Cranioplasty with hydroxyapatite or acrylic is associated with a reduced risk of all-cause and infection-associated explantation

Abstract

Objective

Cranioplasty remains an essential procedure following craniectomy but is associated with high morbidity. We investigated factors associated with outcomes following first alloplastic cranioplasty.

Methods

A single-centre, retrospective cohort study of patients undergoing first alloplastic cranioplasty at a tertiary neuroscience centre (01 March 2010–01 September 2021). Patient demographics and craniectomy/cranioplasty details were extracted. Primary outcome was all-cause explantation. Secondary outcomes were explantation secondary to infection, surgical morbidity and mortality. Multivariable analysis was performed using Cox proportional hazards regression or binary logistic regression.

Results

Included were 287 patients with a mean age of 42.9 years [SD = 15.4] at time of cranioplasty. The most common indication for craniectomy was traumatic brain injury (32.1%, n = 92). Cranioplasty materials included titanium plate (23.3%, n = 67), hydroxyapatite (22.3%, n = 64), acrylic (20.6%, n = 59), titanium mesh (19.2%, n = 55), hand-moulded PMMA cement (9.1%, n = 26) and PEEK (5.6%, n = 16). Median follow-up time after cranioplasty was 86.5 months (IQR 44.6–111.3). All-cause explantation was 12.2% (n = 35). Eighty-three patients (28.9%) had surgical morbidity. In multivariable analysis, the risk of all-cause explantation and explantation due to infection was reduced with the use of both hydroxyapatite (HR 0.22 [95% CI 0.07–0.71], p = .011, HR 0.22 [95% CI 0.05–0.93], p = .040) and acrylic (HR 0.20 [95% CI 0.06–0.73], p = .015, HR 0.24 [95% CI 0.06–0.97], p = .045), respectively. In addition, risk of explantation due to infection was increased when time to cranioplasty was between three and six months (HR 6.38 [95% CI 1.35–30.19], p = .020). Mean age at cranioplasty (HR 1.47 [95% CI 1.03–2.11], p = .034), titanium mesh (HR 5.36 [95% CI 1.88–15.24], p = .002), and use of a drain (HR 3.37 [95% CI 1.51–7.51], p = .003) increased risk of mortality.

Conclusions

Morbidity is high following cranioplasty, with over a tenth requiring explantation. Hydroxyapatite and acrylic were associated with reduced risk of all-cause explantation and explantation due to infection. Cranioplasty insertion at three to six months was associated with increased risk of explantation due to infection.

Keywords:

• Cranioplasty
• morbidity
• mortality
• titanium
• acrylic
• hydroxyapatite

Introduction

Craniectomy continues to be both necessary and effective in modern neurosurgical practice and is increasingly performed for the purposes of cranial decompression in the context of traumatic brain injury (TBI), stroke, subarachnoid haemorrhage and venous sinus thrombus.¹ Craniectomy may also be performed when a bone flap is not viable, for instance, in the context of infiltrative tumours or post-craniotomy bone flap infection. Cranial reconstruction is usually performed following craniectomy but is associated with high rates of complications² including infection,^3,4 hydrocephalus,^5,6 altered cerebrospinal fluid (CSF) dynamics,⁷ and cosmetic intolerance.⁸ Infection often necessitates cranioplasty removal and disposal, which in turn requires a second cranioplasty to restore cranial integrity.

Reducing the rate of explantation of cranioplasty for any reason would improve patient care. Debate continues regarding factors associated with morbidity after cranioplasty (and in particular explantation secondary to infection) and has focused primarily on the choice of alloplastic material.^2–4,9–14 Autologous cranioplasty is generally considered to be an inferior choice.^2,4,10,12

Patient factors and the surgeon’s preference guide the selection of alloplastic material, but each material has its own advantages and disadvantages.^3,14 Custom-made ceramic cranioplasty made with hydroxyapatite (HA) demonstrated a lower rate of infection, but a higher risk of postoperative epidural haematoma when compared to titanium in a randomised controlled trial.¹¹ HA provides a scaffold for supporting bone regeneration and osteointegration, but this advantage comes at the cost of strength, and concern has been reported regarding implant fracture.¹⁵ Titanium on the other hand offers excellent strength, but is reported to be associated with late wound complications, as well as high rates of infection.⁹ Other popular materials include thermoplastic polymers such as polyetheretherketone (PEEK), porous polyethylene (PP) and polymethyl methacrylate (PMMA), but these show limited to no osteointegration.¹⁴ PMMA is utilised in both a prefabricated, custom-designed form (pPMMA) commonly referred to as acrylic, and an ‘off the shelf’ hand-moulded cement form (hPMMA).

The aim of this study was to identify factors associated with surgical morbidity and mortality outcomes following first cranioplasty (utilising any alloplastic material) performed to restore a craniectomy defect (for any indication).

Methods

Study design and setting

This was a single-centre, retrospective cohort study, performed at The Walton Centre NHS Foundation Trust, Liverpool, United Kingdom, a tertiary neuroscience centre which serves a network of 18 hospitals and a population of 3.5 million people. This work falls within the remit of a clinical outcomes service evaluation, and therefore individual patient consent was not required. Institutional review board approval was obtained in order to conduct this study prior to patient identification.

Participants

This study included adult patients (inclusive from 16 years old) who had a first alloplastic cranioplasty operation between 1 January 2010 and 1 September 2021. There was no restriction based on indication for craniectomy, geographical location of craniectomy operation (for instance, those patients undergoing cranioplasty at our institution following repatriation from another neurosurgical centre) or choice of alloplastic cranioplasty material used. Patients were excluded if sufficient clinical data were not available to allow evaluation of craniectomy and cranioplasty associated factors and study outcomes. In addition, patients were excluded if more than one alloplastic material was used at the same time, or if a material was used infrequently (defined as < five occasions), as statistical analysis of the study outcomes would not be possible.

Potentially eligible patients were identified following an electronic search of two prospectively maintained databases, specifically an operative case log with entries categorised by operation type and a surgical implant log. Both databases were searched with the term ‘cranioplasty’ to identify records within the study period. Records were deduplicated and stored in a password protected database in Microsoft Excel^® on a hospital server (Microsoft Inc., Seattle, WA).

Study data

Study data for each patient were extracted from available paper and electronic clinical patient records and imaging. This included baseline and demographic data (age, sex and comorbidities at time of craniectomy), craniectomy variables (date of operation, indication, status as elective/emergency, craniectomy site, use of duroplasty material and closure technique), cranioplasty variables (date of operation, material utilised, seniority and number of operating surgeons, soaking of cranioplasty material in antibiotic solution, use of wound drain, need for CSF diversion and closure technique) and outcome data (complications, need for further surgery, date of further surgery, last-follow-up and alive/deceased status). Complications were categorised as either surgical, neurological or medical. Surgical complications were classified according to the Therapy-Disability-Neurology (TDN) grading system; a novel adverse events classification tool that considers not only the therapy required to treat the adverse event, but associated disability and resultant neurological deficits.¹⁶ Timing and duration of surgical complications were recorded (occurred within 30 days of cranioplasty and resolved, occurred within 30 days of cranioplasty but persisting and occurred beyond 30 days after cranioplasty). For patients with more than one complication, the event that produced the highest score was used for classification.

Study outcomes

The primary outcome was all-cause explantation of a first alloplastic cranioplasty. Secondary outcomes were surgical morbidity after first alloplastic cranioplasty, alloplastic cranioplasty explantation due to infection and overall mortality.

Data analysis

Data analyses were performed using SPSS v24.0 and R v3.5.0. Baseline and demographic data, craniectomy variables and cranioplasty variables were summarised for the study cohort as a whole. The number of patients (with percentages) for each factor was calculated. Categorical variables are summarised with frequencies and percentages. Normally distributed variables are expressed as mean (standard deviation [SD]), whereas skewed variables are expressed as median (interquartile range [IQR]).

To identify factors associated with the primary and secondary outcomes, univariable analyses were first performed, with variables achieving a p value < .1 carried forward to multivariable analysis. Factors associated with surgical morbidity after first alloplastic cranioplasty were assessed with binary logistic regression analysis. For all other outcomes (explantation of first alloplastic cranioplasty, explantation due to infection and overall survival), a Cox proportional hazards model was used. Variables were considered significant in multivariable analysis with a p value < .05. Kaplan–Meier curves were used to demonstrate these outcomes graphically.

Results

Baseline characteristics and operative details of included patients

Two hundred and eighty-seven patients were included for analysis, out of 308 identified as potentially eligible. Twenty-one patients were excluded from the study and subsequent analysis for the following reasons: incorrect operation type (n = 6), insufficient data to analyse study outcomes (n = 6), cranioplasty prior to study start date (n = 4), autologous cranioplasty (n = 2), novel material used only once (n = 1), patient who received both titanium mesh and PMMA cement simultaneously (n = 1) and a patient who had a long-standing history of chronic osteomyelitis as a confounding factor for the analysis of morbidity outcomes (n = 1).

One hundred and eighty-three patients (63.8%) were male, and mean age at time of craniectomy was 42.9 years (SD = 15.4). The most common indications for craniectomy included: TBI (32.1%, n = 92), tumour (any pathology) (23.7%, n = 68) and infected bone flap (18.5%, n = 53). The demographics, baseline characteristics and operative details for the cohort at the time of craniectomy are summarised in Table 1. At time of cranioplasty, the mean age was 44.1 years (SD = 15.2) and the median Age-adjusted Charlson Comorbidity Index (ACCI) was 0 (range: 0–9). The majority of patients’ American Society of Anaesthesiologists (ASA) score at time of cranioplasty was grade 1 (n = 158, 55.1%) or grade 2 (n = 108, 37.6%). The median time from craniectomy to cranioplasty was seven months (IQR 1–15).

Table 1. Craniectomy details.

Variable	Sub-variable	Total number (%)
Number of patients		287
Mean age in years [SD]		42.9 [15.4]
Sex	Male	183 (63.8)
	Female	104 (36.2)
Indication for craniectomy	Traumatic brain injury	92 (32.1)
	Tumour (any pathology)	68 (23.7)
	Infected bone flap	53 (18.5)
	Cerebral infarct	23 (8)
	Intracerebral haemorrhage	13 (4.5)
	Subarachnoid haemorrhage	11 (3.8)
	Cystic lesion	9 (3.1)
	Primary intracranial infection	5 (1.7)
	Other	13 (4.5)
Operative details
Urgency of surgery	Elective	128 (44.6)
	Urgent	152 (53)
	Unknown	7 (2.4)
Craniectomy site	Right hemispheric	78 (27.2)
	Left hemispheric	65 (22.6)
	Right hemicraniectomy	52 (18.1)
	Left hemicraniectomy	41 (14.3)
	Bifrontal	27 (9.4)
	Occipital	7 (2.4)
	Left retrosigmoid	1 (0.3)
	Midline/Vertex	1 (0.3)
	Unknown	15 (5.2)
Duroplasty use	Yes	77 (26.8)
	No	187 (65.2)
	Unknown	23 (8)
Duroplasty material	TissuDura	28 (9.8)
	Spongostan	14 (4.9)
	DuraGen	13 (4.5)
	Resodura	6 (2.1)
	Unknown	16 (5.6)
Closure technique	Monofilament	214 (74.6)
	Clips	49 (17.1)
	Unknown	24 (8.4)

Demographics, baseline characteristics and operative details for the cohort at the time of craniectomy. With respect to indication for craniectomy, tumour includes both benign and malignant lesions whereby craniectomy is not secondary to infection, whereas infected bone flap includes those patients who have undergone craniotomy and subsequently required craniectomy to manage the infection, primary intracranial infection includes both subdural empyema and abscess.

Alloplastic cranioplasty materials included: titanium plate (23.3%, n = 67), hydroxyapatite (22.3%, n = 64), acrylic (20.6%, n = 59), titanium mesh (19.2%, n = 55), hand-moulded PMMA cement (9.1%, n = 26) and PEEK (5.6%, n = 16). The demographics, baseline characteristics and operative details for the cohort at the time of cranioplasty are summarised in Table 2.

Table 2. Cranioplasty details.

Variable	Sub-variable	Total number (%)
Mean age in years [SD]		44.1 [15.2]
Median ACCI [Range]		0 [0–9]
ASA	I	158 (55.1)
	II	108 (37.6)
	III	17 (5.9)
	IV	2 (0.7)
	Unknown	2 (0.7)
Smoking status	Yes	65 (22.6)
	No	216 (75.3)
	Unknown	6 (2.1)
Time to cranioplasty	Median time in months [IQR]	7 [1-15]
	Primary ‘on-table’ cranioplasty	69 (24)
	Less than 3 months	12 (4.2)
	3–6 months	36 (12.5)
	6–12 months	69 (24)
	Over 12 months	101 (35.2)
Cranioplasty material	Titanium plate	67 (23.3)
	Hydroxyapatite	64 (22.3)
	Acrylic	59 (20.6)
	Titanium mesh	55 (19.2)
	Hand-moulded PMMA	26 (9.1)
	PEEK	16 (5.6)
Seniority of surgeon/s	Only junior/trainee	66 (23)
	Senior surgeon present	198 (69)
	Two senior surgeons present	18 (6.3)
	Unknown	5 (1.7)
Cranioplasty soaked in ABX	Total	24 (8.4)
	Soaked in gentamicin	12 (4.2)
	Soaked in vancomycin	12 (4.2)
	Not soaked/no reference to event	263 (91.6)
Drain usage	Yes	68 (23.7)
	No	214 (74.6)
	Unknown	5 (1.7)
CSF diversion	New shunt	3 (1.0)
	Lumbar drain	2 (0.7)
	Shunt in situ	1 (0.3)
	External ventricular drain	1 (0.3)
	No CSF diversion	275 (95.8)
	Unknown	5 (1.7)
Closure technique	Monofilament	248 (86.4)
	Clips	34 (11.8)
	Unknown	5 (1.7)

Demographics, baseline characteristics and operative details for patients undergoing first cranioplasty between 1 January 2010 and 1 September 2021 at The Walton Centre NHS Foundation Trust.

Morbidity after cranioplasty

All morbidity is summarised in Table 3 and stratified by alloplastic cranioplasty material in Table 4. Median follow-up time after cranioplasty was 86.5 months (IQR 44.6–111.3). Of the entire cohort, 97 patients (33.8%) had at least one recorded complication. One hundred and twenty individual complications were recorded in total, and of these 98 were surgical in nature (81.7%), across 83 patients (28.9%). The most frequent surgical complications were: infection of any description (n = 41), a CSF complication e.g., CSF leak, or need for CSF diversion (n = 26), extra-axial haematoma (n = 14), wound complications (n = 11) and cosmetic complaints (n = 6).

Table 3. Morbidity and mortality.

Variable	Sub-variable	Total number (%)
Total No. of patients with a complication		97 of 287 (33.8)
Mean age in years of patients with complication/s [SD]		46.4 [15.6]
Median follow-up time in months after cranioplasty [IQR]		86.5 [44.6–111.3]
Median ACCI of patients with complication/s [Range]		1 [0–5]
No. of complication/s	1	79 of 287 (29.2)
	2	15 of 287 (5.5)
	3 or more	4 of 287 (1.5)
Type of complication/s	Had a surgical complication	83 of 287 (28.9)
	Had a neurological complication	16 of 287 (5.6)
	Had a medical complication	4 of 287 (1.5)
Total no. and no. of specific surgical complications	Total No. complications	120
	Total No. surgical complications	98 (81.7)
	Any infection	41 of 98 (41.8)
	CSF complications	26 of 98 (26.5)
	Extra-axial haematoma	14 of 98 (14.3)
	Wound complications	11 of 98 (11.2)
	Cosmetic complaint	6 of 98 (6.1)
Total no. of patients who had further surgery		57 of 287 (19.9)
Reason for further surgery	Explantation due to infection	27 of 287 (9.4)
	Explantation not due to infection	8 of 287 (2.8)
	Other surgery	22 of 287 (7.7)
Mortality	Total mortality at follow-up	43 (15.0)
	Within 30 days	3 (1.0)

Demographics, baseline characteristics, morbidity counts, further surgery frequencies and mortality for patients undergoing first cranioplasty between 1 January 2010 and 1 September 2021 at The Walton Centre NHS Foundation Trust, Liverpool, UK.

Table 4. Morbidity stratified by cranioplasty material.

Alloplastic material no. cases	All-cause explantation no. cases (% within material)	Explantation due to infection no. cases (% within material)	Any surgical morbidity no. cases (% within material)
Titanium plate (n = 67)	15 (22.4)	12 (17.9)	23 (34.3)
Hydroxyapatite (n = 64)	4 (6.3)	3 (4.7)	16 (25)
Acrylic (n = 59)	3 (5.1)	3 (5.1)	13 (22.0)
Titanium mesh (n = 55)	8 (14.5)	5 (9.1)	18 (32.7)
Hand-moulded PMMA (n = 26)	2 (7.7)	1 (3.8)	7 (26.9)
PEEK (n = 16)	3 (18.8)	3 (18.8)	6 (37.5)
Entire cohort (n = 287)	35 (12.2%)	27 (9.4)	83 (28.9)

Cases of explantation, explantation due to infection and number of patients experiencing any surgical morbidity event stratified by cranioplasty material. Percentages are calculated to demonstrate the proportion of each event for each material; for instance 15 titanium plates were explanted for any reason out of a total of 67 inserted (22.4%). Overall summary figures are provided at the bottom of the table, with percentages representing the proportion of the total cohort.

Fifty-seven patients (19.9%) required further surgery, of which 35 were cranioplasty explantations (12.2%), with 27 of these due to infection (9.4%). The reasons for explantation other than infection included; extra-axial haematoma evacuation (n = 4), brain swelling (n = 1), wound erosion (n = 1), devascularised incision limb (n = 1) and persistent headaches (n = 1).

All-cause explantation and explantation due to infection, stratified by cranioplasty material are summarised in Table 4. When considering those materials utilised most frequently in this cohort, both hydroxyapatite and acrylic had favourable rates of all-cause explantation (n = 4, 6.3% and n = 3, 5.1%, respectively) and explantation due to infection (n = 3, 4.7% and n = 3, 5.1%, respectively).

Twenty-two patients (7.7%) underwent additional surgery for reasons other than explantation of a cranioplasty which included: ventriculoperitoneal shunt insertion (n = 8), external ventricular drain insertion (n = 3), pseudomeningocele repair (n = 2), extra-axial haematoma evacuation (n = 2), reoperation for meningioma (n = 1), lumbar drain insertion (n = 1), ventriculoperitoneal shunt revision (n = 1), replacement of titanium screw (n = 1), exploration and resuspension of temporalis muscle (n = 1), revision of eroded wound (n = 1) and exploration of wound (n = 1).

Figure 1 summarises the TDN grading of surgical morbidity in the 83 patients who experienced at least one surgical complication event, stratified by cranioplasty material. In addition, stratification by timing of highest grading of surgical morbidity is shown: surgical morbidity occurring and resolving within 30 days (Panel A, Figure 1), occurring within 30 days and persisting (Panel B, Figure 1) and occurring after 30 days (Panel C, Figure 1). Finally, TDN grade (independent of time) is provided (Panel D, Figure 1). In summary, titanium plate and mesh morbidity were both more frequent and severe and appear to occur more frequently after 30 days, in comparison to other materials. Hydroxyapatite and acrylic morbidity were less frequent, and for hydroxyapatite, no morbidity persisted beyond 30 days. Hand-moulded PMMA cement was associated with moderate morbidity, whilst PEEK plate was associated with very high morbidity, but this material in particular was used infrequently.

Figure 1. Classification of surgical morbidity. Frequency of patients with each Therapy-Disability-Neurology (TDN) grade, stratified by cranioplasty material which (A) occurred within 30 days and resolved, (B) occurred within 30 days and persisted, (C) occurred after 30 days and (D) for all patients both with and without surgical morbidity events, at any time. The surgical morbidity event that produced the highest grade was ultimately used for grading whee more than one event occurred in a particular patient. Grade 5 events (death) appear in Panel B as they are within 30 days and cannot resolve. The number at the top of each stacked bar states the total number of patients with a surgical morbidity event, within that timeframe. The number in association with each material on the x-axis states the frequency of insertion of that material in total.

Mortality after cranioplasty

Forty-three patients died (15.0%). Three of these deaths (1.0%) were in the peri-operative period (within 30 days) with one being due to hospital-acquired pneumonia, one being due to brain swelling and low GCS necessitating return to theatre and explantation of cranioplasty and a third due to a spontaneous intracerebral haematoma. Craniectomy indications for those that died included; tumour (any pathology) (n = 21), infected bone flap (n = 10), traumatic brain injury (n = 7), cerebral infarct (n = 3), spontaneous intracerebral haemorrhage (n = 1) and primary intracranial infection (n = 1).

Factors associated with primary and secondary outcomes

In multivariable analysis, the risk of all-cause explantation and explantation due to infection was reduced with the use of hydroxyapatite (HR 0.22 [95% CI 0.07–0.71], p = .011, HR 0.22 [95% CI 0.05–0.93], p = .040) or acrylic (HR 0.20 [95% CI 0.06–0.73], p = .015, HR 0.24 [95% CI 0.06–0.97], p = .045), respectively. Risk of explantation due to infection was increased with increasing age at time of cranioplasty (HR 1.04 [95% CI 1.01–1.07], p = .015), cranioplasty at three to six months after craniectomy (HR 6.38 [95% CI 1.35–30.19], p = .020) and for right-sided hemicraniectomy (HR 4.53 [95% CI 1.13–18.24], p = .033). No factors increased the risk of all-cause explantation in this cohort.

Despite investigating a wide range of potentially relevant operative variables, no other craniectomy factors (including indication for craniectomy, urgency of craniectomy, seniority of surgeon/s duroplasty use or closure technique) or cranioplasty factors (including ACCI, ASA, smoking status, seniority of surgeon/s, antibiotic soaking, CSF diversion or closure technique) were associated with the study outcomes. Despite 98 incidents of surgical morbidity across 83 patients (28.9% of the cohort), binary logistic regression analysis did not identify any associated factors associated with the secondary outcome, surgical morbidity.

Risk of mortality was increased with increasing age at cranioplasty (HR 1.47 [95% CI 1.03–2.11], p = .034), use of titanium mesh (HR 5.36 [95% CI 1.88–15.24], p = .002) and use of a surgical drain (HR 3.37 [95% CI 1.51–7.51], p = .003). No factors decreased the risk of mortality in this cohort.

A summary of the univariable and multivariable analyses are provided in the supplementary data file. Kaplan–Meier curves of all-cause explantation free survival, explantation due to infection free survival and overall survival for the entire cohort are demonstrated in Figure 2.

Figure 2. Explantation and mortality. Kaplan–Meier curves demonstrating (A) all-cause explantation free survival, (B) explantation due to infection free survival and (C) overall survival following cranioplasty. Shaded area indicates the 95% CI.

Discussion

This retrospective cohort study evaluated a decade of real-world practice to identify factors associated with morbidity and mortality outcomes following a first alloplastic cranioplasty. All-cause explantation and explantation due to infection were observed in 12.2% and 9.4% of the cohort, respectively. Both hydroxyapatite and acrylic cranioplasty were protective against these outcomes. However, cranioplasty three to six months after craniectomy was associated with an increased risk of explantation due to infection. Surgical morbidity (of any nature and severity) was high, being reported in 28.9% of the cohort, yet no factors were associated with this outcome. Overall mortality was also high (15.0%) and was associated with increasing age at cranioplasty, use of titanium mesh, and use of surgical drains, although this likely reflects the included case mix.

All-cause explantation and explantation due to infection

Explantation of a cranioplasty (whether secondary to cranioplasty infection or not) requires further operations. Understanding associated factors is warranted to reduce the clinical burden of this event for both the patient and healthcare system. Choice of alloplastic cranioplasty material, with respect to surgical morbidity outcomes, has been a central theme within the cranioplasty literature. Autologous bone is associated with a high rate of complications,² notably infection,⁴ complete bone flap resorption,^10,12 reoperation due to bone flap resorption¹² and is now generally out of favour in high-income countries.¹⁷

Titanium was utilised extensively within this cohort in both plate and mesh form. The use of custom-designed titanium plates has, to a large extent, been superseded by the use of ‘in-lay’ cranioplasty materials within our institution, in the form of thermoplastic polymers and ceramics. This has been driven by increasing reports of the superiority of non-titanium alloplastic materials,^{2,9,11,13,15,18,19} but some studies have also reported infection rates that are comparable.^2,12 However, studies evaluating outcomes after cranioplasty frequently neglect to recognise that titanium in plate form is likely to be associated with different outcomes to when titanium is used in mesh form. This was highlighted in a commentary by Ostrov et al.,²⁰ in response to a recent network meta-analysis evaluating complications of cranioplasty in relation to material.⁹ Henry et al., demonstrated that titanium cranioplasty was associated with the highest rate of wound dehiscence and implant exposure at 10%, while conversely HA was not reported on as no data were available for these events in the literature.⁹ Reasons for this are multi-factorial but are reported to include metal hypersensitivity,²¹ scalp atrophy²² and metal fixation hardware causing tension on (in combination with titanium being an on-lay material) or puncture of the overlying skin.¹⁹ Ostrov et al., have previously demonstrated that titanium in mesh form may be associated with lower infection rates in comparison to PEEK, possibly due to reduced adhesion of bacteria or due to a reduced surface area for adhesion.²³ Therefore, we elected to respect the possible differences between titanium in plate and mesh form, in the same way we, and Henry et al., also respected the difference between custom-designed PMMA acrylic cranioplasty and hand-moulded PMMA cement cranioplasty. In return, we did identify differences between our two titanium cohorts. Titanium mesh was generally used with versatility ‘off the shelf’ to restore cranial integrity, especially when craniectomy indication was for tumour, which was the case for over 50% of the titanium mesh cohort. Nearly one in 5 titanium plates and one in 10 titanium mesh cranioplasties were explanted due to infection.

Multivariable analysis demonstrated benefit with the use of hydroxyapatite and acrylic cranioplasty, against all-cause explantation and explantation due to infection. Hydroxyapatite is formed from the same inorganic constituent of bone matrix and is both biocompatible and osteoconductive.^24,25 After custom-made titanium plate, hydroxyapatite was the most frequently utilised material during the study period, used in nearly a quarter of all cases. Only four explantations (three secondary to infection) were observed. This is consistent with the existing literature which demonstrates hydroxyapatite to be highly protective against infection.^9,15

PMMA in a prefabricated, custom-designed form (pPMMA) is commonly referred to as acrylic, but PMMA is also available for use as a hand-mouldable cement (hPMMA). This material offers a customised ‘in-lay’ solution when an ‘off the shelf’ cranioplasty is required, potentially providing an alternative to titanium mesh. Unfortunately, outcomes following PMMA cranioplasty are also rarely presented separately for pPMMA and hPMMA, and this precludes meaningful analysis of these two very different forms of the same chemical material. Henry et al., demonstrated that hand-moulded PMMA in particular was associated with high infection rates and revisional surgery rates, whilst pPMMA outcomes were generally favourable.⁹ Despite moderate use of this material in our study (10%), we only report two explanations, with one being due to infection; less than that observed with titanium mesh.

PEEK has been reported to be favourable with respect to low revisional surgery rates in comparison to autologous, titanium, and both prefabricated and hand-moulded PMMA, and comparable revisional surgery rates to hydroxyapatite.⁹ However, insufficient use within our study period (5%) has hindered analysis. A particular benefit of this material is its strength, and so it may be an ideal option in those patients where HA is deemed to be contraindicated.

A first report of the prospectively maintained United Kingdom and Ireland Cranial Reconstruction Registry (UK CRR) has recently been published,²⁶ which demonstrated that titanium is used extensively in UK practice (64%), followed by autologous (14%), but acrylic and hydroxyapatite are used infrequently (5% each). This is in contrast to our current practice. This initial report describes 30-day outcomes only, and additionally, without consensus outcome definitions, a direct comparison to this dataset is difficult.

An increased risk of explantation due to infection was observed when cranioplasty was performed between three and six months following craniectomy in our cohort. The UK CRR showed that 85% of cranioplasty insertions were performed ‘late’ (>90 days from craniectomy) in the UK, and suggest that a greater delay in time to cranioplasty insertion was associated with increased rate of readmission, although this was not significant. However, Bjornson et al., found no significant difference in the rate of infection or hydrocephalus between patients receiving an ‘early’ vs ‘late’ cranioplasty.²⁷ The importance of cranioplasty timing does remain an important unanswered research question not just for infection but for CSF-related complications and neurological outcome.²⁸

An increased risk of explantation due to infection was also observed when right-sided hemicraniectomy was performed; nearly one in five cases. For comparison, explantation due to infection for left-sided hemicraniectomies was reported once in every eight cases. However, over the study period, right-sided craniectomy was performed slightly more frequently than left-sided craniectomy, possibly due to the reluctance to sometimes intervene surgically for left-sided severe pathology in the form of TBI and cerebral infarct, due to an expectation of worse outcome. In addition, hemicraniectomy is of course a more invasive approach in comparison to all others described in this study (with the exception of bifrontal craniectomy but this was performed less than 10% of the time). Both of these observations may account for the increased risk of explantation due to infection that was observed.

Surgical morbidity

Surgical morbidity (including explantation of cranioplasty) was high with nearly a third of patients experiencing at least one event. The reasons for this are two-fold. First, our patient cohort is highly heterogeneous, and includes patients who have experienced acute vascular events and traumatic brain injuries, along with patients with intracranial tumours and this may likely influence the diversity and frequency of surgical morbidity events that are reported. Second, cranioplasty itself is of course an operation that is historically associated with high levels of surgical morbidity with estimates ranging from 10 to 43%, and this cohort is therefore no exception.^2,9,19

Unfortunately, multivariable analysis did not identify any significant factors associated with the secondary outcome of surgical morbidity, despite an extensive list of factors being evaluated. Given the spectrum and severity of surgical morbidity that may occur, it is not too surprising that individual factors were not identified. This is in contrast to our findings that cranioplasty material and time to cranioplasty were associated with explantation due to infection; as this secondary outcome was very specific.

In order to appreciate the severity of surgical morbidity in this cohort, we elected to grade each patient using the Therapy-Disability-Neurology (TDN) system.¹⁶ This novel grading system has been used successfully to classify adverse events after surgery in patients with diffuse lower-grade glioma, resulting in more adverse events of higher order being attributed when compared to the Landriel–Ibanez classification system.²⁹ In addition, we stratified by time in order separate those events occurring within 30 days of surgery, from those that persist beyond 30 days, or occur late, given the fact that surgical morbidity associated with cranioplasty may not manifest until months, or even years after the peri-operative period has passed. In doing this, it appears that both titanium plate and titanium mesh morbidity events were more frequent, more severe, but also appeared later (after 30 days), when compared to other materials. This may suggest that titanium is associated with surgical morbidity that manifests late, which is not observed to the same extent with ‘in-lay’ materials such as hydroxyapatite and acrylic. Whilst this may reflect a longer implant duration for titanium in this cohort (given our historical preference for this material), and hence ‘time for morbidity event to occur’, this finding is consistent with the current literature.⁹ For this reason, we feel that an ‘in-lay’ alloplastic material is preferable to a custom-designed titanium plate in most circumstances.

Twenty-four patients had 26 post-cranioplasty CSF complications, on a continuum of severity ranging from CSF leaks that could be managed easily with the use of sutures, to those necessitating surgical intervention for CSF diversion. Understanding predictors of hydrocephalus is challenging because both the craniectomy indication (and in particular TBI and SAH),³⁰ and restoration of cranial integrity itself,³¹ can affect CSF dynamics. This is an important research question, as described in the consensus statement from the international consensus meeting on post-traumatic cranioplasty,²⁸ as predicting those patients that may require CSF diversion following cranioplasty could prevent unexpected and severe adverse events.

The consequences of post-cranioplasty extra-axial haematoma range from minimal to profoundly important. Small extra-axial haematomas observed on routine postoperative imaging, or imaging performed for minimal neurological impairment can most likely be managed conservatively. However, for larger extra-axial haematomas, or those associated with substantial neurological impairment, evacuation is likely to be necessary, as was the case for 8 of the 14 patients in this cohort. Cranioplasty explantation may or may not be necessary as judged by the operating surgeon, considering the degree of preoperative neurological impairment and intra-operative brain swelling observed. Ultimately, the wound and cranioplasty (if not explanted) are potentially subject to an increased risk of infection after reoperation for extra-axial haematoma evacuation.³⁰ It is difficult to identify predictors of post-cranioplasty extra-axial haematoma formation based on the incidence observed within this cohort and is likely multi-factorial. A previous study reported male sex, African American ethnicity and hypertension as predictive factors, but this study did not consider cranioplasty material in its analysis. There have however, been reports of HA in particular, being associated with extra-axial haematoma formation.¹¹ We observed two cases where return to theatre was necessary for management of haematoma. In one case, the cranioplasty was replaced after washout, but in the second case, the cranioplasty was explanted due to the need to evacuate a subdural haematoma. A second commonly discussed disadvantage of HA relates to risk of fracture, reported to occur in 2.1% of cases.¹⁵ We did not observe this event in our cohort, possibly due to our comparatively smaller experience with this material. However, given our awareness of the possible increased risk of fracture, we have been cautious to not use this material where there may exist an increased risk of head injury; such as in the context of epilepsy.

There are likely to be other important factors not reported in this study that positively affect clinical outcome but are difficult to quantify. For instance, individual surgeon experience, choice of cranioplasty material relative to the specifics of an individual clinical scenario, and dual senior surgeon representation (for instance, from plastic and aesthetic surgery or maxillofacial surgery) to incorporate technical expertise across subspecialties, especially for more complicated cranial reconstruction cases. Given that dual senior surgeon representation was almost exclusively confined to the HA cohort within this study, it is possible that this reflects a more complex group in our practice, and that reduced incidence of explantation in this group is clinically relevant.

Mortality

In this study, peri-operative mortality was low, with only three patients dying within 30 days of cranioplasty (1%). However, overall mortality at time of follow-up was high (15.0%). The study cohort is heterogeneous with respect to indication for craniectomy but does represent a decade of real-world practice. Nearly half of the patients that died had craniectomy for tumour, with the remainder largely consisting of patients who had craniectomy for an infected bone flap or TBI. It is plausible that craniectomy indication accounts for the mortality observed within this cohort, but the exact cause of death for these patients is not known. Our analysis demonstrated that overall mortality was associated with increasing age at cranioplasty, use of titanium mesh, and use of surgical drains. Three out of every five titanium mesh cranioplasties were inserted when craniectomy indication was tumour, which accounted for half of all the tumour cases. It is therefore unsurprising that the use of titanium mesh was associated with overall mortality in this study. Similarly, we do not conclude that the use of drains is causal of mortality, moreover, that drain use was prevalent in clinical scenarios where mortality was more likely, due to clinical condition and extent of surgery; for instance, in half the cases where a drain was used, indication for craniectomy was TBI and hemicraniectomy was subsequently performed.

Study limitations and future directions

Given the retrospective nature of this study, a number of patients had missing data points. We did however, exclude patients with missing data which would have precluded analysis of the primary and secondary outcomes, as well as patients who had cranioplasty created from more than one material, or a material used less than five times. The number of patients receiving each type of cranioplasty material was relatively balanced across the cohort (with the exception of hand-moulded PMMA cement and PEEK) and reflects a period of evolving practice over the last decade, with a preference for ‘in-lay’ alloplastic materials in more recent years.

Considering that multivariable analysis did not identify factors associated with surgical morbidity, it could be argued that a larger cohort (whether single- or multi-centre) could allow for the analysis of additional, specific morbidity outcomes, where the event of interest may be less common; for instance, need for CSF diversion. In addition, if timing of such surgical morbidity could be accurately extracted, cox proportional hazards model could be used, rather than binary logistic regression analysis.

This study does not consider patient satisfaction, but cosmetic intolerance was a significant issue for six patients in this cohort, and for one this necessitated revisional surgery. However, cosmetic complaints are likely to be underreported. These issues are poorly defined within the literature, especially with respect to cranioplasty material. To this end, we are currently conducting a cross-sectional cohort study investigating health-related quality of life after cranioplasty, which includes satisfaction with cosmetic outcome. We believe that this in an important factor to also consider when selecting a reconstruction material.

Conclusions

Cranioplasty continues to be an operation associated with high levels of surgical morbidity, which extends beyond the most reported outcome of explantation secondary to infection. Consensus does not exist regarding the best material to use, but this should be a case-specific decision. We report that both hydroxyapatite and acrylic are protective against all-cause and infection-associated explantation, and that cranioplasty between three and six months after craniectomy may be associated with an increased risk of infection-associated explantation. Prospective, comparative studies which incorporate health-related quality of life assessments are necessary.

Acknowledgements

Portions of this work were presented in poster form at the Society of British Neurological Surgeons Spring Meeting, Manchester, March 2019, and orally at the online congress of the European Association of Neurological Surgeons, October 2021.

Disclosure statement

CPM is a clinical research fellow at the University of Liverpool, funded by a grant from The Brain Tumour Charity. There are no known conflicts of interest to declare from any author.

Additional information

Funding

The authors reported there is no funding associated with the work featured in this article.