What Is the Survivorship of TKA With a Twin-peg or Spikes-and-keel Cementless Implant Compared With Cemented? A Registry-based Cohort Study

Abstract

Background

Use of cementless TKA has grown after encouraging data from contemporary implants. Yet registry studies have shown inferior survivorship of cementless fixation when treated as a monolithic class aggregating contemporary and historic designs. Two contemporary cementless TKA designs with distantly different fixation strategies have emerged in the last 2 decades, mostly focused on tibial fixation: porous tantalum and twin-peg tibia and, more recently, porous titanium and a spikes-and-keel tibia. However, their survivorship in comparison with cemented options and between each other remains to be thoroughly delineated.

Questions/purposes

(1) Is there a difference in aseptic survivorship between cementless twin-peg TKA constructs compared with cemented options? (2) Is there a difference in aseptic survivorship between cementless spikes-and-keel TKA constructs compared with cemented options? (3) Is there a difference in aseptic survivorship between cementless twin-peg TKA constructs compared with cementless spikes-and-keel TKA constructs?

Methods

We conducted a cohort study using data from the Kaiser Permanente Total Joint Replacement Registry (TJRR). The TJRR prospectively collects patient, perioperative, and implant details on all patients who undergo TKA in a multiregional organization with 12 million members; patients included in the TJRR are longitudinally monitored for outcomes after TKA, and identified outcomes are manually validated through chart review. Patients who underwent primary TKA for osteoarthritis from 2009 to 2023 with a fully cementless construct of either twin-peg or spikes-and-keel, and their fully cemented options, were included (n = 136,443). TKA with hybrid or unknown fixation (2.6% [3571]), rotating or unknown mobility (0.8% [1081]), or fully constrained or unknown stability (1.1% [1549]) were excluded. The final study sample included 130,242 primary TKAs performed by 388 surgeons at 60 hospitals. Of the 125,414 patients receiving twin-peg TKAs, 9.2% who received cemented and 12.3% who received cementless were lost to follow-up. Of the 4828 patients receiving spikes-and-keel TKAs, 11.3% who received cemented versus 11.2% who received cementless were lost to follow-up. Those who terminated membership during the study period tended to be younger than those who did not (65 versus 68 years for both groups). A comparison between cementless (575 twin-peg and 1574 spikes-and-keel) versus cemented (124,839 twin-peg and 3254 spikes-and-keel) fixation was performed for each pair of analogous implants. For the twin-peg comparison, when comparing the cementless group to the cemented group, the cementless group was younger (61 versus 68 years), included more male patients (77% versus 39%), and more patients who self-reported White race (76% versus 66%). For the spikes-and-keel comparison, when comparing the cementless group to the cemented group, the cementless group was younger (65 versus 70 years), had a higher BMI (32 versus 31 kg/m²), included more male patients (47% versus 29%), and had a higher proportion of minimally stabilized implants (57% versus 33%). A secondary comparison was performed between the two fully cementless fixation constructs. When comparing the twin-peg to spikes-and-keel cementless groups, the twin-peg group was younger (61 versus 65 years), included more male patients (76% versus 47%), had a higher proportion of those with an American Society of Anesthesiologist (ASA) classification of 1 to 2 (70% versus 44%), and had a higher proportion of posterior stabilized implants (56% versus 43%). Before outcome evaluation, propensity score weights were calculated using multivariable logistic regression models that included patient age, gender, BMI, race, smoking status, ASA classification, implant stability, and operative year as predictors of treatment assignment. After the application of propensity score weighting, all factors were balanced between the comparison groups except for male gender for the cementless versus cemented twin-peg comparison and twin-peg versus spikes-and-keel cementless comparison. Propensity score–weighted Cox proportional hazards regression was used to evaluate aseptic revision risk of the TKA construct during follow-up for all comparisons. A secondary adjustment was applied in the regression model for male gender as balance was not achieved after propensity score weighting alone.

Results

In the adjusted analysis for the twin-peg group, cementless fixation had a higher risk of aseptic revision compared with cemented fixation (HR 2.13 [95% confidence interval (CI) 1.30 to 3.50]; p = 0.003). In the adjusted analysis for the spikes-and-keel group, no difference in aseptic revision risk was observed between cementless and cemented fixation (HR 0.94 [95% CI 0.56 to 1.58]; p = 0.82). When directly comparing cementless twin-peg to spikes-and-keel, twin-peg was associated with a higher risk of aseptic revision compared with spikes-and-keel (HR 2.56 [95% CI 1.13 to 5.79]; p = 0.02).

Conclusion

In this large registry-based cohort study, no cementless TKA construct was associated with better implant survivorship compared with their cemented counterpart options. Cementless fixation may continue to evolve and improve, as cementless spikes-and-keel performed better than twin-peg. But given that reliable results have been observed over longer periods with established cemented TKA, at present, the prudent choice remains cemented TKA. Future recommendations may change but require a demonstrable advantage in implant survivorship and outcomes of future cementless constructs over cemented fixation in carefully designed studies.

Level of Evidence

Level III, therapeutic study.

Introduction

Although cementation remains the dominant mode of fixation for TKA, interest in and usage of cementless TKA has grown in recent years. Advocates of cementless fixation suggest advantages in decreased surgical time, less reliance on a tourniquet [17], and the potential for eventual biological ingrowth and incorporation [2].

In the last decade, use of cementless TKA fixation has increased from just 1.9% of primary TKAs in the American Joint Replacement Registry in 2012 to 20.5% in 2022 [3]. In 2022, two manufacturers represented 89% of cementless TKAs implanted in the United States [8]. Released in 2013, Triathlon Tritanium® (Stryker) represented 75% of cementless implants sold and featured a 3D-printed highly porous titanium tibial surface with a spikes-and-keel design. Zimmer Biomet introduced the NexGen® in 1999 and later followed with the Persona® implant lines, featuring a tibia with porous tantalum trabecular metal (TM) hexagonal twin-pegs [8] that captured 14% of the same market.

Registry-based studies focused individually on either spikes-and-keel or twin-peg implants have reported encouraging results [5, 19, 25]. Yet when registries examined cementless TKA as a monolithic class, survivorship remained concerning potentially because of the poor performance of older implants [3, 5, 11, 15, 21, 24]. Therefore, as cementless fixation is now growing in utilization, a comparative registry study focused on more contemporary cementless designs, both between design types and compared with established cemented options, is needed. As historic implants may negatively influence aggregated registry survivorship studies, we sought to evaluate the survivorship of cementless TKA focused on two more modern classes of designs.

We asked: (1) Is there a difference in aseptic survivorship between cementless twin-peg TKA constructs compared with cemented options? (2) Is there a difference in aseptic survivorship between cementless spikes-and-keel TKA constructs compared with cemented options? (3) Is there a difference in aseptic survivorship between cementless twin-peg TKA constructs compared with cementless spikes-and-keel TKA constructs?

Patients and Methods

Study Design and Data Source

This was a retrospective, comparative, large database cohort study using data from the Kaiser Permanente Total Joint Replacement Registry (TJRR). This healthcare system provides care to more than 12 million patients in eight geographical regions of the United States (Colorado, Georgia, Hawaii, Mid-Atlantic states, Northern California, Northwest [including Oregon and Southern Washington], Southern California, and Washington). Patients are representative of the population in which the healthcare system operates [9]. Information on TJRR coverage, data collection procedures, and quality assurance have been detailed previously [22, 23]. Using electronic forms that are completed at the point of care, the operating surgeon report records information on the patient, procedure, implant, surgeon, and hospital information for all TKAs performed within the healthcare system. Additional data elements are obtained using the integrated electronic health record (EHR), administrative databases, other institutional databases, and mortality records. All TKAs included in the TJRR are longitudinally monitored for outcomes using electronic screening algorithms until either membership termination or death, and data are validated by trained research associates.

Covariates

We obtained information on patient and surgical covariates that were considered as potential confounders from the TJRR. Patient covariates included age (continuous), BMI (continuous), gender (women, men), self-reported race (Asian, Black, Hispanic, Other races, White), smoking status (current smoker, quit/never smoker), and American Society of Anesthesiologist (ASA) classification (1 to 2, ≥ 3, missing). Implant covariates included implant stability (minimally stabilized [collectively including cruciate retaining, anterior stabilized, and medial stabilized] and posterior stabilized).

Patients

The study sample included patients who underwent primary TKA for the diagnosis of osteoarthritis with a fully cemented Persona / NexGen TKA, a fully cementless Persona TM / NexGen TM (Zimmer Biomet), a fully cementless Triathalon, or fully cementless Triathlon Tritanium (Stryker Howmedica) implant from 2009 to 2023 (Fig. 1). In total, 148,662 cases were identified. Procedures using implants that were involved in a recall were excluded [3, 17]. TKA with hybrid or unknown fixation (2.6% [3571]), rotating or unknown mobility (0.8% [1081]), or constrained (that is, condylar constrained) or of unknown constraint (1.1% [1549]) were excluded. The final study cohort included 130,242 primary TKAs (125,414 twin-peg and 4828 spikes-and-keel) performed by 388 surgeons across 60 facilities in the healthcare system. Persona prostheses were the most frequently used of the Zimmer Biomet models included in the twin-peg cohort (59% cementless and 67% cemented) (Supplemental Table 1; http://links.lww.com/CORR/B378).

Fig. 1.

Study flow diagram. ^a = recalled implants included NexGen nonaugmentable option cemented tibial components.

Of the 125,414 patients receiving twin-peg TKAs, 9.2% who received cemented and 12.3% who received cementless were lost to follow-up. Of the 4828 patients receiving spikes-and-keel TKAs, 11.3% who received cemented versus 11.2% who received cementless were lost to follow-up. Those who terminated membership during the study period tended to be younger than those who did not (65 versus 68 years for both groups).

Descriptive Data

The Zimmer Biomet twin-peg cohort included 575 cementless and 124,839 cemented TKAs (Supplemental Table 1; http://links.lww.com/CORR/B378). During the study, cementless twin-peg was used infrequently compared with its cemented options (Fig. 2A). Patients who underwent cementless twin-peg fixation were younger (61 versus 68 years) and more represented by male gender (77% versus 39%) and self-reported White race (76% versus 66%); average BMI (32 versus 31 kg/m²) was similar between the two groups, as well as the proportion of current smokers (5% versus 3%), those with an ASA classification of 1 to 2 (70% versus 63%), and posterior stabilized constraint (56% versus 59%). After propensity score weighting, balance in the distribution of all factors was improved between the study groups (standardized mean difference [SMD] < 0.2), with the exception of gender (SMD 0.280) (Table 1). For the twin-peg cohort, the 10-year cumulative aseptic revision probability was 4.8% for cementless versus 2.2% for cemented TKA (Fig. 3A).

Fig. 2.

Volume of primary TKA performed within a US integrated healthcare system using (A) a twin-peg construct and (B) a spikes-and-keel construct.

Table 1.

Characteristics of 125,414 patients who underwent primary TKA using a twin-peg prosthesis (2009 to 2023) by fixation status before and after the application of propensity score weighting

Characteristic	Cementless (n = 575)	Cemented (n = 124,839)	SMDa
			Before weighting	After weighting
Age in years	61 ± 8	68 ± 9	0.778	0.162
BMI in kg/m2b	32 ± 6	31 ± 5	0.113	0.060
Men	77 (441)	39 (48,118)	0.837	0.280
Racec			0.282	0.151
White	76 (434)	66 (82,587)
Black	5 (31)	9 (10,992)
Hispanic	16 (90)	16 (20,019)
Other	4 (20)	9 (11,241)
Current smoker	5 (29)	3 (3145)	0.133	0.066
ASA classification			0.155	0.020
1-2	70 (401)	63 (79,178)
≥ 3	29 (168)	34 (42,748)
Missing	1 (6)	2 (2913)
Implant constraint			0.070	0.156
Minimally stabilized	44 (254)	41 (50,830)
Posterior stabilized	56 (321)	59 (74,009)

Data presented as mean ± SD or % (n).

^aSMD is used to understand the distribution for a covariate across study groups; an SMD of < 0.2 suggests that the distribution of the covariate is balanced between the study groups. Before propensity score weighting, the distribution of age, gender, and race was imbalanced between fixation groups. After weighting, all covariates except for gender were balanced.

^bBMI was missing in < 0.1% (131).

^cRace was based on self-reported information extracted into the registry from the integrated electronic health record.

Fig. 3.

The crude cumulative aseptic revision incidence (solid line) and 95% CI (shaded area) after cemented and cementless TKA using (A) a twin-peg construct and (B) a spikes-and-keel construct.

The Stryker spikes-and-keel cohort included 1574 cementless and 3254 cemented TKAs. Use of spikes-and-keel cementless fixation grew starting in 2015 and surpassed cemented fixation by 2021 (Fig. 2B). Patients who underwent cementless spikes-and-keel fixation were younger (65 versus 70 years), had higher BMI (32 versus 31 kg/m²), were more represented by male gender (47% versus 29%), and more often used minimally stabilized implants, including cruciate-retaining or cruciate-stabilized (57% versus 33%). After propensity score weighting, balance was improved for all covariates (SMD < 0.2) (Table 2). The 5-year cumulative aseptic revision probability was 1.2% for cementless versus 1.4% for cemented TKA (Fig. 3B). Crude incidences of revision for both twin-pegs and spikes-and-keel TKA have been provided (Supplemental Table 2; http://links.lww.com/CORR/B378).

Table 2.

Characteristics of 4828 patients who underwent primary TKA using a spikes-and-keel prosthesis by fixation status before and after the application of propensity score weighting

Characteristic	Cementless (n = 1574)	Cemented (n = 3254)	SMDa
			Before weighting	After weighting
Age in years	65 ± 8	70 ± 9	0.603	0.023
BMI in kg/m2b	32 ± 5	31 ± 5	0.204	0.015
Men	47 (735)	29 (940)	0.374	0.021
Racec			0.152	0.039
White	82 (1297)	86 (2784)
Black	5 (80)	4 (129)
Hispanic	10 (163)	7 (228)
Other	2 (34)	4 (113)
Current smoker	4 (65)	3 (107)	0.045	0.013
ASA classification			0.184	0.040
1-2	44 (696)	45 (1470)
≥ 3	38 (602)	44 (1415)
Missing	18 (276)	11 (369)
Implant constraint			0.509	0.003
Minimally stabilized	57 (900)	33 (1062)
Posterior stabilized	43 (674)	67 (2192)

Data presented as mean ± SD or % (n).

^aSMD is used to understand the distribution for a covariate across study groups; an SMD of < 0.2 suggests that the distribution of the covariate is balanced between the study groups. Before propensity score weighting, the distribution of age, BMI, gender, and implant constraint was imbalanced between fixation groups. After weighting, all covariates were balanced.

^bBMI was missing in < 0.1% (31).

^cRace was based on self-reported information extracted into the registry from the integrated electronic health record.

When comparing cementless twin-peg with cementless spikes-and-keel, we observed that the twin-peg group was younger (61 versus 65 years), more represented by male patients (76% versus 47%), had lower ASA classes (70% versus 44%), and more often used posterior stabilized implants (56% versus 43%). After propensity score weighting, all factors were balanced between the comparison groups with the exception of male gender (SMD 0.201) (Supplemental Table 3; http://links.lww.com/CORR/B378).

Primary and Secondary Outcomes

The primary outcome was aseptic revision surgery after the index TKA. This was procedurally based, defined as any operation after the primary TKA in which any (tibia, femur, liner, or patella) implant component was removed and replaced for noninfectious reasons. Revisions after the index TKA were monitored until death, healthcare membership termination, or the study end date of December 31, 2023. All identified revisions and revision reasons were manually validated using the EHR.

Our secondary goal was to evaluate indicated diagnoses for revision, specifically of loosening, instability, and periprosthetic fractures. Revision reasons were not mutually exclusive; patients could have had more than one reason for revision reported, and each revision reason was evaluated separately. A manual chart review was performed for the present study among the more limited cementless twin-peg and cementless spikes-and-keel cohorts to confirm failure mode.

Ethical Approval

This study was approved by Kaiser Permanente Institutional Review Board (#5488).

Statistical Analysis

Propensity score weighting was used to balance the distribution of covariates between the two study groups before revision risk assessment. Propensity scores were calculated using multivariable logistic regression models that included the covariates specified previously as predictors of treatment assignment. Propensity scores are the conditional probability of the patient receiving the treatment of interest given the covariates specified [4]. The standardized mean difference (SMD) for each covariate was measured before and after the application of propensity score weighting. The SMD provides information on the magnitude of the difference in the distribution of a covariate between two study groups; an SMD of < 0.2 was used to indicate a small difference, or rather, balance in the covariate distribution between the study groups [4]. We used the Rosenbaum approach to address missing covariate information; we created separate levels for nominal variables with missing values and we created a missing indicator for continuous variables with missing information while also imputing the mean. A secondary adjustment was applied in the regression model for gender for the twin-peg cementless versus cemented comparison and cementless twin-peg versus spikes-and-keel comparison as balance was not achieved after propensity score weighting [20].

We evaluated revision risk as a time to event using survival analyses. Follow-up time was defined as the difference between the index TKA and the revision date for those who experienced a revision or the difference between the index TKA and the date of healthcare membership termination, death, or study end date (December 31, 2023), whichever came first, for those who did not have a revision. Patients who had a septic revision, membership termination, death, or reached the study end date were censored at the last surveillance date. We calculated cumulative revision incidence at 10 years of follow-up as 1 minus the Kaplan-Meier estimate; given that spikes-and-keel cementless systems were not widely used until 2015, revision incidence was reported at 5 years of follow-up for this cohort. Average treatment effect propensity score–weighted Cox proportional hazards regression was used to assess the revision risk by fixation type using cemented fixation as the reference group. As there can be correlation in TKAs that were performed by the same surgeon, robust standard errors were used by including operating surgeon as a cluster term in the regression model. The proportional hazards assumption for the Cox regression model was checked, and the assumption was met. Outcomes with fewer than five events in at least one of the comparison groups were not modeled in regression analyses. Regression analysis only looked out to 5 years of follow-up for the spikes-and-keel comparison and the cementless twin-peg versus spikes-and-keel comparison. A p value < 0.05 was the statistical significance cutoff used for this study, and all tests were two-sided. All analyses were performed using RStudio, version 2021.09.1+372 (RStudio).

Secondary Analysis

We performed a secondary analysis comparing the two cementless TKA designs. For the twin-peg versus spikes-and-keel comparison, we used average treatment effect propensity score–weighted Cox proportional hazards regression with spikes-and-keel as the reference group. At 5 years of follow-up, the crude cumulative aseptic revision probability was 3.2% for twin-peg cementless TKA and 1.1% for spikes-and-keel cementless TKA (Fig. 4). Upon quality control chart review, in cementless twin-peg TKA, there were three revisions for tibial loosening (two with NexGen and one with Persona) and six revisions of femoral loosening (five with NexGen and one with Persona). With cementless spikes-and-keel TKA, there were three revisions for tibial loosening and one for femoral loosening.

Fig. 4.

The crude cumulative aseptic revision incidence (solid line) and 95% CI (shaded area) after cementless TKA.

Sensitivity Analysis

We examined the potential for surgeon experience and practice patterns confounding the association between fixation and revision risk, including surgeon preference for cementless or cemented TKA and a sensitivity analysis when restricting to surgeons with an annual procedure volume greater than 15 TKAs.

Results

Aseptic Survivorship of Cementless Twin-peg TKA Versus Cemented Option

In the propensity score–weighted regression analysis, cementless fixation had a higher risk of aseptic revision procedure compared with cemented fixation (HR 2.13 [95% confidence interval (CI) 1.30 to 3.50]; p = 0.003). When evaluating revision diagnoses, we observed a higher risk for aseptic loosening with cementless fixation compared with cemented (HR 4.31 [95% CI 2.15 to 7.93]; p < 0.001) (Table 3).

Table 3.

Adjusted risk for revision surgery after primary TKA by group

Revision reason	Adjusted HR (95% CI)a	p value
Twin-peg cementless versus cemented options (n = 125,414)
Aseptic revision	2.13 (1.30-3.50)	0.003
Aseptic loosening	4.31 (2.15-7.93)	< 0.001
Spikes-and-keel cementless versus cemented option (n = 4828)
Aseptic revision	0.94 (0.56-1.58)	0.82
Instability	0.92 (0.44-1.92)	0.82
Twin-peg cementless versus spikes-and-keel cementless
Aseptic revision	2.56 (1.13-5.79)	0.02

^aPropensity score–weighted Cox proportional hazards regression model accounted for age, BMI, gender, race/ethnicity, smoking status, ASA classification, and implant stability. A cluster term at the surgeon level was also included in the model to address correlation between TKAs that were performed by the same surgeon. Spikes-and-keel was the reference group.

Aseptic Survivorship of Cementless Spikes-and-keel Versus Cemented Option TKA

In the propensity score–weighted regression analysis, we observed no difference in 5-year aseptic revision risk for cementless compared with cemented TKA with the spikes-and-keel design (HR 0.94 [95% CI 0.56 to 1.58]; p = 0.82). Similarly, when evaluating specific revision diagnoses, we observed no difference in specific revision reasons with sufficient events for comparison (Table 3).

Aseptic Survivorship of Cementless Twin-peg Versus Cementless Spikes-and-keel

In the propensity score–weighted analysis, cementless twin-peg had a higher 5-year aseptic revision risk compared with cementless spikes-and-keel TKA (HR 2.56 [95% CI 1.13 to 5.79]; p = 0.02) (Table 3).

Sensitivity Analysis

When restricted to surgeons who had an annual volume of ≥ 15 TKAs, our results remained consistent. In the twin-peg cohort, no surgeons used a cementless-dominant pattern, and high-volume surgeons utilized cement > 80% of the time (Supplemental Figure 1A; http://links.lww.com/CORR/B377). In the spikes-and-keel cohort, only two surgeons invariably used cementless fixation, but they accounted for only a small number of TKAs in the sample. The remainder of surgeons in the spikes-and-keel cohort utilized cement selectively (Supplemental Figure 1B; http://links.lww.com/CORR/B377).

Discussion

Cementless TKA use has increased in the last decade, paralleled by encouraging survivorship data of modern implants compared with previous implants. Historically, cementless TKA demonstrated revision rates of 7% to 13% at long-term follow-up [6, 7, 11, 13]. Evaluation of modern cementless implants with single surgeons and single institutions, including the implants in this study, have been more encouraging, with survivorship comparable to their cemented counterparts [10, 14]. Yet despite improvements, it remains to be shown whether cementless TKA (including more recent generations) can outperform cemented TKA in a large registry-based cohort study. While the older twin-peg group was associated with inferior revision risk to cemented fixation, the cementless spikes-and-keel did not outsurvive its cemented option. Therefore, given that good results have been observed over longer periods of time in studies with cemented TKA, at present we recommend against wide adoption of cementless TKA over established cemented fixation.

This was a large registry study of cementless TKA with comparisons restricted to designs of more direct modern relevance. We were able to exclude TKAs in which implants involved in recalls were used. In this case, two notable recalls occurred: the NexGen cemented nonaugmentable tibia with a smooth undersurface was found to have increased cement debonding, while cementless Persona TM had a brief period in time in which the tibial tray instrumentation may have caused increased radiolucent lines [3, 16, 17, 26].

In addition, the study cohort was derived from a US-based, multiregional, integrated healthcare system where patients have previously been found to be representative of the areas served. Data used for the study came from a longitudinally maintained registry, in which detailed information on patient demographics, comorbidities, implants, outcomes, and coverage is available [22, 23].

Limitations

There are limitations to this study. The study design allowed us only to make claims of association as opposed to causality between implant design and revision risk. However, we have controlled for several patient demographic variables; therefore, between-group differences identified likely represent the performance in real-world practice.

Despite propensity score weighting, a gender imbalance with more male representation remained in the cementless twin-peg group when compared with its cemented option and to cementless spikes-and-keel. To address this lingering imbalance after weighting, gender was included as a covariate in the regression models. However, current evidence favors male gender for cementless TKA survivorship [3], which is likely the reason there was a heavy male selection bias. This would likely not change our results, as we observed inferior survivorship despite more male representation after propensity score weighting.

As the study is observational, there is the potential for additional confounding from unmeasured factors. Although we considered ASA classification, age, and gender, we did not directly control for specific diagnoses that can affect osteointegration, such as osteopenia/osteoporosis, metabolic bone diseases, or inflammatory arthritis.

Ambiguity also remains in the technique necessarily used with implant constraint. Historically, liner constraint came in only posterior cruciate-retaining (CR) and posterior-stabilized and cruciate-sacrificing (PS) implants. However, the emergence of intermediate and technique-flexible liner options such as “anterior stabilized,” “medial congruent,” and “cruciate stabilized” obfuscates the handling and integrity of the posterior cruciate ligament. We therefore dichotomized our analysis between femur type, as only in a PS femur can we be certain that the posterior cruciate ligament is sacrificed. This was accounted for in the propensity score weighting, and we did not find an interaction between liner stability and fixation type when defined in this way.

Preoperative deformity or postoperative alignment could not be controlled. Although there is growing attention surrounding alternative alignment philosophies and alignment phenotypes [1], at present, these are not tracked in the registry nor documented in the EHR or claims. One device included in this study period is FDA approved for alternative target alignment [27], but we are aware that the practice of alternative targets has been increasing in this study period and itself warrants further registry study.

There was a diversity in component design nuances in the twin-peg cohort. These included geometry (“Narrow,” “Gender,” or “Flex” femurs and “MIS” tibias) and femoral fixation technology (fiber-metal mesh on NexGen and TM on Persona). Given that survivorship is similar despite nuanced geometric differences, we did not disaggregate beyond PS versus CR [3]. On individual chart review, we found more femoral-sided loosening with the NexGen cementless femur (which features fiber-metal mesh): 2% (4 of 223) compared with 0.3% (1 of 340) Persona TM and 0.06% (1 of 1547) Triathlon. Although attention is typically paid to the tibia, this finding reflects a generational evolution in femoral fixation as well and ought to remain a topic of implant survivorship focus.

We could not evaluate additional outcomes of clinical relevance that did not result in revision (such as subsidence, stiffness, or radiologic evidence of loosening), as this information is not collected in the registry. However, revision surgery is a well-defined critical endpoint in which both the surgeon and patient have agreed that additional surgical intervention is needed.

Finally, although the study spanned 14 years, there was a low event rate of aseptic revisions, and there were too few events to evaluate some specific reasons for revision. Additional study in larger cohorts and longer follow-up with specific revision indications would be needed. However, the industry-wide migration away from twin-peg to spikes-and-keel designs may preclude that chance, and focused comparison between comparable next-generation spikes-and-keel designs would be of interest in the future.

Discussion of Key Findings

Both twin-peg and spikes-and-keel groups may have the support of respective randomized clinical trials finding survivorship comparable with, if not better than, cemented options [12, 18]. Yet results from real-world registries suggest a possible discrepancy in the performance of cementless TKA [3, 5]. We therefore sought to evaluate the performance of these two more recently used common cementless TKA designs within a large US-based healthcare system, both against their cemented options and between each other. Overall, we failed to find a survivorship advantage when comparing any cementless TKA design to its cemented option. Given this, we recommend against the routine and widespread adoption of cementless TKA at present.

We found a higher aseptic revision risk for cementless twin-peg TKA when compared both to its own cemented TKA options and with the newer cementless spikes-and-keel TKA. The most common location of failure may be with femoral component fixation, for which TM was added later and which itself reflects generational improvement. Therefore, despite more attention on the tibia component, we must continue to scrutinize both components’ performance in cementless TKA.

The spikes-and-keel design may have driven much of the recent increased utilization of cementless TKA, but to date, the survivorship of this design does not surpass cemented TKA. As several manufacturers have recently updated their cementless TKA to similar designs, registry monitoring of the next generation of implants will be crucial for directing future fixation strategies.

Conclusion

Cementless TKA, including recent designs, failed to demonstrate better survival than cemented TKA in our registry analysis; therefore, we recommend against the routine use of cementless TKA over established cemented designs at present. We acknowledge that there have been generational improvements with newer implants. Future utilization recommendations may change with the next generation of cementless designs if they can show improved survivorship (in either all or at least certain populations) compared with established implants.

Supplementary Material

abjs-483-1288-s002.docx ^{(29KB, docx)}