Effect of robotic milling on periprosthetic bone remodeling

Effect of Robotic Milling on Periprosthetic Bone Remodeling Takehito Hananouchi,1 Nobuhiko Sugano,1 Takashi Nishii,1 Nobuo Nakamura,2 Hidenobu Miki,1Akihiro Kakimoto,2 Mitsuyoshi Yamamura,2 Hideki Yoshikawa1 1Department of Orthopedic Surgery, Osaka University Graduate School of Medicine, Osaka, Japan 2Center of Arthroplasty, Kyowakai hospital, Osaka, Japan Received 6 June 2006; accepted 26 December 2006 Published online 24 April 2007 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jor.20376 ABSTRACT: The ROBODOC system has provided better fit and fill of the stem and less destructionof the bony architecture than with manual surgery. These benefits might affect femoralperiprosthetic bone remodeling. We evaluated the effects of robotic milling in cementless total hiparthroplasty (THA) in a longitudinal 24-month follow-up study using dual energy X-rayabsorptiometry (DEXA) and plain radiographs of 29 patients (31 hips) after ROBODOC THA and24 patients (27 hips) after manual THA with the same stem design. To minimize the influence of otherfactors on bone remodeling, only female osteoarthritis patients, who had no drugs that might affectbone metabolism were enrolled. Significantly less bone loss occurred at the proximal periprostheticareas in the ROBODOC group. In zone 1, the decrease was 15.5 versus 29.9% using conventionalrasping; in zone 7, the loss was 17.0% with ROBODOC compared to 30.5% with conventional rasping(p < 0.05). On radiographs, endosteal spot welds in the proximal medial portion were morepronounced in the ROBODOC group (48 vs. 11% in the conventional group, p < 0.05). Our resultssuggest that robotic milling is effective in facilitating proximal load transfer around the femoralcomponent and minimizing bone loss after cementless THA. ß 2007 Orthopaedic Research Society.
Published by Wiley Periodicals, Inc. J Orthop Res 25:1062–1069, 2007Keywords: ROBODOC system; periprosthetic bone remodeling our knowledge, however, the effects of roboticsurgery on periprosthetic bone remodeling have ROBODOC (Integrated Surgical Systems, Davis, not been investigated quantitatively.
CA) is an active system of computer-assisted The aim of this study was to investigate surgery that helps surgeons to prepare the femoral periprosthetic bone remodeling after ROBODOC canal for cementless total hip arthroplasty (THA) THA in a longitudinal follow-up study using by using a five-axis robotic arm and a high- dual-energy X-ray absorptiometry (DEXA) and speed milling device.1 Based on postoperative plain radiographs in comparison with a control radiographic assessment, ROBODOC surgery pro- THA group performed without ROBODOC. To vided better alignment and fit and fill of the stem minimize the influence of other factors on bone than conventional surgery.1–3 In addition, the remodeling, only female patients with a preopera- robotic milling system provided less destruction tive diagnosis of osteoarthritis were enrolled.
of intramedullary trabecular architecture4 andreduced the amount of fatty and bony debrisintroduced into the venous system, resulting in a MATERIALS AND METHODS lower incidence of severe intraoperative embolicevents.5 Because proximal bone remodeling is From March 2001 to December 2003, 53 patients considered to be related to stem fit and fill6 and (58 hips) gave informed consent to be enrolled in this preservation of endosteal trabecular architecture prospective study and underwent primary THA. Eligiblepatients included females with a diagnosis of osteoar- and vascularity,7 robotic surgery could reduce the thritis with no previous surgery and no administration periprosthetic bone loss after THA. To the best of of drugs that affect bone mineral metabolism. Thepatients were allowed to choose which group (with andwithout ROBODOC) to join. The femoral canal was pre- Correspondence to: T. Hananouchi (Telephone: þ81-6-6879- pared with the ROBODOC in 29 patients (31 hips), and 3552; Fax: þ81-6-6879-3559; with a standard rasp in 24 patients (27 hips). There were no significant differences between the two groups in ß 2007 Orthopaedic Research Society. Published by Wiley Periodicals,Inc.
age (ROBODOC ¼ 56.7
9.2 years, conventional ¼ 57.4
JOURNAL OF ORTHOPAEDIC RESEARCH AUGUST 2007 BONE REMODELING BY ROBOTIC MILLING 7.1 years) and body mass index (ROBODOC ¼ 22.6
3.4, Periprosthetic bone mineral density (BMD) surround- convetional ¼ 22.5
2.8).
ing the femoral component was measured in the 7 Gruen In all patients, a VerSys Fiber Metal Taper (Zimmer, zones16 using DEXA (DPX-L, Lunar Madison, WI) at Warsaw, IN) titanium alloy stem was inserted without 3 weeks and 12 and 24 months after surgery. Because the cement. This stem is a straight type component with a DEXA measurement was affected by the rotational distal tapered design and has a proximal titanium fiber position of the leg,17 and patients with severe osteoar- metal porous coating with hydroxyapatite and a polished thritis had difficultly placing the leg in neutral position distal portion. To minimize the difference in the decision due to contracture, preoperative DEXA measurements criteria for stem size between the two groups, preopera- were not performed. Instead, the DEXA measurements tive planning was done using the ORTHODOC system, a 3 weeks postoperatively, when patients were free from preoperative planning workstation of the ROBODOC,8,9 contracture, were used as baseline values.
which provides the optimal size and position of the Patients were placed supine on the table with femoral component for appropriate fit and fill. In standard knee and foot supports so that the femur was the conventioal group, the surgeons were informed of in a neutral position (08 of neck anteversion). DEXA the results of preoperative planning before the operation measurements and the evaluation of the scans were all and were encouraged to use the recommended stem size.
performed by the same medical assistant, who was Because a larger size proximal porous-coated stem is blinded to patient selection. For quality control of the associated with greater proximal bone loss,10 we eval- DEXA machine, the assistant performed the calibration uated whether the size distribution between the two before the initial examination of the measurement groups was comparable. The acetabular component day. Longitudinal precision on a phantom estimated a (Trilogy; Zimmer) was implanted using a press-fit coefficient of variation of <2% during the study period.
method after underreaming of 1 mm in both groups.
According to the previous DEXA report, initial BMD The postoperative protocol was the same in both groups, 3 weeks postoperatively in zone 4 had significant with immediate full weight bearing as tolerated. From influence on subsequent proximal bone loss.17 Thus, postoperative day 1, patients were allowed to transfer BMDs around the distal portion of the prosthesis were from bed to wheel chair with full weight bearing under compared between both groups. To analyze regional bone control of a physical therapist. Then, patients were able density change in each zone, the BMD ratio was to walk with a walker or T-cane, depending on the calculated 12 and 24 months postoperatively in each patient's recovery.
zone as a percentage of the 3 week values.17 The patients were followed using clinical, radio- In previous articles assessing periprosthetic femoral graphical, and bone mineral assessments. The clinical bone remodeling with different factors or interven- rating before the operation and postoperatively at tions,18–24 a 5 to 10% difference in BMD decrease by 24 months was determined using the Merle d'Aubigne´ DEXA between test groups was identified as significant.
hip score.11 All plain radiographs were scanned digitally Therefore, we calculated a sample-size to detect the 5% with 300 dpi and were evaluated using Image-J software difference in the BMD decrease. Twenty-five hips in each (NIH). On immediate postoperative radiographs, stem group were sufficient to determine whether there was a alignment was measured by means of the varus/valgus significant difference (power ¼ 0.8 and p < 0.05).
stem angle, defined as the difference between the axes of For comparison between the two groups, we used the femur and stem.1 The medial fit of the stem1 at the univariate analysis for data grouped into distinct neck cut level was graded as good if there was contact categories (spot welds, medial proximal fit, radiolucent between the stem and femoral medial cortex, fair if there lines, ectopic ossification, and intraoperative fracture) was a 1- to 2-mm space, and poor if the space was more with Fisher's exact test. We used the Mann-Whitney than 2 mm, based on a modification of the classification by U-test for continuous data (age, BMI, Merle d'Aubigne´ Woolson et al.12 The fill of the stem at the level of the score, stem alignment, proximal fill ratio, and stem size).
lesser trochanter1 was measured as the percentage of the The difference of the BMD ratios between the two groups width of the femoral component to that of the medullary was examined using repeated measure ANOVA.
cavity. Radiolucent lines around the proximal poroussurface13 and endosteal spot welds14 were assessed on the2-year follow-up radiographs. Heterotopic ossification was evaluated according to Brooker's classification.15 The radiographic assessment was performed by two The median Merle d'Aubigne´ hip scores11 before orthopaedic surgeons in a blind fashion. We evaluated surgery was 9.5 (pain 1.9, mobility 4.1, ability 3.5) intraobserver and interobserver variability using Pear- in the ROBODOC group and 9.9 (pain 1.8, mobility son correlation coefficient (r) for stem alignment and fit 4.2, ability 3.9) in the conventional group. At and Kappa coefficients (k) for the appearance of spot 24 months, these scores improved to 17.8 (pain 6.0, welds. In the intraobserver variability of alignment, fit,and spot welds (from two measurements separated by a mobility 5.9, ability 5.9) and 17.7 (pain 6.0, 3-week interval), r was 0.74 and 0.78, respectively, and k mobility 5.8, ability 5.9), respectively (Table 1).
was 0.76. In the interobserver variability, r was 0.70 and There was no significant difference in these scores 0.83, and k was 0.77.
between the two groups at 24 months. In the JOURNAL OF ORTHOPAEDIC RESEARCH AUGUST 2007 HANANOUCHI ET AL.
Comparison of Clinical Results in ROBODOC and Conventional Groups Conventional Group Clinical score (Merle d'Aubigne´, average
SD) 24 months after surgery The average (range) size of used stem No. of stems using different size compared with No. of intraoperative femoral fracture *p < 0.05 with Fisher's exact test.
ROBODOC group, a stem of the same size proximal zones 24 months postoperatively was as suggested by preoperative planning was significantly smaller in the ROBODOC group (zone implanted in all hips. In the conventional group, 1; 15.5%, zone 7; 17.0%) than the conventional a stem of the same size as suggested by preopera- group (zone 1; 29.9%, zone 7; 30.5%) (Table 4).
tive planning was implanted in 22 hips, and an The difference of the decrease in the BMD ratio at undersized stem was implanted in 5 hips. There distal zones (zone 4 and 5) between the two groups were no significant differences in sizes between was only 6%, but the ROBODOC group also showed the groups. No intraoperative femoral fractures a significantly smaller decrease of the BMD ratio.
occurred in the ROBODOC group during stemfixation, while two patients in the conventionalgroup had intraoperative fractures that requiredtreatment with cable cerclage. No patients in either group underwent revision surgery by24 months.
Periprosthetic bone remodeling is influenced Radiographic examples are shown in Figure 1.
by gender,18,19 perioperative density,17,20,21 stem Immediate postoperative radiographs showed that size,10 and stem material.22 To clarify the effects of stem alignment was better in the ROBODOC group robotic milling on adjacent bone remodeling, the (p ¼ 0.01) (Table 2). The number of hips rated with effects of these other factors were minimized by good proximal medial stem fit was significantly enrolling only women with osteoarthritis and by larger in the ROBODOC group (p < 0.0001). A using the same stem design and material and significant difference in fill was found in favor of the the same preoperative planning procedure in all ROBODOC group (p ¼ 0.04). No radiolucent lines patients. In addition, patient ages, stem size, and in the proximal portion or progressive subsidence postoperative BMD of the femoral distal portion occurred in any hip at 24 months. Endosteal spot were comparable between groups.
welds in the proximal portion (zones 1 and 7) were DEXA assessments showed significant dif- detected in 26 (84%, zone 1) and 15 (48%, zone 7) of ferences between the two groups in almost all the ROBODOC cases and 23 (85%, zone 1) and 3 zones, especially proximally, suggesting that dif- (11%, zone 7) of the conventional cases. The ferences in femoral preparation between robotic difference in endosteal spot welds in zone 7 was milling and manual rasping affected periprosthetic (p ¼ 0.004). The radiographs also bone remodeling. Two mechanisms may account for showed class 1 ectopic ossification in one patient the more favorable bone remodeling after robotic in each group.
milling. First, 95% of the ROBODOC group had a There was no significant difference between the good fit compared to 63% in the conventional group.
groups in initial BMD in all zones (Table 3).
Similarly, better fill occurred in the ROBODOC However, repeated-measure ANOVA in all zones but zone 3 showed significant differences between reports.1–3 A favorable proximal fit and fill provides the two groups (zone 1, p ¼ 0.010; zone 2, p ¼ 0.012; better bone ingrowth proximally.6 Therefore, the zone 3, p ¼ 0.23; zone 4, p ¼ 0.044; zone 5, p ¼ 0.034; better proximal fit and fill in the ROBODOC group zone 6, p ¼ 0.020; zone 7, p ¼ 0.0012) (Fig. 2).
might facilitate proximal load transfer from the Especially, the decrease in the BMD ratio at the stem to the surrounding bone with a high rate of JOURNAL OF ORTHOPAEDIC RESEARCH AUGUST 2007

BONE REMODELING BY ROBOTIC MILLING (a–c) Radiographs of a 66-year-old woman in the ROBODOC group. The 2-year postoperative radiograph (b) shows spot welds around the hydroxyapatite coating (c; white and blackarrows). The medial femoral neck is unchanged from the immediate postoperative radiograph (a).
(d,e) Radiographs of a 56-year-old woman in the conventional group. The 2-year postoperativeradiograph (e) shows severe osteopenia in the proximal portion (zone 7) compared with thepostoperative film (d).
ingrowth or ongrowth of periprosthetic bone. This Second, robotic milling provides nearly intact favorable biomechanical environment could be trabecular architecture between the stem and reflected in the preservation of the BMD ratios in adjacent cortical bone, whereas with rasp prepara- zones 1, 2, 6, and 7.
tion, the bone microstructure is grossly destroyed.4 JOURNAL OF ORTHOPAEDIC RESEARCH AUGUST 2007

HANANOUCHI ET AL.
Thus, with rasp preparation, vascular injury Less destruction of adjacent bony microstructure and decreased blood supply in association with around the stem using robotic milling might the destruction of trabecular architecture may enhance recovery or renewal of periprosthetic bone after surgery. We speculate that BMD ratios in Comparison of Radiographic Results in ROBODOC and Conventional Groups ROBODOC (n ¼ 31) Conventional (n ¼ 27) Immediate postoperative radiographic assessment Stem (varus/valgus) alignment (8) Proximal medial fit assessment of stem (good/fair/poor) (% graded poor) Proximal fill assessment of stem (%) 24-month postoperative radiographic assessment No. of cases with radiolucent lines No. of cases with spot welds in proximal zone No. of cases with ectopic ossifications *p < 0.05 with Mann-Whitney U-test.
**p < 0.05 with Fisher's exact test.
JOURNAL OF ORTHOPAEDIC RESEARCH AUGUST 2007

BONE REMODELING BY ROBOTIC MILLING The Initial BMD (Average
SD) at of the ROBODOC group demonstrated proximal Postoperative 3 Weeks in the Two Groups bone ingrowth fixation without radiolucent lines,and spot welds in zone 7 indicate a good biologic response. Bone thermal damage may be less evident in a primary THA using the ROBODOC system since there is no need to remove cement or to mill sclerotic bone from previous primary surgery.
We did not compare the time to recovery of walking ability postoperatively between the two groups, although time might influence peripros- thetic bone remodeling. One report showed that late weight bearing resulted in a large reduction inBMD around the stem.28 However, we previouslyfound no significant difference comparing roboticand manual surgery with respect to the time the distal zones 4 and 5 of the ROBODOC group required to regain the ability to walk more than were better partly because bony architecture was six blocks without a cane.29 Thus, weight bearing was not considered to be associated with the Heat damage by robotic milling may adversely difference of bone remodeling between the two groups in this study.
radiolucency on plain radiographs.25,26 A risk of Our study has limitations. First, the follow-up heat injury throughout the milling procedure in period was only two years. In previous reports, revision THA using the ROBODOC system was long-term BMD changes in periprosthetic femoral reported.27 However, in our study, radiographs bone have been controversial.22,30 One report Time-related changes in median BMD ratios in Gruen's zones 1 to 7. ROBODOC surgery (Robo) and conventional manual surgery (Rasp).
JOURNAL OF ORTHOPAEDIC RESEARCH AUGUST 2007 HANANOUCHI ET AL.
Comparison of Postoperative BMD Ratios (Average
SD) between the showed that BMD in zone 7 decreased continuously Another report showed that bone loss in zone We thank Dr. Yasuhisa Hayaishi, Gratia Hospital, 7 2 years postoperatively recovered progressively Dr. Kohei Yabuno, Osaka police Hospital, and Dr. Ichiro by 10 years.22 Further follow-up of our patients is Nakahara, North Osaka police Hospital for support required to determine the long-term effect of during the gathering of data. No competing interests robotic milling on BMD change. Second, this study was not randomized; patients enrolled in our studychose robotic milling or manual THA themselves,although the two groups were comparable in termsof preoperative and postoperative factors that female patients were enrolled. In our country, 1. Bargar WL, Bauer A, Borner M. 1998. Primary and patients with hip osteoarthritis are predominantly revision total hip replacement using the Robodoc system.
Clin Orthop Relat Res 354:82–91.
female.31 Previous reports showed that the extent 2. Nishihara S, Sugano N, Nishii T, et al. 2004. Clinical of the stress shielding was different between male accuracy evaluation of femoral canal preparation using the and female patients,18,19 so we avoided this effect ROBODOC system. J Orthop Sci 95:452–461.
by enrolling no male patients. Finally, caution is 3. Schneider J, Kalender W. 2003. Geometric accuracy in necessary when applying our results to other robot-assisted total hip replacement surgery. ComputAided Surg 83:135–145.
studies of periprosthetic bone remodeling. Bone 4. Jerosch J, Peuker E, von Hasselbach C, et al. 1999.
remodeling is affected by patient-related factors Computer assisted implantation of the femoral stem in such as bone chemical markers,32 BMD of other THA—an experimental study. Int Orthop 234:224–226.
body parts,21 menopause, and stem related- 5. Hagio K, Sugano N, Takashina M, et al. 2003. Effective- factors such as design (e.g., anatomical stem), ness of the ROBODOC system in preventing intraopera-tive pulmonary embolism. Acta Orthop Scand 743:264– surface finish and the extent of porous coating.33 Further trials in a large number of patients 6. Laine HJ, Puolakka TJ, Moilanen T, et al. 2000. The effects stratified by multiple factors are required to of cementless femoral stem shape and proximal surface examine how these other factors affect the impact texture on ‘‘fit-and-fill'' characteristics and on bone of robotic milling impact on periprosthetic bone remodeling. Int Orthop 244:184–190.
7. Santavirta S, Ceponis A, Solovieva SA, et al. 1996.
Periprosthetic microvasculature in loosening of total hip In conclusion, this study of female osteoarthritic replacement. Arch Orthop Trauma Surg 1155:286–289.
patients evaluated periprosthetic bone remodeling 8. Haraguchi K, Sugano N, Nishii T, et al. 2001. Comparison and found that robotic milling reduced postopera- of fit and fill between anatomic stem and straight tapered tive bone loss of the proximal femur after cement- stem using virtual implantation on the ORTHODOCworkstation. Comput Aided Surg 65:290–296.
less THA using a straight type stem with 9. Nishihara S, Sugano N, Nishii T, et al. 2003. Comparison of proximal porous coating and polished distal taper the fit and fill between the Anatomic Hip femoral component and the VerSys Taper femoral component JOURNAL OF ORTHOPAEDIC RESEARCH AUGUST 2007 BONE REMODELING BY ROBOTIC MILLING using virtual implantation on the ORTHODOC work- 22. Karachalios T, Tsatsaronis C, Efraimis G, et al. 2004. The station. J Orthop Sci 83:352–360.
long-term clinical relevance of calcar atrophy caused by 10. Skoldenberg OG, Boden HS, Salemyr MO, et al. 2006.
stress shielding in total hip arthroplasty: a 10-year, Periprosthetic proximal bone loss after uncemented hip prospective, randomized study. J Arthroplasty 194:469– arthroplasty is related to stem size: DXA measurements in 138 patients followed for 2–7 years. Acta Orthop 77:386– 2004. Effects of discontinuation as well as intervention 11. Merle d'aubigne R, Postel M. 1954. Functional results of of cyclic therapy with etidronate on bone remodeling hip arthroplasty with acrylic prosthesis. J Bone Joint Surg after cementless total hip arthroplasty. Bone 35:217– Am 36-A3:451–475.
12. Woolson ST, Adler NS. 2002. The effect of partial or full 24. Arabmotlagh M, M, Hennigs T. 2006.
weight bearing ambulation after cementless total hip Alendronate prevents femoral periprosthetic bone loss arthroplasty. J Arthroplasty 177:820–825.
following total hip arthroplasty: prospective randomized 13. Engh CA, Bobyn JD, Glassman AH. 1987. Porous-coated double-blind study. J Orthop Res 24:1336–1341.
hip replacement. The factors governing bone ingrowth, 25. Berman AT, Reid JS, Yanicko DR Jr, et al. 1984. Thermal- stress shielding, and clinical results. J Bone Joint Surg Br ly induced bone necrosis in rabbits. Relation to implant failure in humans. Clin Orthop Relat Res 186:284–292.
14. Geesink RG, Hoefnagels NH. 1995. Six-year results of 26. Mjoberg B, Pettersson H, Rosenqvist R, et al. 1984. Bone hydroxyapatite-coated total hip replacement. J Bone Joint cement, thermal injury and the radiolucent zone. Acta Surg Br 774:534–547.
Orthop Scand 556:597–600.
15. Brooker AF, Bowerman JW, Robinson RA, et al. 1973.
27. Nogler M, Krismer M, Haid C, et al. 2001. Excessive Ectopic ossification following total hip replacement. Inci- heat generation during cutting of cement in the Robodoc dence and a method of classification. J Bone Joint Surg Am hip-revision procedure. Acta Orthop Scand 726:595– 16. Gruen TA, McNeice GM, Amstutz HC. 1979. ‘‘Modes of 28. Boden H, Adolphson P. 2004. No adverse effects of early failure'' of cemented stem-type femoral components: a weight bearing after uncemented total hip arthroplasty: a radiographic analysis of loosening. Clin Orthop Relat Res randomized study of 20 patients. Acta Orthop Scand 17. Nishii T, Sugano N, Masuhara K, et al. 1997. Longitudinal 29. Nishihara S, Sugano N, Nishii T, Miki H, Nakamura N, evaluation of time related bone remodeling after cement- Yoshikawa H. Comparison between hand-rasping and less total hip arthroplasty. Clin Orthop Relat Res 339:121– robotic milling for stem implantation in cementless total hip arthroplasty. J Arthroplasty (in press).
18. Sychterz CJ, Engh CA. 1996. The influence of clinical 30. Aldinger PR, Sabo D, Pritsch M, et al. 2003. Pattern of factors on periprosthetic bone remodeling. Clin Orthop periprosthetic bone remodeling around stable uncemented Relat Res 322:285–292.
tapered hip stems: a prospective 84-month follow-up study 19. Venesmaa PK, Kroger HP, Miettinen HJ, Jurvelin JS, and a median 156-month cross-sectional study with DXA.
Suomalainen OT, Alhava EM. 2001. Monitoring of peri- Calcif Tissue Int 732:115–121.
prosthetic BMD after uncemented total hip arthroplasty 31. Inoue K, Wicart P, Kawasaki T, et al. 2000. Prevalence with dual-energy X-ray absorptiometry—a 3-year follow- of hip osteoarthritis and acetabular dysplasia in french up study. J Bone Miner Res 166:1056–1061.
and japanese adults. Rheumatology 39:745–748.
20. Arabmotlagh M, Hennigs T, Warzecha J, et al. 2005.
32. Yamaguchi K, Masuhara K, Yamasaki S, et al. 2003.
Bone strength influences periprosthetic bone loss after hip Predictive value of a preoperative biochemical bone arthroplasty. Clin Orthop Relat Res 440:178–183.
marker in relation to bone remodeling after cementless 21. Rahmy AI, Gosens T, Blake GM, et al. 2004. Periprosthetic total hip arthroplasty. J Clin Densitom 6:259–265.
bone remodelling of two types of uncemented femoral 33. Yamaguchi K, Masuhara K, Ohzono K, et al. 2000.
implant with proximal hydroxyapatite coating: a 3-year follow-up study addressing the influence of prosthesis cementless total hip arthroplasty. The influence of the design and preoperative bone density on periprosthetic extent of porous coating. J Bone Joint Surg Am 82:1426– bone loss. Osteoporos Int 154:281–289.
JOURNAL OF ORTHOPAEDIC RESEARCH AUGUST 2007

Source: http://www.robodoc.com.sg/references/25_Effect_of_Robotic_Milling_on_Periprosthetic_Bone_Remodeling_2006_P.pdf