged from one to thirty cubic centimeters. Grafting was typically performed at the time of internal or external fixation. One hundred and thirty (74 per cent) of the 176 fractures that were treated with a collagen-ceramic graft and 122 (72 per cent) of the 169 that were treated with an autogenous graft were described by the surgeon as comminuted.
Of the 176 collagen-ceramic grafts, 112 (64 per cent) were made with bone marrow aspirated from the iliac crest, and sixty-four (36 per cent), from bone marrow aspirated from the site of the fracture. Two milliliters of bone marrow is required for each kit. One to four kits were used for each fracture, with 167 (95 per cent) of the 176 fractures requiring one or two kits. For the 169 fractures that were treated with an autogenous graft, the graft was typically obtained from the posterior iliac crest or, in some instances, from the anterior iliac crest.
Fixation
The types of fixation devices that were used in the initial operation did not differ significantly between the groups (p = 0.60). The principal devices used for the 176 fractures that were treated with a collagen-ceramic graft were plates and screws (143 fractures), an intramedullary nail (thirteen fractures), an intramedullary nail and a plate (one fracture), a wire (one fracture), and an external fixator (eighteen fractures). The principal devices used for the 169 fractures that were treated with an autogenous graft were plates and screws (147 fractures), an intramedullary nail (nine fractures), an intramedullary nail and a plate (one fracture), and an external fixator (twelve fractures).
Duration of the Operation
The average duration of the operation was 133 minutes for the patients who received a collagen-ceramic graft and 153 minutes for those who received an autogenous graft (p = 0.053, Wilcoxon signed-rank test).
Union of the Fracture
The progression of healing as seen radiographically was similar in the two treatment groups. Early formation of callus and new-bone formation at the site of the graft were followed by reconstitution of the cortex and eventual obliteration of the fracture line. Gradual disappearance of ceramic particles that had been visible on early radiographs of fractures treated with a collagen-ceramic graft suggested either resorption or incorporation of particles into new bone.
Union of the fracture was defined as the first radiographic evidence of healing as evaluated by the surgeon and was quantified as a score of 9 points or more on the 16-point radiographic scale. At six (p = 0.94), twelve (p = 0.80), eighteen (p = 0.67), and twenty-four months (p = 0.89) postoperatively, the rates of union were statistically equivalent for the two graft materials (Table I). By six months, 138 (91 per cent) of 152 fractures that had been treated with a collagen-ceramic graft had united radiographically, compared with 122 (92 per cent) of 133 fractures that had been treated with an autogenous graft. By one year after the injury, 140 (97 per cent) of 144 fractures that had been treated with a collagen-ceramic graft had united radiographically, compared with 116 (99 per cent) of 117 that had been treated with an autogenous graft. The rates of union also were equivalent between the two groups when the data were stratified according to weight-bearing as compared with non-weight-bearing bones.
The determination of healing made by the independent radiologist who was blinded with respect to the type of treatment, the interval since the operation, and clinical function was considerably more conservative than the determination made by the surgeon (Table I). Nevertheless, the radiologist's determinations of healing were also statistically equivalent between the two groups at all time-intervals (p > 0.10).
Clinical Assessment
Wound-healing was rapid in both treatment groups and was not significantly different at any time-point. At three months postoperatively, there was healing of the wound in association with 137 (95 per cent) of 144 fractures that had been treated with a collagen-ceramic graft and in association with 131 (94 per cent) of 139 that had been treated with an autogenous graft (p = 0.93). At two years, all of the wounds had healed.
The absence of pain at the site of the fracture during activities of daily living did not differ significantly between the group treated with a collagen-ceramic graft and that treated with an autogenous graft (104 [71 per cent] of 146 at twelve months and ninety-three [71 per cent] of 131 at twenty-four months, compared with eighty [66 per cent] of 121 and eighty-one [70 per cent] of 116 at the same time-intervals; p = 0.44 and 0.95, respectively). The lack of use of analgesics also did not differ significantly between the group treated with a collagen-ceramic graft and that treated with an autogenous graft (133 [91 per cent] of 146 at twelve months and 118 [90 per cent] of 131 at twenty-four months, compared with 106 [88 per cent] of 121 and 108 [93 per cent] of 116 at the same time-intervals; p = 0.47 and 0.53, respectively). The absence of impairment in activities of daily living was similar for the group treated with a collagen-ceramic graft and that treated with an autogenous graft (ninety-seven [66 per cent] of 146 at twelve months and 105 [80 per cent] of 131 at twenty-four months, compared with seventy-two [60 per cent] of 121 and eighty-five [73 per cent] of 116 at the same time-intervals; p = 0.30 and 0.26, respectively). Approximately one-third of the patients in each group had pain at the site of the fracture and impairment in activities of daily living one year after the injury; these rates improved only slightly by two years after the injury. It is notable that, of 147 fractures that had been treated with an autogenous graft, forty-six (31 per cent) were associated with pain at the donor site six weeks after the operation. At two years posteroperatively, six (5 per cent) of 114 fractures that had been treated with an autogenous graft continued to be associated with residual pain at the donor site.
Complications
There was no significant difference between the treatment groups with respect to the prevalence of complications (p = 0.33), except for infection (defined as local signs of infection and a positive culture). The rate of infection in the group that had been managed with an autogenous graft was higher than that in the group that had received a collagen-ceramic graft (Table II); this difference was significant regardless of whether two infections that occurred at the iliac-crest donor site in the former group were included (p = 0.003) or excluded (p = 0.008). Of the nine infections at the sites of fractures that had been treated with a collagen-ceramic graft, six resolved after irrigation, débridement, and administration of antibiotics and three necessitated extensive débridement and removal of the graft material. Of the twenty-two infections at the sites of fractures that had been treated with an autogenous graft, five resolved after administration of antibiotics alone whereas seventeen necessitated operative intervention. Of the two infections that occurred at the iliac-crest donor site, one resolved after administration of antibiotics alone and the other resolved after incision and drainage in addition to administration of antibiotics.
There were seven refractures. One, in a patient who had been managed with a collagen-ceramic graft, was iatrogenic, occurring during removal of a plate from the radius eighteen months postoperatively. The remaining six refractures were the result of new, low-energy trauma: four occurred within five months after the original injury, when healing was incomplete, and two (one in each treatment group) occurred three years after the original fracture, through the site of the original fracture.
Immunological Findings
An analysis was performed on serum samples from 236 patients (128 who had been managed with a collagen-ceramic graft and 108 who had been managed with an autogenous graft) from the initial group of 297. The other sixty-one patients did not have this analysis because serum samples had not been obtained at the appropriate intervals. Of the 236 patients who were thus analyzed, thirteen (sixteen fractures) demonstrated a positive antibody titer to bovine collagen. Twelve of these patients had been managed with a collagen-ceramic graft, and one had been managed with an autogenous graft. None of the thirteen patients had measurable antibodies to human type-I or type-III collagen.
A sample of thirteen patients who did not exhibit a positive antibody titer to bovine collagen was identified to serve as matched controls for the thirteen patients who had a positive antibody titer. Of these twenty-six patients, fifteen (seven who had a positive antibody titer and eight who had a negative [control] titer) agreed to have intradermal challenge with bovine collagen. As expected4, none of the patients who had a negative titer demonstrated a skin response. One of the seven patients who had a positive titer demonstrated a positive reaction to the skin challenge; this patient did not have any complications related to fracture-healing.
Our findings indicate that a collagen-ceramic-marrow composite graft material is as effective and as safe as autogenous bone graft in the treatment of fracture defects of long bones. The validity of this inference depends on an absence of bias in the design and execution of the study. We attempted to eliminate selection bias by carrying out a randomized, prospective, controlled trial in which both the surgeon and the patient were blinded with respect to the treatment until after the patient had been enrolled. By carrying out the trial concurrently at eighteen centers, the possible effects of an undetected bias in treatment or in assessment at any one center were minimized. The rate of fractures lost to follow-up before the twenty-four-month visit to the clinic was similar in the two groups (fourteen [8 per cent] of the 176 that were treated with a collagen-ceramic graft as compared with fifteen [9 per cent] of the 169 that were managed with an autogenous graft); we have no reason to suspect a treatment-related bias with respect to the outcomes in these two self-selected groups of dropouts.
Our sample comprised a heterogeneous collection of injuries, operative techniques, and postoperative regimens. The rationale for designing such a broadly based study was to represent the full spectrum of bone-grafting practices currently in use for acute fractures. As expected when subjects are randomized, the diversity was similar in the two treatment groups with respect to the demographic features, the characteristics of the fracture, the operation, and the postoperative care. Therefore, comparison of the two treatment groups is reasonable despite their heterogeneity.
Ideally, an additional, concurrent control group of fractures that were not treated with grafting should have been included in the study. It is possible that a number of these fractures would have healed satisfactorily in the absence of any graft; insofar as this might have been the case, the current study fails to demonstrate the effectiveness of either graft material. However, there is no ethical way to design a clinical study that includes the failure to graft as a treatment option. We relied on the judgment of the surgeons to determine the necessity for a graft. Although we did not prospectively define the requirement for a graft, a post hoc definition of the indication for grafting in this study was a clinically important volumetric defect of one cubic centimeter or more, due to comminution or loss of a fragment, after reduction of the fracture.
We noted a considerable discrepancy between the surgeons and the blinded radiologist with respect to the evaluation of fracture-healing. Clearly, a surgeon, whose interpretation of a radiograph is tempered by knowledge of the clinical course and the findings of physical examination, is more likely to consider a fracture healed than is a radiologist. Nonetheless, this difference appears to be systematic and not influenced by the type of treatment, and we do not consider it to be a source of bias; both the surgeons and the radiologist found the two groups to have essentially the same rates of healing.
The complications that we noted were what one might expect for a large series of fractures of this type. We have no cogent explanation for the high rate of infection in the patients who were managed with an autogenous graft. It is noteworthy that the soft-tissue injury was rated as severe in association with twenty-one (12 per cent) of the 169 fractures that were treated with an autogenous graft as compared with fifteen (9 per cent) of the 176 that were treated with a collagen-ceramic graft; this difference, although not significant, might have contributed to a higher rate of infection in association with the fractures that were treated with an autogenous graft. Nine infections in patients who had been managed with an autogenous graft were associated with a delay in fracture-healing.
The development of two infections at the iliac-crest donor site, the two instances of drainage of the wound at the donor site resulting from a hematoma and dehiscence, and the persistence of pain at the donor site at twenty-four months postoperatively in six patients who had been managed with an autogenous graft underscore the potential morbidity of obtaining a bone graft from the iliac crest9,13,15. Indeed, avoidance of this morbidity is one of the attractions of use of a synthetic graft and provides some justification for its expense.
The refracture that occurred through the original fracture site three years after the index injury in a patient who had received a collagen-ceramic graft raises the suspicion that the graft in this patient may have contributed to a persistent locus of weakness. A similar refracture occurred in a patient who had been managed with an autogenous graft.
The rarity of an immune response to the bovine collagen, and the absence of complications associated with such a response, are consistent with the experience of many years of use of bovine type-I collagen in operative and dermatological applications4. As in those applications, the occurrence of an immune reaction was not in itself predictive of a poor outcome in the current series.
The exact clinical role of a synthetic bone-graft material such as the one that we studied remains to be defined. Autogenous graft has three overlapping clinical roles: it can provide immediate structural support, it can provide an osteoconductive scaffold for the filling of a defect, and it can provide an osteogenic stimulus from both cells and growth factors. The synthetic graft material that we studied is unable to fulfill the first of these roles (immediate structural support); however, it functions well as an osteoconductive gap-filling material, and the addition of the patient's marrow to the composite graft provides an osteogenic stimulus2,12. For patients who have an acute traumatic defect of a long bone, use of the collagen-ceramic graft material instead of an autogenous graft appears to be justified on the grounds of safety, efficacy, and elimination of the increased operative time and risk involved in obtaining a graft from the iliac crest.
Black, R. J.; Zardiackas, L. D.; Teasdall, R.; and Hughes, J. L., Jr.: The mechanical integrity of healed diaphyseal bone defects grafted with calcium hydroxyapatite/calcium triphosphate ceramic in a new animal model. Clin. Mater.,6: 251-264, 1990.6251
1990
[PubMed]
Burwell, R. G.: The function of bone marrow in the incorporation of a bone graft. Clin. Orthop.,200: 125-141, 1985.200125
1985
[PubMed]
Cornell, C. N.; Lane, J. M.; Chapman, M.; Merkow, R.; Seligson, D.; Henry, S.; Gustilo, R.; and Vincent, K.: Multicenter trial of Collagraft as bone graft substitute. J. Orthop. Trauma,5: 1-8, 1991.51
1991
[PubMed]
DeLustro, F.; Dasch, J.; Keefe, J.; and Ellingsworth, L.: Immune responses to allogeneic and xenogeneic implants of collagen and collagen derivatives. Clin. Orthop.,260: 263-279, 1990.260263
1990
[PubMed]
Grundel, R. E., and Chapman, M. W.: Type I bovine fibrillar collagen and porous biphasic calcium phosphate ceramic for repair of segmental bone defects in the canine ulna. Read at the Annual Meeting of the Canadian Orthopaedic Research Society, Toronto, Ontario, Canada, May 28, 1989.
Grundel, R. E.; Chapman, M. W.; Yee, T.; and Moore, D. C.: Autogeneic bone marrow and porous biphasic calcium phosphate ceramic for segmental bone defects in the canine ulna. Clin. Orthop.,266: 244-258, 1991.266244
1991
[PubMed]
Gustilo, R. B., and Anderson, J. T.: Prevention of infection in the treatment of one thousand and twenty-five open fractures of long bones. Retrospective and prospective analyses. J. Bone and Joint Surg.,,58-A: 453-458, June 1976.58-A453
1976
Lane, J. M.; Muschler, G. F.; Werntz, J.; Nottebaert, M.; Gebhart, M.; Piergentili, C.; and Burstein, A.: The use of composite bone graft materials in a segmental femoral defect model in the rat. J. Orthop. Trauma,2: 57-58, 1988.257
1988
Laurie, S. W.; Kaban, L. B.; Mulliken, J. B.; and Murray, J. E.: Donor-site morbidity after harvesting rib and iliac bone. J. Plast. and Reconstr. Surg.,73: 933-938, 1984.73933
1984
McIntyre, J. P.; Shackelford, J. F.; Chapman, M. W.; and Pool, R. R.: Characterization of a bioceramic composite for repair of large bone defects. Am. Ceramic Soc. Bull.,70: 1499-1503, 1991.701499
1991
Moore, D. C.; Chapman, M. W.; and Manske, D.: The evaluation of a biphasic calcium phosphate ceramic for use in grafting long-bone diaphyseal defects. J. Orthop. Res.,5: 356-365, 1987.5356
1987
[PubMed]
Ohgushi, H.; Goldberg, V. M.; and Caplan, A. I.: Heterotopic osteogenesis in porous ceramics induced by marrow cells. J. Orthop. Res.,7: 568-578, 1989.7568
1989
[PubMed]
Summers, B. N., and Eisenstein, S. M.: Donor site pain from the ilium. A complication of lumbar spine fusion. J. Bone and Joint Surg.,71-B(4): 677-680, 1989.71-B(4)677
1989
Werntz, J.; Lane, J. M.; Piez, C.; Seyedin, S.; and Burstein, A.: The repair of segmental bone defects with collagen and marrow. Trans. Orthop. Res. Soc.,11: 108, 1986.11108
1986
Younger, E. M., and Chapman, M. W.: Morbidity at bone graft donor sites. J. Orthop. Trauma,3: 192-199, 1989.3192
1989
[PubMed]