Complications of Alloplastic Graft Materials Used in Cranioplasty: Systematic Review and Network Meta-Analysis

Introduction

Cranioplasty is one of the oldest surgical procedures aimed at repairing cranial defects. The first recorded cranioplasty was performed around 3000 BC using gold plates. Later, in 1668, a Dutch surgeon reported repairing a cranial defect in a Russian soldier using bone derived from a canine skull [1]. Since then, neurosurgeons worldwide have used various grafting materials to restore cranial defects.

Cranioplasty is performed for functional and aesthetic purposes, primarily to protect the underlying brain. However, this procedure has a higher complication rate, ranging from 25% to 50%, compared with routine neurosurgical procedures, which have a complication rate of approximately 1% to 5% [2–4]. The 2 most common complications associated with cranioplasty are infection and graft resorption, highlighting the importance of selecting appropriate graft materials [5–8]. Other complications include hematoma formation, delayed healing, cerebrospinal fluid leakage, poor cosmetic outcomes, prolonged surgical duration, extended hospital stays, and high costs [9–12]. The increasing demand for cost-effective and efficient alloplastic graft materials has led to the introduction of various alternatives in the market.

Several alloplastic materials have been recently introduced, claiming to reduce the complications commonly associated with autogenous bone grafts. The most widely used alloplastic materials for cranioplasty are polymethylmethacrylate (PMMA), titanium (Ti), polyetheretherketone (PEEK), and hydroxyapatite cement (HA). Ideally, alloplastic materials should possess biomechanical and biochemical properties that allow them to replace lost tissue, be chemically inert, and fit cranial defects precisely [13–15]. Additionally, they should not interfere with imaging modalities such as magnetic resonance imaging and computed tomography scans. The choice of graft material depends on multiple factors, including patient age, pathology, and location of the cranial defect [16–18].

Approximately 2 to 3 million patients each year require bone grafts to repair bony defects caused by underlying diseases [19–21]. Consequently, there is a growing demand for developing patient-specific, biocompatible alloplastic materials. Economic factors significantly influence the production of cranioplasty materials, and the global cranial implant market is projected to reach approximately $1.4 billion by the end of 2025 [22–25]. The demand is driven by complications associated with existing materials and their varying properties. However, the inconsistent clinical outcomes of different grafting materials make it challenging to determine which material results in the fewest complications.

Existing systematic reviews and meta-analyses generally favor alloplastic materials over autogenous bone grafts [24,26]. However, these reviews have limitations, including reliance on literature reviews without meta-analysis, due to heterogeneity among studies or using fixed-effect models rather than random-effect models when analyzing complications associated with these surgeries. Most meta-analyses have directly compared materials, limiting their ability to evaluate individual materials comprehensively [7,24]. Network meta-analysis, which allows for direct and indirect comparisons, provides a more robust method for assessing the efficacy of various alloplastic materials [27]. While previous network meta-analyses have compared materials based on their adaptability to cranial defects, they have not comprehensively analyzed infection rates, implant exposure, overall complications, and re-surgery [7]. Thus, a thorough comparative evaluation of alloplastic materials used in cranioplasty remains lacking.

This systematic review and network meta-analysis aims to compare post-cranioplasty complications, including infection, implant exposure, and re-surgery, across different alloplastic materials to identify the most clinically effective and biocompatible options. The findings will provide valuable evidence to help neurosurgeons, cosmetic surgeons, and reconstructive surgeons select the most appropriate material for cranioplasty, ensuring patient-specific choices that minimize complications.

Material and Methods

SEARCH STRATEGY AND ARTICLE SELECTION:

We searched PubMed, Web of Science, Embase, and Scopus for articles published from January 2015 till January 2025. The protocol was prepared according to the PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analysis) guidelines and the PRISMA extension for network meta-analysis [28,29] (Figure 1). The protocol was submitted on the PROSPERO (reg No. CRD42024503879).

The search terms were combined under various sub-headings with commonly used alloplastic materials for cranioplasty surgery. Studies published in English in peer-reviewed journals were included. Two authors (AS and SQ) screened the title and abstracts independently. Discrepancies during the screening process were resolved by consulting a third author (MA).

INCLUSION AND EXCLUSION CRITERIA:

Of studies that reported complications of mainly infection, implant exposure, and re-surgery associated with using alloplastic materials following cranioplasty surgeries for patients over 18 years of age, the studies that reported and compared 2 or more alloplastic materials were included. Studies reporting fewer than 5 patients and examining only 1 material were excluded. Case reports, case series, pediatric populations, and reviews were excluded. Studies published before 2015 and in languages other than English were also excluded.

DATA EXTRACTION:

Extracted data included authors, year of publication, country, study design, material used, comparison of material, follow-up, indication of cranioplasty, type of complications, re-surgery, and cosmetic outcomes. Only outcomes of interest that were sufficient to construct network meta-analysis were included in data extraction. These outcomes were mainly related to infection, overall complications, implant exposure, and re-surgery.

RISK OF BIAS:

The Newcastle-Ottawa Scale (NOS) was used to measure the quality of observational studies included in the current review. Scores for NOS were below the standards measured by the Agency for Health Care Research and Quality Services. The NOS is designed explicitly for non-randomized studies, including cohort and case-control studies [30]. It evaluates bias across 3 domains: selection of participants (maximum of 4 stars), comparability of groups (maximum of 2 stars), and outcome assessment (maximum of 3 stars).

DATA SYNTHESIS:

Individual materials were compared, and the meta-analysis was performed, as previous studies have not compared the surgical outcomes of alloplastic graft materials. The absolute risk of binary outcomes such as infection, complication, and healing problems was calculated to obtain the data. Second, relative risk was calculated with a 95% confidence interval (CI) for the binary outcomes. For continuous outcomes such as length of hospital stay and duration of surgery, standardized mean difference was considered. A P value <0.05 was considered statistically significant.

The relative risk ratio for overall complications was calculated using the random-effect model to obtain a pooled point estimate with a 95% CI. A network meta-analysis for pooled data was performed using the frequentist model. This model uses a P-score that represents the chance that one material is better than another based on the average of all the materials. A P-score of 1 indicates a good material, while a P-score of 0 indicates the chance of complications [31]. The ranking and P-score of treatment should be interpreted cautiously, as CIs might coincide, increasing the chances of bias. Moreover, network meta-analysis does not provide significant differences between the materials. Hence, it is important to create a table to compare the materials. A random-effects model was used for network meta-analysis.

Network meta-analysis for randomized control trials was done separately, and heterogeneity was assessed using I2. Low heterogeneity was measured for values between 0% and 25%, moderate, between 25% and 50%, and substantial, over 50%. Data was analyzed using R software (version 4.0.7) with a metafor package [31].

For every network analysis, the direct and indirect evidence were placed parallel, and transitivity was assessed for potential effects in the studies. Meta-regression analysis was performed to assess the moderators, such as follow-up time and infections, by combining the outcomes of the included studies. Multi-collinearity was assessed using intercorrelation analysis. Network graphs were prepared to evaluate the geometry of the material, where the thick line indicates direct comparison, and the dotted line represents the number of multi-arm studies.

Results

RISK OF BIAS:

Risk of bias assessment using the Newcastle-Ottawa Scale revealed that eight studies (33%) had low risk, 13 (54%) moderate risk, and 3 (13%) high risk. While most evidence was of acceptable quality, conclusions are tempered by methodological limitations in over half the studies, necessitating cautious interpretation (Table 1).

OVERALL COMPLICATIONS:

Sixteen included studies reported overall complications, with an incidence rate of 26.79% (629/2348) with all alloplastic materials [4,11,18,32–45]. However, the highest rate of complications was associated with PMMA and Ti, compared with other materials. Resorption of bone was reported in a few studies, but it was not considered a complication. The overall incidence of complications associated with different alloplastic materials was PEEK, 6.39% (41/641); PMMA, 28.2%(134/461); HA, 7.5% (14/186); Ti, 16.6% (108/637), polypropylene, 10% (3/34) and 3D-printed Ti and PEEK materials, 5% and 2%, respectively (Table 1). No network meta-analysis or meta-analysis was performed for overall complications, due to heterogeneity among the included studies.

INFECTION:

Infection with alloplastic craniofacial material was reported in 21 studies [4,7,11,18,33–40,42–50]. For PEEK materials, 6.41% (44/641) of patients had an infection following cranioplasty surgery, compared with 13.8% (144/1096) for other alloplastic materials combined (Table 1). Statistically significant mean differences (P value=0.005) favoring specific materials were observed in several studies (Figure 2A). In multiple studies, Ti outperformed PEEK, with mean differences indicating superior performance.

Figure 2B shows the Surface Under the Cumulative RAnking curve (SUCRA) plot [51], which illustrates the relative ranking probabilities of 7 biomaterials – PEEK, ultra-high-molecular-weight polyethylene (UHMWPE), Ti, HA, autologous material, hard PMMA (hPMMA), and porous PMMA (pPMMA) – in terms of clinical performance based on network meta-analysis. The x-axis denotes the ranking positions (from 1=best to 7=worst), while the y-axis represents the cumulative probability that a given material occupies a particular rank or better. Each curve corresponds to a specific material, and the gradient bar on the right visualizes SUCRA values, ranging from 0% (worst performance) to 100% (best performance). PEEK and UHMWPE exhibit the highest SUCRA values (approaching 100%), with steep cumulative curves that rapidly approach a probability of 1 by rank 1 and 2. These findings suggest that the minimum incidence of infection is reported in materials like PEEK and UHMWPE. Ti and HA have mid-range SUCRA values (approximately 50%–60%), indicating moderate clinical performance. hPMMA and pPMMA have the lowest SUCRA values (close to 0%), with their curves skewed toward the higher ranks (higher infection), suggesting these materials are least likely to be used as cranioplasty materials.

NETWORK META-REGRESSION ANALYSIS:

Figure 2C presents the results of a network meta-regression, assessing the influence of a covariate (follow-up duration, age, or defect size) on the relative treatment effects of different biomaterials, compared with PEEK. The y-axis displays the relative effect (mean difference) vs PEEK, while the x-axis shows the covariate value. The regression lines represent the relationship between covariate values and treatment effects for each biomaterial. The blue line (Ti) crosses the x-axis and approaches a zero mean difference at higher covariate values, suggesting its relative performance improves over time or with greater covariate presence. Autologous material (red line) maintains a moderate slope, indicating a stable but consistently inferior performance relative to PEEK. pPMMA and hPMMA (cyan and green lines) maintain greater mean differences across all covariate values, suggesting these materials are consistently less effective than PEEK, regardless of covariate shifts. UHMWPE (magenta), despite strong baseline performance, shows an increasing mean difference with covariate increase, potentially indicating diminished benefit in specific subgroups. The meta-regression reveals that the relative effectiveness of materials, compared with PEEK, may be moderated by contextual or patient-specific covariates, highlighting the importance of considering such factors when choosing biomaterials. Ti appears more favorable at higher covariate values, while pPMMA and hPMMA consistently underperform. Figure 2D indicates network analysis representing a strong association between the alloplastic materials used in the studies.

EXPOSURE:

Implant exposure was studied in 13 studies [4,11,12,33,35–38,42–44,46,48,50,52,53], of which 11 used Ti as an alloplastic material [7,11,32,33,35–37,42–44,46,48,50]. Of the patients in the included studies, only 2 patients who had undergone cranioplasty with PEEK materials had implant exposure, while approximately 20 patients with Ti had implant exposure. Figure 3A presents a network meta-analysis forest plot comparing the efficacy of the biomaterials PEEK, pPMMA, UHMWPE, hPMMA, and Ti based on their clinical performance or outcome measure (implant exposure). The analysis ranks these materials based on their mean difference estimates, 95% CIs, and predictive intervals. PEEK emerged as the top-performing material (rank 1), showing the most favorable effect size across comparisons, with mean differences ranging from −1.00 to −4.00, with wide CIs (−13.66 to 11.86) (Figure 3B), indicating statistical non-significance in some pairwise comparisons. pPMMA was ranked second (rank 2), demonstrating moderate effectiveness, particularly when compared with hPMMA and Ti, with mean differences ranging from −1.00 to −5.00. Some comparisons show narrower CIs, suggesting more consistent outcomes. UHMWPE and hPMMA ranked third and fourth, respectively, with mixed results and overlapping CIs. Notably, UHMWPE exhibited a mean difference of −2.00 (−16.96 to 12.96) when compared with PEEK, indicating variability in performance. Ti was the lowest-ranked material (rank 5), consistently associated with less favorable outcomes, as seen with mean differences ranging from −2.00 to −6.00 in comparisons with other materials, and with CIs suggesting potential inferiority (−14.08 to 10.88) (Figure 3C).

The SUCRA scores [51] (Figure 3D) and the network ranking suggest that materials like PEEK and pPMMA have lower chances of implant exposure, while Ti exhibits higher chances of implant dehiscence. The wide predictive intervals in some cases imply underlying heterogeneity or limited sample sizes, recommending cautious interpretation.

RE-SURGERY:

The incidence of re-surgery was reported in 10 studies [32,33,36,37,39,40,42–44,48–50], and the highest incidence was reported with pPMMA alloplastic material. Figure 4A shows the forest plot for comparative analysis of treatment effects on re-surgery. The plot displays effect estimates (point estimates) with their corresponding 95% CIs for re-surgery with various treatments.

Figure 4B presents the baseline risk plot for all treatment arms across the included studies. The baseline risk in the reference treatment arm is shown on a logarithmic scale to cover the wide range of values. Horizontal bars indicate 95% CIs derived from the outcome and standard error. The red dashed line indicates the median baseline risk across all studies. This plot visually compares baseline event rates between studies and material types, highlighting potential heterogeneity and supporting the interpretation of comparative treatment effects in network meta-analysis. pPMMA (P-score=0.79) and HA (P-score=0.73) groups demonstrated higher chances of re-surgery across multiple studies, while Ti exhibited greater variability and included studies with lower baseline risks and minimal chances for re-surgery.

Figure 4C represents point estimates (0.62 for Ti) with different ranges (0.05 to 5.71) showing 95% CIs followed by hPMMA and PEEK. The wide CIs suggest uncertainty in effect estimates, possibly due to small sample sizes or heterogeneous studies.

Discussion

OVERALL COMPLICATIONS:

Overall complications were studied in 16 studies [32,33,36,37,39,40,42–44,48–50], including studies reporting minimum bone resorption among alloplastic materials. Among all the materials included, HA has minimal or no reported bone resorption, indicating osteointegration properties of the materials. While 3 studies [7,11,32] including HA as an alloplastic material showed promising results, more studies should be conducted to support their findings. Minor bone resorption is common with all materials, which is acceptable. However, previously published literature supports minimal bone resorption with alloplastic materials [2,5,25,52,54,55]; this complication was not considered in the present network meta-analysis.

The complications associated with different alloplastic materials were highest in PMMA, at 28.2% (134/461), and Ti, at 16.6%. However, no significant difference was reported in the overall complications of the different materials. Similar findings were reported in the network meta-analysis of all the cranioplasty materials. Conversely, a review on PMMA materials with autologous bone graft reported significantly fewer complications. This could be because the present review includes only alloplastic materials, not autologous bone grafts.

INFECTION:

The pooled analysis revealed that both pPMMA and hPMMA were associated with higher infection rates than were PEEK and Ti. This was reflected in the SUCRA rankings, whereby pPMMA and hPMMA displayed values close to 0%, indicating a greater likelihood of postoperative infection and suggesting limited suitability for cranioplasty. The SUCRA curves for these materials were skewed toward higher infection incidence, reinforcing this trend.

However, this finding should be interpreted cautiously, as the number of patients receiving pPMMA and hPMMA implants was relatively small, potentially limiting the statistical power and generalizability of the results [7,56]. Furthermore, despite these trends, no statistically significant difference in infection rates was observed across the included studies, highlighting the need for more robust comparative data.

Notably, PEEK, Ti, and HA ranked highest for minimal infection risk, consistent with previous meta-analyses. This may be attributed to their favorable osseointegration properties, which promote stable integration with the surrounding bone and reduce the risk of bacterial colonization. PEEK offers a smooth, inert surface that is less prone to biofilm formation, while HA supports bone regeneration and bonding, potentially enhancing resistance to infection. These material characteristics may partly explain their superior performance in infection-related clinical outcomes.

IMPLANT EXPOSURE:

One of the key findings of this network meta-analysis was the significantly higher incidence of implant exposure (or dehiscence) associated with Ti materials, compared with other alloplastic graft materials. This result is consistent with that of previous systematic reviews, which reported higher exposure rates with Ti implants in cranioplasty procedures [7,24,54,55]. In our network meta-analysis, Ti ranked lowest regarding implant exposure outcomes, indicating its relatively unfavorable performance.

The increased risk of implant exposure with Ti may be attributed to its biological and mechanical properties. While biocompatible and widely used in various surgical disciplines, Ti is a rigid and non-resorbable material [41,57]. This rigidity can lead to mechanical stress at the bone-implant interface, especially in areas with thin, soft tissue coverage or in patients with compromised wound healing capacity. Moreover, Ti implants can elicit a localized inflammatory response, potentially resulting in soft tissue breakdown, bone resorption, and subsequent implant dehiscence or exposure [41,57].

Histopathological and clinical studies have also suggested that Ti-induced inflammation can disrupt the delicate balance required for osseointegration and soft tissue integration [33,43,50,58]. This makes implant exposure not only more likely but also more difficult to manage, often necessitating re-surgery.

RE-SURGERY:

There was no significant difference in re-surgery among the alloplastic materials. This result was not similar to the findings of systematic reviews comparing alloplastic and autologous grafts [7]. Studies have suggested that bone resorption is the main reason for re-surgery, which is mainly seen with autologous bone graft materials, and in the present review, these materials were not studied [6,32–34]. A randomized clinical trial reported that Ti and autologous bone graft materials have a higher incidence of re-surgery [41]. However, in the present review, pPMMA material ranked lowest, suggesting that the use of this material led to re-surgery. The inclusion criteria of the present review limited the number of studies evaluating re-surgery to 10, and the higher incidence of re-surgery was reported in the studies with immediate cranioplasty. More studies are required to confirm the relationship between re-surgery and different materials used for cranioplasty.

LIMITATIONS:

The present systematic review and network meta-analysis has several limitations, including the studies’ low to moderate quality. All the included studies were observational (retrospective and prospective cohort), subjecting them to confounding bias. Additionally, reports on newer alloplastic materials such as 3D-printed PEEK and Ti were limited and sparse; hence, bias was seen in the analysis of the included studies. Most of the included studies were rated as having moderate to high risk of bias, particularly due to issues such as inadequate control for confounders, lack of blinding, small sample sizes, and insufficient follow-up. Moreover, outcomes of interest explained in studies were defined in different terminologies. Therefore, separating the outcomes of interest into different sub-groups was not possible. Smaller differences in CIs in network meta-analysis can lead to overlapping results. The ranking of materials for outcomes such as infection and re-surgery should be interpreted cautiously.

While the present review compared only 3 outcomes, other outcomes, such as time for surgery, hospital stay, cosmesis, seizures, and cost-effectiveness, should be studied in future meta-analyses. Moreover, studies on newer materials, such as HA and 3D-printed alloplastic materials, were lacking in the literature; hence, future studies should focus on outcomes related to these materials.

Conclusions

In the present network meta-analysis of alloplastic cranioplasty materials and associated postoperative complications such as infection, implant exposure, and re-surgery, moderate heterogeneity was observed across all 3 constructed networks. Overall, PEEK emerged as a more favorable alloplastic graft materials, compared with the others. Due to the novelty and limited data availability, 3D-printed grafting materials were excluded from the network meta-analysis. Randomized controlled trials are necessary to validate these findings and better inform clinical decision-making. Such studies will help determine which alloplastic material offers the best outcomes with minimal postoperative complications.