A Predictive Model for Graft Failure in Femtosecond Laser-Assisted Penetrating Keratoplasty Among Chinese Patients: A 2-Year Study

Penetrating keratoplasty (PKP) stands as a cornerstone in the treatment of various corneal diseases and has undergone significant evolution over the years [1]. The integration of femtosecond laser-assisted penetrating keratoplasty (Fs-PKP) marks a pivotal advancement, demonstrating significant enhancements over conventional PKP techniques in terms of surgical precision and outcomes. The precision afforded by femtosecond laser cutting has been instrumental in enhancing the integration between grafts and hosts, a factor of paramount importance for the success and longevity of corneal transplants [2‐5]. Comparative studies between traditional methods and Fs-PKP have demonstrated a significant reduction in graft rejection and endothelial cell loss rates among patients undergoing Fs-PKP, underscoring the marked benefits of this advanced technology in improving transplant outcomes [2‐5].

Despite these technological advancements, graft failure—characterized by irreversible changes in the corneal stroma such as edema, haze, or scarring that negatively affect visual acuity—remains a major challenge in Fs-PKP. In conventional PKP, graft success hinges on various recipient and donor factors. Issues like active ocular inflammation, corneal neovascularization, and ocular comorbidities (e.g., glaucoma) can elevate graft rejection risks or compromise graft stability due to effects on intraocular pressure. Likewise, donor attributes, notably endothelial cell density and cornea health, critically influence the graft's initial viability and function. Surgical specifics, such as graft size and precise positioning, also play a key role, impacting the graft–host integration. Correct sizing and alignment are essential for smooth graft integration, reducing edge irregularities and potential complications like poor healing. These interconnected factors significantly determine transplant outcomes, emphasizing the need for meticulous risk factor management to enhance PKP success [6‐11].

However, these studies have predominantly targeted individual risk factors, frequently omitting an encompassing evaluation. The emergence of predictive models—integrating varied data types and employing sophisticated statistical and machine learning methods to enhance precision and facilitate individualized risk assessments—has offered new insights. For instance, a study conducted in 2005 aimed at predicting short-term graft survival but was constrained to a 1-year postoperative period and did not incorporate a diverse range of data sources [12]. More recently, a 2021 study in the United States developed a predictive nomogram for conventional PKP, utilizing a larger patient sample and extending the follow-up period up to 5 years [13]. While existing predictive models for PKP boast advancements beyond the scope of traditional epidemiological approaches, their primary design caters to Western populations, focusing predominantly on conventional PKP techniques. Such geographical and technological specificities necessitate a reassessment of these models’ applicability in the context of Fs-PKP, particularly in light of the latest technological improvements and the adoption of precise preventative measures for well-documented risks.

Moreover, the unique demographic and clinical characteristics of Chinese patients, who represent a significant portion of the Fs-PKP recipient base, are not fully captured by models primarily designed for Western populations. Factors such as better overall ocular health, access to more sophisticated surgical techniques, and the economic aspects of accessing state-of-the-art medical care set Chinese Fs-PKP candidates apart from their counterparts considered for traditional PKP. The demographic consistency of the Chinese population, mainly composed of Han Chinese and characterized by a low immigration rate, reduces the relevance of the variability that international models, intended for ethnically and clinically diverse groups, aim to address. This suggests that a predictive model tailored to the specific features of the Chinese population would offer greater accuracy and relevance, thereby improving its utility for this particular group.

This divergence highlights the urgent necessity for developing predictive models that are specifically attuned to the nuances of the Chinese clinical environment, ensuring they are both relevant and applicable to the patient population in China. Therefore, our research aims to fill this critical void by developing and validating a comprehensive nomogram for graft failure in Chinese patients undergoing Fs-PKP. By integrating a broad spectrum of risk factors, we anticipate that our model will more accurately reflect the unique healthcare needs and clinical scenarios of the Chinese demographic, thereby providing a refined instrument for preoperative assessment and informed decision-making in the domain of Fs-PKP.

In this study, we have innovatively developed a nomogram for predicting corneal graft failure in Chinese patients undergoing Fs-PKP. This model, which incorporates factors such as recipients’ systemic autoimmune disorders and ocular trauma history, prior PKP, donors’ diabetes history, and endothelial cell density of donor corneas, uniquely integrates recipient, donor, and surgical variables. Such a comprehensive approach offers clinicians a valuable tool for assessing and mitigating the risk of graft failure.

The observed 1-year and 2-year survival rates for Fs-PKP (71.69 and 68.07%, respectively) are particularly significant given the limited literature in the Chinese demographic [14‐21]. These rates are slightly lower than those reported in the Singapore Corneal Transplant Study for conventional PKP and align with the global survival rate range for this procedure [14‐21]. This discrepancy might reflect the unique socio-economic and healthcare factors in China, underscoring the importance of considering regional differences in healthcare delivery and patient demographics when evaluating corneal transplantation outcomes.

While conventional PKP’s graft failure risk factors are well documented, data on Fs-PKP remain notably scarce [6‐11]. Our research focused solely on the use of the mushroom-shaped incision technique, a method refined by the precision of femtosecond laser technology. This technique provides consistently uniform and regular incisions, crucial for achieving optimal wound alignment. It offers several benefits to standard trephine: it replaces a significant amount of the anterior stromal tissue, reducing corneal astigmatism by enlarging the anterior stromal segment and lowering the risk of graft rejection with a smaller posterior segment [2]. However, the requirement for suction on transparent corneal areas during laser application limits Fs-PKP’s applicability in cases with corneal perforation or severe opacification, where traditional corneal transplantation remains necessary. This limitation highlights the need to reevaluate the risk factors for Fs-PKP, as they might differ from those associated with conventional PKP methods [2‐5, 16‐18, 22, 23]. Our study identified specific risk factors relevant to Chinese patients undergoing Fs-PKP, including systemic autoimmune disorders, history of ocular trauma, prior PKP, diabetes history in donors, and the endothelial cell density of donor corneas. These findings are in line with recognized risk factors from previous studies on PKP [6‐11], thus reinforcing the credibility of our results.

We conscientiously addressed known high-risk factors for graft failure in conventional PKP, such as preoperative infections, elevated intraocular pressure, and systemic autoimmune disorders, driven by ethical imperatives. Despite these targeted efforts, our multivariate Cox regression analysis highlighted that systemic autoimmune disorders remain a significant risk factor for graft failure in Fs-PKP. This highlights the intricate and impactful nature of systemic autoimmune diseases in the context of corneal transplantation. The findings imply that while our strategies effectively managed certain risk factors, systemic autoimmune disorders necessitate more focused and perhaps specialized postoperative care for Fs-PKP patients, underscoring the need for tailored medical attention in these cases.

Interestingly, while anterior chamber hypopyon and corneal neovascularization initially showed significance as risk factors in univariate Cox regression analysis (P values < 0.05), their significance diminished in the multivariate Cox analysis. This suggests that these factors may not independently contribute to graft failure when considered alongside other variables, highlighting the complexity of their influence in corneal transplantation outcomes. This observation may be attributed to the unique nature of Fs-PKP surgery, where patients typically present with less severe forms of these clinical signs [2].

Our study presents a novel nomogram developed and validated for predicting graft failure in patients undergoing Fs-PKP. This nomogram, incorporating five aforementioned risk factors, has demonstrated effective predictive capability, as evidenced by its concordance index (C-index) of 0.72 in the training group and 0.66 in the validation group. Its reliability across various time intervals and clinical scenarios, affirmed by time-dependent ROC curves and calibration plots, reflects its accuracy in mirroring real-world clinical outcomes. The clinical utility of the nomogram, underscored by DCA, confirms its substantial net benefit in diverse clinical settings, reinforcing its role in effectively guiding the management of Fs-PKP patients. This innovation marks a significant contribution to the field of ophthalmic surgery in China, offering precise prognostic insights in an area where the application of predictive models has been limited [11, 24, 25]. While a renowned predictive model from the United States for conventional PKP, identified eleven variables significantly associated with graft failure. These variables encompass a range of conditions, from active microbial infection at the time of PK to the presence of intraocular silicone oil and extensive corneal neovascularization. However, our findings reveal that the severe conditions highlighted in previous studies, such as intraocular silicone oil and widespread corneal neovascularization, are notably absent in our Fs-PKP patient cohort. This discrepancy could be attributed to the relatively milder ocular conditions of patients selected for Fs-PKP. Additionally, guided by ethical considerations, our approach included preemptive interventions for specific, well-defined risk factors (e.g., active microbial infection at the time of PK, glaucoma) prior to surgery. Given these observations, we posit that the nomogram developed for traditional PK may not be directly applicable to predicting outcomes in Fs-PKP patients. This assertion is supported by our model’s distinct focus and its adaptation to the unique clinical profiles observed in the Chinese Fs-PKP patient demographic. Consequently, our study not only fills a critical gap in the existing literature but also highlights the need for specialized predictive models that cater to the specific requirements of different surgical techniques and patient populations. This nuanced approach is essential for optimizing patient outcomes in the realm of corneal transplantation surgery.

A key strength of our study is its distinction as the most extensive research focused on developing a clinical predictive model for corneal transplant failure specifically in Chinese patients undergoing Fs-PKP. We conducted an in-depth analysis of risk factors contributing to graft failure, covering a broad spectrum that encompasses recipient characteristics, donor attributes, and surgical variables. Moreover, our study extends beyond the mere development of the model, emphasizing rigorous validation of its multidimensional effectiveness. This thorough methodology enhances the model’s clinical applicability and reliability, making a significant contribution to the domain of ophthalmic transplantation within the Chinese healthcare context.

Our study represents one of the most comprehensive investigations into Fs-PKP within the Chinese demographic, despite its limitations arising from a modest cohort size. This limitation largely stems from the nascent stage of Fs-PKP technology adoption in China, which restricts the pool of suitable candidates and the length of the feasible follow-up period. Recognizing the importance of early dissemination, we consider it crucial to publish our data now, even though our follow-up period is limited to 2 years. Our study’s initial findings, prompted by its innovative contributions and potential clinical relevance, merit prompt sharing. However, as a single-center study, the specificity of our clinical practices and patient demographics might limit the broader applicability of our results across different geographical and clinical settings. In acknowledging these limitations, we emphasize the preliminary yet significant value of our work, laying the groundwork for future research in this area.

To optimize the application of Fs-PKP within clinical settings, it is imperative that future investigations aim to surmount the existing methodological limitations. A pivotal step in this direction involves the augmentation of study cohorts and the elongation of observational periods. Such measures would permit a more exhaustive appraisal of Fs-PKP’s long-term therapeutic efficacy and safety profile. Extending follow-up durations is crucial for the accurate delineation of the sustained impacts of Fs-PKP and the vigilant surveillance for any late-onset adverse effects. Moreover, the intersection of deep learning and artificial intelligence methodologies with clinical datasets, inclusive of imaging modalities and diverse multi-modal data, represents a promising frontier for enhancing prognostic models associated with Fs-PKP. The application of these sophisticated analytical tools can significantly improve the precision and dependability of outcome predictions for Fs-PKP by facilitating the nuanced identification of patient-specific attributes and response patterns. This approach heralds the advent of tailored therapeutic strategies, offering nuanced decision-making support for the administration of Fs-PKP. Further, building on the foundational work of nomograms, there exists an opportunity to construct and empirically validate network-based scoring frameworks aimed at the pre-treatment appraisal of risk and therapeutic outcome probabilities. The development of such scoring systems promises to enrich clinicians’ insights into patient-specific prognoses and underpin informed clinical decision-making. The establishment of empirically validated frameworks would not only fortify the decision-making process but also contribute to the refinement of Fs-PKP treatment paradigms. Importantly, these scoring systems would enable the robust validation and reproducibility of research findings, thereby reinforcing the evidence base and facilitating the broader adoption of Fs-PKP in clinical practice.

Our study has successfully developed and validated a predictive model specifically designed for Chinese patients undergoing Fs-PKP. By incorporating critical factors, including the recipient's autoimmune history, history of ocular trauma, prior PKP, the donor's history of diabetes, and the endothelial cell density of the donor corneas, our model represents a significant advancement in the field of corneal transplantation. This development markedly enhances the preoperative prediction and management strategies for Chinese patients at risk of Fs-PKP graft failure, offering a tailored approach to mitigate potential complications and improve surgical outcomes.