Over the past 25 years, an IRB-approved program has been in place at the authors’ institutions (Department of Veterans Affairs Salt Lake City Health Care System and Department of Orthopaedics, University of Utah School of Medicine) to analyze retrieved postmortem TKAs of the same specific design [2
] implanted by one surgeon (AAH) who had clinically followed the patients who ultimately donated [22
]. At the time of index operation, between 1985 and 1988, 76 patients consented to the retrievals upon death if possible. Good bone quality and the absence of osteopenic or osteoporotic bone were indications for selection of this implant. Bone quality was assessed by preoperative radiographs and intraoperatively by compressing the bone. During the time of these 76 implantations, the surgeon performed TKA in 360 patients. Clinical followup studies [22
] in this type of implant design at 10 to 14 years have demonstrated implant survivorship comparable to that of cemented designs [23
]: the initial patient population had high modified Hospital for Special Surgery Knee Score and a 95% survivorship at 10- to 14-year followup [22
]. At the time of the current study, we had accumulated 19 postmortem-retrieved primary TKA implants (patellar, tibial, and femoral) from 14 donors (12 men, two women) of which five were bilateral retrievals (Table ). The average (± SD) age at death was 77 ± 7 years (range, 66–91 years). The implants were in situ for an average of 11 ± 7 years (range, 1–25 years) (Table ). Previous data have been reported on five of these donor retrievals [2
]; however, the previous studies did not include analysis of all outcome variables reported in the current study.
Clinical data obtained from 14 donors’ medical records of which five were bilateral
A single orthopaedic surgeon (AAH) performed the surgeries according to a standardized operative approach using instruments specifically designed for the implants. The total knee components used in this study were of the Natural-Knee®
design with a commercially pure titanium porous coating with an average pore size of 530 μm and 55% porosity. In addition, the surgeon applied a layer of autograft bone chips, similar to milled bone, onto the resected host bone surface during the surgery and before implant placement [15
]. After preparing the milled bone from the underside surface of the resected tibia, the surgeon placed the bone slurry on the prepared undersurface of the patella and reversed the reamer to uniformly spread the slurry. For the femur and tibia, the surgeon spread the slurry over the surface of the resected bone using the scalpel handle, resulting in a 1.5- to 2-mm-thick layer of milled bone from the implant surface into the trabecular spaces.
At postmortem retrieval, the knee components (patella, tibia, and femur) were recovered in situ while ensuring the interface between the implants and host bone tissues remained undisturbed. We accessioned, photographed, and radiographed each component and removed all soft tissues and excess bone without compromising the bone-implant interface. Components were then fixed in 70% ethanol, dehydrated in ascending grades of ethanol, infiltrated, and embedded in methylmethacrylate using standard techniques [6
]. We then cut the components into the maximum number of 3-mm-thick sections (Fig. ), averaging seven patellar, eight tibial, and 10 femoral wafers per component, on a custom water-cooled, high-speed cutoff saw [5
] equipped with a diamond-edged blade (Rockazona Inc, Peoria, AZ, USA). The number of sections was dictated by the size of the individual components. The sections were ground and polished to a scratch-free finish using a variable-speed grinding wheel (Buehler Inc, Lake Bluff, IL, USA) using standard techniques. All histologic analyses were conducted in a blinded fashion by two observers (BMW, KEK).
Fig. 2 The image shows high-resolution contact radiographs of 3-mm-thick sections from the right patella of one of the donors used to measure the ABI. This patellar implant had been in situ for 84 months and had a high percentage of bone in apposition (more ...)
We sputter-coated each 3-mm-thick section with a conductive layer of gold for approximately 1 minute (Hummer VI-A; Anatech Ltd, Alexandria, VA, USA) and examined it using a scanning electron microscope (JSM 6100; JEOL Inc, Peabody, MA, USA) with a backscattered electron (BSE) detector (Tetra; Oxford Instruments Ltd, Cambridge, UK) and associated image capture software (ISIS; Oxford Instruments) at ×40 magnification. The entire length of the porous coating along the implant surface dictated the number of images captured from each component. We performed BSE imaging analysis to determine the percent area of bone present in the porous-coated region, periprosthetic interface region, and host bone region. The porous-coated region was defined as the 2-mm region within the porous coating. The periprosthetic region was defined, as in previous studies [6
], as being directly adjacent to the porous coating and extending for 2 mm from the coating surface. The periprosthetic interface region was composed of the autograft bone chips. The host bone region was defined as starting 3 mm from the porous coating surface and also extending 2 mm with the 1-mm distance between the periprosthetic and host bone region serving as a buffer to differentiate between these two regions (Fig. ). The total space imaged included the area occupied by implant porous coating, bone, and marrow [6
]. We calculated the percent void area within the porous coating, or area available for bone ingrowth that was not occupied by the implant, as the area occupied by bone divided by the available void in the pore space of the image field and multiplied by 100. Additional calculations of percent void area in the porous coating filled with bone and percent total area filled with bone were also performed for comparison with previously reported bone ingrowth literature utilizing BSE imaging [11
Fig. 3A–C BSE images illustrate the three regions used for analysis of bone ingrowth: (A) porous-coated region, (B) periprosthetic region, and (C) host bone region (original magnification, ×40). The periprosthetic interface region was in direct apposition (more ...)
We took high-resolution contact radiographs (Fig. ) of each 3-mm-thick section using high-resolution film (Kodak SO253; Eastman Kodak, Rochester, NY, USA) at 55 kV, 1.0 mA for 1800 seconds in a radiography cabinet (Torrex 120D; Scanray, Hawthorne, CA, USA) [3
]. The appositional bone index (ABI) of each component [3
] was measured by viewing the high-resolution contact radiographs at ×8 magnification on a radiographic light box and taking measurements using a calibrated, hand-held digital caliper (CD-6B; Mitutoyo Corp, Painesville, OH, USA). Radiolucencies interposed between the bone and implant (ΣL2
) were subtracted from the total linear length of the porous-coated interface (ΣL1
). We then calculated the ABI for the entire component as the percentage of bone that appeared in direct contact with the porous-coated regions of the components using the following equation:
This method was developed to better discriminate the tissue response and fibrous tissue formation at the interface of each component and for inspection of regional analysis to determine the extent of osteolysis compared to the limitations of gross radiographs, which interfere with visibility due to projection effects (Fig. ). These contact radiographs were then used to document regional osteolysis (Fig. ). After this process, the 3-mm-thick sections were subsequently ground down to 50 μm to confirm the presence of wear particulate in the osteolytic region using transmitted and polarized light microscopy.
Fig. 4A–B (A) A gross radiograph and (B) high-resolution contact radiograph from the same specimen show projection effects inherent with gross radiography, making detailed regional observations difficult but showing the superior detail with high-resolution contact (more ...)
High-resolution contact radiographs of the (A) femur, (B) tibia, and (C) patella show the regions used for reporting osteolysis (Table ).
The area of bone (bone ingrowth, periprosthetic bone, and host bone) from sections of femur, patella, and tibia were averaged to obtain a single regional value per donor (n = 19: patellas, tibias, and femurs). A paired t-test was used when comparing regions (percent bone ingrowth, percent periprosthetic bone, and percent host bone) within the patellar, tibial, or femoral components. Univariable regression was performed with implantation time as the independent variable and percent bone ingrowth or ABI as the dependent variable, with n = 19 donor implants for each of the three types of components (patellar, tibial, and femoral). We performed all analyses with SAS® software (SAS Institute, Inc, Cary, NC, USA).