1 Background

Knee osteoarthritis (OA) represents one of the most prevalent joint disorders worldwide, particularly affecting older adults. With the increase in life expectancy and the aging population, the incidence of knee OA has risen, necessitating a greater demand for effective therapeutic interventions. Total knee arthroplasty (TKA) is regarded as the gold standard for the treatment of end-stage knee osteoarthritis, aimed at alleviating pain and restoring functionality when conservative methods, such as physical therapy or intra-articular injections, prove insufficient.¹

Despite the designation of TKA as the gold standard intervention, approximately 15 to 20 percent of patients remain dissatisfied with their outcomes. A myriad of factors, encompassing both patient-specific elements and surgeon-controlled variables, significantly influence patient-reported outcome measures and satisfaction following TKA.²

While traditional TKA techniques have demonstrated long-term success, challenges persist in achieving precise component alignment and optimal joint kinematics. Advancements in implant design, surgical techniques, and perioperative care have positively impacted patient outcomes over the years; however, issues related to variability in alignment, soft tissue balancing, and implant positioning continue to pose challenges. These factors are critical as they directly influence postoperative function, pain relief, and the durability of the implants.^3,4

Robotic-assisted total knee arthroplasty has transformed surgical methodologies by addressing the limitations associated with conventional manual techniques. Robotic systems are engineered to enhance surgical precision, improve consistency in alignment, and provide superior balancing of the knee joint.^5–9 The available robotic systems can be classified into three categories: active systems, where the surgeon positions the robotic arm near the patient and the robot performs the surgery autonomously; haptic systems, where the surgeon controls the instrument by activating a “go” button while the robot maintains the movement within pre-defined parameters; and collaborative systems, in which the surgeon retains control of the procedure and collaborates with a smart robotic tool.¹⁰

Furthermore, robotic systems can be divided into two types: “closed” platforms, which are specifically designed to operate with a particular implant, and “open” platforms, which provide flexibility by allowing the use of various implants and designs based on the surgeon's preferences or the patient's requirements.¹¹

Historically, robotic-assisted TKA has depended on image-based systems that require preoperative imaging—such as computed tomography (CT), magnetic resonance imaging (MRI), or X-rays. These systems facilitate detailed three-dimensional reconstructions of the patient's anatomy, enabling more accurate surgical planning and execution.^12–14 Although these imaging techniques enhance preoperative planning for precise bone resections and optimal implant positioning, they also present inherent drawbacks, including costs, radiation exposure, processing time, and the necessity of reproducing the landmarks required by the system.^15,16 Such challenges, in conjunction with the complexities of integrating imaging data into surgical workflows, have propelled interest in alternative methods.

Image-free robotic systems have emerged as a promising and efficient solution for surgical procedures. Notably, the ROSA Knee System currently stands out as the sole system that offers surgeons the option to utilize image-based techniques, employing long leg 2D X-rays, or to operate without images. Preliminary studies suggest that image-free systems may yield comparable precision and clinical outcomes to their image-based counterparts.^17,18

This systematic review aims to provide a comprehensive analysis of imageless robotic assisted TKA focusing on clinical efficacy, patient safety, economic impact, practical feasibility and precision.

2 Materials and methods

The review followed the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, ensuring a thorough and systematic approach to data collection and analysis.¹⁹ This systematic review has also been registered with the International Prospective Register of Systematic Reviews (PROSPERO), under registration number 1161673.

2.3

2.3 Data extraction and analysis

The titles and abstracts were independently screened by two reviewers. For abstracts that either met the inclusion criteria or caused uncertainty, full-text articles were obtained. These full texts were subsequently re-evaluated by the same two independent reviewers. Any discrepancies were resolved through assessment by the senior author, S.M.P.R (Fig. 1).

Close

Fig. 1 PRISMA flowchart.

Export to PPT

The methodological quality of each study was assessed using the Modified Coleman Score, which offers a maximum of 100 points. Two authors, independently assigned Modified Coleman scores and reached a consensus on the final score.

The risk of bias (ROB) for each included study was assessed by 2 independent reviewers using design-specific ROB tools (ROB 2 for RCTs, ROBINS-I (Risk of Bias in Nonrandomized Studies of Interventions) for nonrandomized studies of interventions.^20,21

Descriptive synthesis was performed: data were systematically taken out, tabulated, and narratively abstracted to determine overall patterns, similarities, and differences among studies. Reported effect sizes and confidence intervals, where they existed, were taken directly from the original articles. Categorical variables are displayed as frequencies and percentages, while continuous variables are presented as means with their standard deviations. All numerical data have been rounded to one decimal place for enhanced precision. A weighted average was calculated to control for the relative contribution of each study to the overall estimate. Specifically, individual study values were multiplied by their respective weights (sample size) and then combined. This was divided by the sum of the assigned weights, giving a summary measure that captures both the size and precision of each study included.

A pooled (weighted) mean and its 95% confidence interval (CI) were calculated to provide a quantitative synthesis of continuous outcomes across studies. For each included study, the mean (X‾ᵢ), standard deviation (SDᵢ), and sample size (nᵢ) were extracted.

The pooled mean (X‾pooled) was calculated as:X‾pooled=[Σ(nᵢ×X‾ᵢ)]/Σnᵢ

The pooled standard deviation (SDpooled) was calculated as:SDpooled={[Σ((nᵢ−1)×SDᵢ2)]/[Σ(nᵢ−1)]}

The standard error of the pooled mean (SEpooled) was calculated as:SEpooled=SDpooled/(Σnᵢ)

Finally, the 95% confidence interval was obtained as:CI95%=X‾pooled±1.96×SEpooled

This approach allows for the estimation of an overall mean value that accounts for differences in sample size and variability among studies, providing a more reliable and representative measure of central tendency.

3 Results

4 Discussion

This systematic review offers a nuanced examination of the clinical performance, accuracy, and safety profile of imageless RATKA. By rigorously evaluating performance metrics such as alignment accuracy, PROMs, and complication rates, this work contributes to a more critical understanding of how imageless robotic platforms are shaping the evolving landscape of TKA.

Alignment accuracy consistently emerged as a significant strength of imageless RATKA in comparison with conventional manual techniques. Rajasekaran et al. (2024) reported notably smaller deviations in crucial alignment parameters, including the HKA angle, LDFA, and tibial slope, when using the Velys robotic platform. Similar trends were documented by Albelooshi (2023) and Adamska (2023), both of whom found that robotic systems produced fewer outliers in component alignment relative to standard instrumentation. The improved control over posterior tibial slope reported by Eerens (2022) further supports the precision of these systems. Of particular interest, Rajgor et al. (2024) performed a direct comparison between the ROSA imageless system and the widely adopted MAKO image-based system in TKA.²⁹ Their retrospective analysis, encompassing 100 consecutive procedures performed by experienced surgeons, found no significant difference in joint biomechanics restoration, including joint line height, patella height, tibial slope, and posterior condylar offset, between the two platforms. This result underscores the technical equivalence of imageless and image-based approaches regarding component positioning, which is critical for long-term implant function. To support the validity of imageless systems in component placement, it is appropriate to reference the research by Murphy et al., in 2023, comparing the MAKO and OMNIbot platforms.³⁰ The authors proved that the OMNIbot platform had a greater percentage of components placed within the desired femoral coronal range, though both platforms provided great precision to reach the desired range (99.7% vs. 98%, p = 0.04). The same study also recorded that both systems achieved a very high percentage (99%) intraoperative coronal and sagittal alignment goals. The accumulating evidence suggests that imageless RATKA can achieve a high degree of surgical precision, with the potential to positively impact implant longevity and patient satisfaction.

In terms of functional outcomes, most studies demonstrated postoperative improvements in PROMs following imageless RATKA.^31–33 Early advantages were observed in some series, such as the superior FJS and pain relief reported by Alton et al. for the robotic-assisted cohort during the first three months, though these differences tended to equalize by the one-year follow-up.³⁴ Similarly, Adamska et al. found higher KOOS scores in the robotic group in the immediate postoperative period. These findings collectively suggest that robotic assistance may provide certain short-term functional benefits, but long-term superiority over traditional methods remains to be proven and warrants further investigation with high-quality longitudinal data.

One recurrent challenge reported across studies was the increased operative duration associated with imageless RATKA, particularly during the initial phase of adoption. Adamska et al. and Alton et al. directly linked this to the learning curve inherent in integrating new robotic technology into routine practice. Importantly, Burgio et al. (2024) demonstrated that longer operative times did not correspond with a heightened risk of periprosthetic joint infection, suggesting that extended surgical duration in itself may not compromise short-term safety, especially once surgeons progress along the learning curve.³⁵

Complication rates following imageless RATKA were generally low in the analyzed studies, reaffirming the acceptable safety profile of these systems. Nevertheless, surgical site infections, periprosthetic fractures, and soft tissue impingement were observed in isolated cases, highlighting the ongoing need for technical vigilance and careful perioperative management. The low complication rates reported by Burgio et al., in particular, support the safety of imageless RATKA when performed in a well-controlled setting.

Radiological outcomes corroborated the enhanced accuracy associated with robotic-assisted techniques; however, the heterogeneity in radiographic assessment protocols limited the potential for direct inter-study comparison. The studies reviewed illustrate substantial heterogeneity in the parameters used to assess accuracy: some focus on global coronal alignment (HKA), others on specific component angles such as LDFA, PTS, or rotational positioning. This variability complicates cross-study comparisons and limits the ability to draw definitive conclusions about the superiority of one approach over another.

Moreover, even within the framework of robotic-assisted techniques, outcomes may be influenced not only by the system employed but also by the surgeon's chosen alignment philosophy. This introduces an intrinsic bias, as radiographic “accuracy” may differ depending on the intended target rather than purely on the execution of the surgical plan.

Taken together, these findings underscore the need for a standardized and universally accepted set of radiological and functional outcome measures in TKA research. Such harmonization would facilitate comparability across studies and ensure that differences in reported accuracy reflect true technological advancements rather than methodological variability. At the same time, it should be recognized that surgical intent—particularly alignment strategy—remains a major determinant of the final radiographic outcome, and therefore must be accounted for when interpreting results.

A quantitative assessment of publication bias, i.e., funnel plot analysis, was not feasible due to the limited number of studies included. However, it should be mentioned that the risk of publication bias cannot be excluded, particularly in the context of most small, single-center retrospective studies reporting positive results. Also, sensitivity analyses could not be performed in a systematic way; but excluding studies with highest risk of bias or smallest sample sizes, the direction of the overall results remained the same as in the primary findings. Therefore, although the strength of the conclusions is limited by heterogeneity of the studies and potential publication bias, the trends that were found appear adequately dependable within the current evidence.

Several limitations must be acknowledged. First, the overall methodological quality of the included studies was modest, with considerable heterogeneity in patient selection, reporting standards, and outcome measures. Second, our review is susceptible to publication bias and does not capture unpublished data. Third, the diversity of PROMs used across studies may limit the comparability of functional results. Fourth, larger and multicenter investigations using standardized outcome assessment tools would enable a more robust appraisal of imageless RATKA. Fifth, the inclusion of multiple imageless RATKA platforms in the present synthesis could represent a source of bias. Lastly, variability in postoperative protocols across studies complicates efforts to directly compare outcomes.

Future investigations should prioritize randomized controlled trials with rigorous methodological standards and standardized outcome assessments. Furthermore, studies focusing on health economics—such as cost-effectiveness and quality-adjusted life years (QALYs)—will be essential for shaping both clinical practice and policy.

5 Conclusions

In summary, imageless RATKA is associated with notable improvements in alignment accuracy and early postoperative functional outcomes, with a reassuring safety profile. On average, PROMs improved by 40–50 points in the studies included, indicating uniform and clinically relevant functional recovery following imageless RATKA. Although the initial learning curve may contribute to longer operative times, ongoing technical refinement and increased surgeon familiarity are likely to address these limitations. High-quality comparative research remains essential to fully define the clinical role of imageless robotic systems within the broader context of total knee arthroplasty. Finally, it would be desirable for more comparative studies to be conducted between imageless platforms and their direct competitors, namely image-based systems.

Informed consent

Not needed for the type of study (Systematic Review)

Ethical approval

No ethical approval needed for the type of study (systematic Review)

Authors’ contribution

four authors (A.E.M. V.C, LA and G.T.G.) independently reviewed the abstracts, obtaining the full texts for any abstracts that were inconclusive. Any differences between the reviewers were discussed, and if disagreements remained, the senior author (S.M.P.R.) was consulted. The reference lists of the selected articles were manually checked to identify additional relevant studies. All selected studies were then analyzed retrospectively by three authors (A.E.M., G.T.G., R.S.), who extracted and entered the data into an Excel worksheet. Finally, the data sheet was reviewed by four authors (FB., R.S., S.M.P.R., A.E.M.), who reached an agreement on the extracted data. Additionally, the references of the identified papers were searched to find further relevant articles, and all journals were considered. Three reviewers (A.E.M., G.T.G., R.S.) independently conducted the review process, evaluating the full texts of the selected articles to assess their eligibility and gather relevant data. When there was uncertainty regarding whether to include a study, the senior author made the final call. Additionally, three authors (A.E.M., G.T.G., V.C) independently evaluated the risk of bias using standardized criteria. Any disagreements were settled The titles and abstracts were independently screened by two reviewers, A.E.M. and L.A. For abstracts that either met the inclusion criteria or caused uncertainty, full-text articles were obtained. These full texts were subsequently re-evaluated by the same two independent reviewers. Any discrepancies were resolved through assessment by the senior author, S.M.P.R. The methodological quality of each study was assessed using the Modified Coleman Score, which offers a maximum of 100 points. Two authors, A.E.M. and G.T.G., independently assigned Modified Coleman scores and reached a consensus on the final score.

Funding

No funding was received for the present study.