Carbon fibre reinforced PEEK versus traditional metallic implants for orthopaedic trauma surgery: A systematic review

Abstract

Introduction

There is no literature review comparing outcomes of fixation using carbon-fibre-reinforced polyetheretherketone (CFR PEEK) compared to metal implants used in orthopaedic extremity trauma surgery. A systematic review was performed to compare CFR PEEK to metal implants for clinically-important fracture outcomes.

Methods

A search of the online databases of PubMed/Medline, EMBASE and Cochrane Database was conducted. A systematic review was performed following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. A meta-analyses was performed for functional outcomes in proximal humerus fractures converting the score differences to standard mean difference units. GRADE approach was used to determine the level of certainty of the estimates.

Results

Two prospective randomised controlled trials and seven comparative observational studies with a total of 431 patients were included. Of the nine studies included, four compared the use of CFR PEEK against metal plates in proximal humerus fractures. Aggregated functional scores across the proximal humerus studies, there was a small signal of better improvement with CFR PEEK (SMD 0.22, 95% CI -0.03 to 0.47, p = 0.08, low certainty). Greater odds of adverse events occurred in the metal group (OR 2.34, 95% CI 0.73 to 7.55, p = 0.15, low certainty).

Conclusions

Low to very low certainty evidence suggests a small improvement in functional recovery with CFR PEEK in proximal humerus fractures. This may be mediated through a small reduction in major adverse events related to fracture healing and stability. There is currently insufficient evidence to support the widespread use of CFR PEEK implants in fracture fixation.

Level of evidence

Level IV.

1. Introduction

The standard of care for fracture fixation continues to be typically implants composed of metal alloys. Recently, there has been an emergence of carbon-fibre reinforced polyetheretherketone (CFR PEEK) implants with the rationale that they reduce adverse events and may therefore improve functional outcomes in trauma care.¹ CFR PEEK is a thermoplastic reinforced composite construct produced by adding variable lengths of carbon fibre to improve its mechanical properties.² Predictable and alterable biomechanical properties can be achieved by varying the percentage mix of carbon-fibre to the base material as well as modifying the length and orientation of the carbon fibre.² The composition of CFR PEEK implants can be modified to offer strength and stiffness similar to cancellous or cortical bone,³ and even common metallic implants such as steel and titanium. The human cortical bone typically has a tensile strength of 104–121 MPa and a Young's Modulus of 14 GPa,⁴ CFR PEEK has been shown to match these values with a tensile strength and Young's Modulus of 120MPa and 18 GPa respectively.⁵ Some have argued that with the material's mechanical properties being closer to that of human bone, adverse effects on bone such as stress shielding, bone resorption and non union, which are common problems associated with metallic implants, may be avoided.⁶

CFR PEEK implants have the advantage of being completely radiolucent which allows for better radiological assessment of fracture site and healing.⁷ Due to their non-magnetic properties, CFR PEEK implants have the added benefit of being compatible with Magnetic Resonance Imaging.² Over the past ten years, various studies have investigated the biomechanical properties of CFR PEEK and its use in dental implants,⁸ spinal cages⁹ and for bearing surfaces in hip and knee replacement surgery.¹ A previous systematic review by Li and Bhandari et al., in 2015¹ covers other indications for CFR PEEK implants. It also provides a thorough and comprehensive review of the biomechanical and laboratory research that is not the focus of this review. Our focus is on comparative research of CFR PEEK versus traditional metal implants for clinically-important outcomes of extremity fracture fixation. We believe due to the recent emergence of many fracture studies reporting on the comparative efficacy of CFR PEEK, an updated systematic review, and plausible a meta-analysis was warranted.

This systematic review aims to compare the clinical outcomes of CFR PEEK implants, including plates and intramedullary nails, to conventional metallic implants for fracture fixation in extremity orthopaedic trauma surgery.

2. Methods

A systematic review was completed in February 2021. A meta-analyses were performed where possible. These meta-analytic approaches were discussed between group members beforehand to ensure some apriori design and consensus on the approaches. The project was registered on the Open Science Framework DOI 10.17605/OSF.IO/75DMC.

2.1. Search criteria and strategy

A systematic search of PubMed/Medline, EMBASE and Cochrane Database was conducted to find all English language studies which mentioned the use of CFR PEEK in their title or abstract on February 7, 2021. The Boolean search strategy used for PubMed search is available below. The inclusion and exclusion criteria are displayed in Table 1. The search was conducted by two independent reviewers (F.A & U.H.) who identified the studies for inclusion. Only studies which directly compared the outcomes of metallic and CFR PEEK implants were included.

Table 1.

Study inclusion and exclusion criteria.

Inclusion Criteria	Exclusion Criteria
All comparative studies in English language which compared CFR PEEK to metal implants in treatment of upper and lower limb extremity fractures	•Non-comparative cohort studies
	•Studies using PEEK that is not specifically Carbon Fibre Reinforced (CFR) PEEK
	•Studies where implants were not used in Orthopedic Trauma Surgery (e.g. used in dentistry, neurosurgery and elective orthopaedic surgery) or used in fractures related to malignancies
	•Studies where implants were not used in clinical settings/on live patients (includes studies on cadaveric or animal models and studies in vitro investigating biomechanical properties of CFR PEEK)

Fig. 1 shows a flow diagram of the identification and selection process for the studies included in the review complying with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline.

Fig. 1.

PRISMA (Preferred Reporting Items for Systematic reviews and Meta-analyses) flow diagram of the identification and selection process for the studies included in the review.

2.2. Boolean search strategy

2.2.1. Search string

(cfr OR carbon fiber reinforced OR carbon fiber OR carbon fiber reinforced OR carbon fiber OR carbon) AND (peek OR polyetheretherketone OR polyether-ether-ketone OR poly ether ether ketone OR polyaryletherketone OR polyaryletherketones OR polyaryl-ether-ketone OR poly aryl ether ketone).

2.2.2. Data extraction and analysis

Data from each study were extracted and documented into a pre-defined spreadsheet by two independent readers (F.A & U.H). Any differences in opinions were reconciled by the senior author (K.T.) and with consensus.

2.2.3. Assessment of risk of bias

The quality of the studies was assessed independently by 2 reviewers (F.A & U.H) using the Methodological Index for Non-randomized Studies (MINORS) appraisal tool for observational studies and the Cochrane risk-of-bias tool for randomized controlled trials (Table 4). Disagreements were resolved through consensus or discussion with a third reviewer (K.T.). The MINORS appraisal tool was used to evaluate the quality of observational studies.¹⁰ A score of 0, 1, or 2 was assigned to each of the 12 criteria on the MINORS checklist, resulting in a maximum score of 24 for comparative studies. The MINORS scores for comparative studies were categorized as follows: 0–6 indicated very low quality of evidence; 7–10, low quality of evidence; 10–16, fair quality of evidence; and greater than 16, good quality of evidence. The internal validity of randomized controlled trials was assessed using the Cochrane risk-of-bias tool.¹¹ A rating of low, high, or unclear risk was given for selection bias, performance bias, detection bias, attrition bias, reporting bias, and other bias based on in-text evidence.

Table 4.

Risk of bias of prospective randomised trials.

	Adequate sequence generation?	Allocation concealment?	Blinding?	Incomplete outcomes data addressed?	Free of selective reporting?	Other bias
Ziegler [10]	Yes	Yes	No	No	No	Yes-Confounding, Attrition, Detection Biases
Perugia [14]	Yes	Yes	No	No	No	Yes-Confounding, Attrition, Detection Biases

2.2.4. Meta-analyses

We were limited in our ability to aggregate the data in a comprehensive meta-analysis due to the heterogenous reports of outcome measures used in the studies. Therefore, we performed a meta-analysis exclusively for functional outcomes in proximal humerus fractures converting the score differences to standard mean difference (SMD) units. Two-sided p-values, odds ratios, and 95% confidence intervals were provided. We used the GRADE^12,13 approach to determine the level of certainty of the estimates. GRADE is a way of evaluating the source and details of available data to help interpret the quality of evidence and confidence in the effect estimate(s). The framework considers the study design, risk of bias, consistency of the magnitude of the effect, considerations of variability in the patients, interventions, or outcomes being studied (i.e. indirectness), and the precision of the effect estimate. Other considerations include publication bias and a dose-response gradient. The Gradepro online software was used to facilitate the application of the GRADE criteria¹⁴

3. Results

3.1. Included studies

The most recent search (February 7, 2021) yielded 384 results. 347 articles were excluded by reading the title and 24 were excluded by reading the abstract. The remaining 13 articles were fully read. Four of these were excluded (two were non-comparative studies and two focused on the treatment of pathological fractures related to malignancies) (Fig. 1). A total of nine articles were included in this qualitative review (Table 2).

Table 2.

Study Characteristics and Demographic data. Order of listing: CFR PEEK sample vs Metal sample.

Lead author	Year & Country	Journal	Sample	Mean Age	Male Sex	Dominant Limb Affected	BMI	Fracture Type
Ziegler10	2019, Germany	Deutsches Arzteblatt International	63 (32 vs 31)	61.8 ± 12.4 vs 60.9 ± 12.4	6/32 vs 7/31	15/32 vs 14/31	26.4 ± 5.7 vs 26.2 ± 4.7	Proximal Humerus Fractures
								Neer 2 part: 6 vs 5
								Neer 3 part: 22 vs 13
								Neer 4 part: 4 vs 13
Padolino11	2018, Italy	Musculoskeletal Surgery	42 (21 vs 21)	57.4 (41–78) vs 55.8 (22–78)	9/21 vs 7/21	8/21 vs 13/21	27.5 (21.5–38.9) vs 26.8 (17.5–38.1)	Proximal Humerus Fractures
								Neer 3 part: 14 vs 17
								Neer 4 part: 7 vs 4
Katthagen12	2017, Germany	Eur J Orthop Surg Traumatol	42 (21 vs 21)	66.8 ± 9.9 vs 67.4 ± 9.7	7/21 vs 7/21	NR	NR	Proximal Humerus Fractures
								Neer 2 part: 2 vs 2
								Neer 3 part: 9 vs 12
								Neer 4 part: 10 vs 7
Schliemann13	2015, Germany	J Shoulder Elbow Surg	58 (29 vs 29)	66.4 (23–87) vs NR	7/29 vs NR	NR	NR	Proximal Humerus Fractures (Neer parts 3 and 4)
Perugia14	2017, Italy	Injury	30 (15 vs 15)	56.8 ± 7.1 vs 52.6 ± 8.4	5/15 vs 4/15	4/15 vs 6/15	NR	Distal radius fractures
Mitchell15	2018, USA	J Orthopaedic Trauma	22 (11 vs 11)	71.7 (51–89) vs 57.3 (27–86) p = 0.03	3/11 vs 3/11	NA	NR	Distal femur fractures
Byun16	2020, USA	Eur J Orthop Surg Traumatol	31 (10 vs 21)	49.8 (23–80) vs 54.9 (18–89)	6/10 vs 10/21	NA	28.8 (19.1–44.1) vs 27.1 (18.9–47.5)	Distal femur fractures
Guzzini17	2017, Italy	Injury	87 (46 vs 41)	56.8 ± 2.34 vs 58.3 ± 3.55	14/46 vs 11/41	NR (but groups reported as homogenous)	NR (but groups reported as homogenous)	Ankle fractures
								AO class A: 4.3% vs 4.9%
								AO class B: 73.9% vs 75.6%
								AO class C: 21.7% vs 19.5%
Ziran18	2020, USA	J Orthopaedic Trauma	56 (26 vs 30)	36 (20–79) vs 39 (15–74)	24/26 vs 18/30	NA	NR	Diaphyseal Tibial fractures
								AO class A: 5 vs 13
								AO class B: 10 vs 9
								AO class C: 11 vs 8

KEY.

NR: not recorded.

NA: not applicable.

3.2. Study characteristics and demographic data

A total of 431 patients with 431 fractures underwent surgical fixation. This included 211 treated with CFR PEEK implants and 220 with metal implants. The pooled mean age was 58.2 years in the CFR PEEK groups and 55.8 years in the metal groups. All nine studies included in this review compared outcomes were small comparative studies involving two cohorts. Study characteristics and demographic data of the studies included in this review are displayed in Table 2.

3.3. Results of assessment of risk of bias

Assessment of randomised trials with the Cochrane Risk of Bias Tool (Table 4),¹¹ observational comparative studies with the MINORS criteria,¹⁰ all indicated moderate to high risk of bias across studies. The average MINORS score across the seven non-randomized comparative studies (Table 5) was 11.7 (range 6–18). Of the observational comparative studies, two studies had very low quality evidence, four had fair quality evidence and one had good quality evidence.

Table 5.

MINORS—risk of bias assessment of nonrandomized studies.

Variable	Padolino11	Katthagen12	Schliemann13	Mitchell15	Byun16	Guzzini17	Ziran18
1.A clearly stated aim	2	2	2	2	2	2	2
2. Inclusion of consecutive patients	0	0	2	0	0	2	2
3. Prospective collection of data	0	2	2	0	0	2	0
4. Endpoints appropriate for aim of study	2	2	2	2	2	2	2
5. Unbiased assessment of the study endpoint	0	0	2	0	0	0	2
6. Follow-up period appropriate for aim of the study	2	2	2	2	2	2	2
7. Loss of follow-up less than 5%	2	0	0	0	0	0	0
8. Prospective calculation of the study size	0	0	0	0	0	0	0
9. An adequate control group	2	2	2	0	0	2	2
10. Contemporary groups	2	0	0	0	0	0	0
11. Baseline equivalence of groups	2	0	2	0	0	0	2
12. Adequate statistical analyses	0	0	2	0	0	0	2
TOTAL (maximum score: 24)	14	10	18	6	6	12	16
Quality of evidence	Fair	Fair	Good	Very Low	Very Low	Fair	Fair

Score: 0 (not adequate), 2 (reported and adequate).

3.4. Material of implant

The materials and manufacturers of the implants used in the included studies are listed in Table 3.

Table 3.

Materials and manufacturers of implants used.

Type of fixation	Study	CFR PEEK	Metal
Proximal Humeral fractures (plate fixation)	Ziegler10	PEEK Power Humeral Fracture Plate; Arthrex, Naples, Florida, USA	Titanium Proximal Humerus Internal Locking Systes (PHILOS) (Depuy Synthes, West Chester, Pennsylvania, USA)
	Katthagen12
	Padolini11	Diphos H, Lima Corporate, San Daniele del Friuli, Italy
	Schliemann13
Distal radius fractures (plate fixation)	Perugia14	CarboFix distal radiusm volar locking plate (CarboFix Orthopedics Ltd., Herzeliya, Israel)	Titanium Acu-Lock Volar Distal Radius Plate (Acumed Ltd., Hillsboro, Oregon, USA)
Distal femoral fractures (plate fixation)	Mitchell15	CarboFix distal femoral plate (CarboFix 35 Orthopaedics Ltd, Herzeliya, Israel)	Stainless steel variable angle distal femur locking condylar plate (VA-LCP 34) (DePuy-Synthes, Paoli, USA)
	Byun16
Tibial shaft fractures (Intramedullary nailing)	Ziran18	Carbofix Tibial Nail (CarboFix, Herzeliya, Israel	Titanium alloy nail (Depuy Synthes, Paoli, USA)
Lateral malleolus ankle fractures (plate fixation)	Guzzini17	‘Radiolucent (CFR PEEK) plate’	‘Stainless steel plate’
		Name of manufacturer and type of plate not stated

Detailed study outcomes of the nine included studies are available in Table 6.

Table 6.

Study outcomes of all included studies.

Study	Design	Surgical use	Sample	Outcomes			CFR PEEK	Metal	P value
Ziegler et al.10	Randomised controlled trial	Proximal humeral fractures (plate fixation)	63 32 CFR PEEK 31 Metal	Complete bony consolidation		12 weeks	32/32	31/31	NR
				Significant loss of reduction			0/32	0/32	NR
				Head shaft angle		Post op	142.46 ± 6.39°	139.97 ± 7.70°	NR
						6 weeks	142.13 ± 6.38°	139.82 ± 7.70°	NR
						12 weeks	142.53 ± 6.45°	138.81 ± 8.21°	NR
				Disabilities of the arm, shoulder and hand (DASH) Score		6 weeks	56.5 ± 19.3	59.8 ± 15.6	NR
						12 weeks	38.4 ± 21.4	37.7 ± 16.2	NR
						6 months	27.5 ± 20.5	28.5 ± 17.9	0.82
				Simple Shoulder Test		6 weeks	30.0 ± 20.8	29.4 ± 18.9	NR
						12 weeks	54.9 ± 24.8	51.5 ± 16.5	NR
						6 months	62.5 ± 22.3	65.0 ± 20.1	NR
				Oxford Shoulder Score		6 weeks	20.3 ± 9.8	20.4 ± 8.5	NR
						12 weeks	33.8 ± 10.0	33.3 ± 6.5	NR
						6 months	37.7 ± 8.8	38.6 ± 6.8	NR
Padolino et al.11	Retrospective observational study	Proximal humeral fractures (plate fixation)	42 21 CFR PEEK 21 Metal	Inadequate fracture reduction			3/21	4/21	>0.05
				Cortical thinning (mm); change from pre-op values in brackets			4.7 (1.3)	3.9 (0.13)	0.002
				Adequate calcar reduction			11/21	13/21	0.548
				>50% tuberosity resorption			3/21	9/21	0.040
				Varus/valgus malalignment			2/21	2/21	>0.05
				Humeral head collapse/necrosis			1/21	1/21	>0.05
				Simple Shoulder Test	Final follow up		7 ± 2.25	7 ± 2.16	>0.05
				Constant Murley Score	Final follow up		66.3 ± 20.5	63.3 ± 19.6	>0.05
				Active anterior elevation	Final follow up		142.8 ± 38°	127.6 ± 33°	0.075
				Active lateral elevation	Final follow up		134.1 ± 40°	113.8 ± 38°	0.099
				External rotation (points)	Final follow up		32.6 ± 12	36.6 ± 26	0.737
				Internal rotation (points)	Final follow up		2.8 ± 2	2.4 ± 2	0.664
Katthagen et al.12	Retrospective observational study	Proximal humeral fractures (plate fixation)	42 21 CFR PEEK 21 metal (matched from literature)	Constant Murley Score	3 months		57.5 ± 16.2	52.4 ± 15.5	0.31
					12 months		73.8 ± 15.4	69.4 ± 18.5	0.43
				One or more screw perforation			NR	5/21	NR
				Revision surgery due to screw perforations			Less	More	0.048
Schliemann et al.13	Prospective observational study	Proximal humeral fractures (plate fixation)	58 29 CFR PEEK 29 Metal	Simple Shoulder Test	24 months		59	48	NR
				Oxford Shoulder Score	24 months		27.4 (8–45)	21.6 (9–43)	0.029
				Constant Murley Score	24 months		71.3 (44–97)	59.2 (28–86)	0.038
Perugia et al.14	Randomised Controlled Trial	Distal radius fractures (plate fixation)	30 15 CFR PEEK 15 Metal	Mean follow-up for outcomes (months)			15.7 (12–19)	16.1 (13–21)
				DASH Score			15.3 (2.5–58.9)	12.2 (10.6–54.8)	>0.05
				Visual Analogue Scale (VAS) pain score			3.6 (0–7)	2.9 (0–6)	NR
				Complete recovery of range of movement			10/15	12/15	NR
				Wrist extension			64° (44°–76°)	61° (42°–75°)	>0.05
				Wrist flexion			78° (59°–80°)	80° (62°–80°)	>0.05
				Radial deviation			18° (7°–20°)	15° (5°–20°)	>0.05
				Ulnar deviation			39° (23°–45°)	42° (26°–45°)	>0.05
				Supination			87° (72°–90°)	88° (70°–90°)	>0.05
				Pronation			80°	77° (70°–80°)	>0.05
				Hand grip strength as percentage of unaffected limb at final follow up			92.30%	94.40%	NR
				Key pinch strength as percentage of unaffected limb at final follow up			90.40%	90.70%	NR
				Radiological restoration of normal radial inclination (21°–25°)			74.5% (15°–27.5°)	73% (14°–29°)	>0.05
				Radiological restoration of normal Radial height (10–13 mm)			66.6% (6.8–17.3 mm)	70% (6.3–18.2 mm)	>0.05
				Radiological restoration of normal volar tilt (7°–15°)			90.2% (3°–187°)	91.3% (5°–185°).	>0.05
				Radiological restoration of normal ulnar variance (0.7–1.5 mm)			86.3% (0.7–4.1 mm)	85.8% (0.5–4.8 mm)	>0.05
Mitchell et al.15	Retrospective observational study	Distal femur fractures (plate fixation)	22 11 CFR PEEK 11 Metal (stainless steel)	Non-union			1/11 (9.1%)	4/11 (36.4%)	0.12
				Hardware failure			0/11 (0%)	2/11 (18.2%)	0.14
				Re-operation			1/11 (9.1%)	4/11 (36.4%)	0.12
				Mean time to Radiographic Union			18.7 weeks	12.4 weeks	0.26
				Mean time to Full Weight-bearing			9.9 weeks	12.4 weeks	0.23
Byun et al.16	Retrospective observational study	Distal femur fractures (plate fixation)	31 10 CFR PEEK 21 Metal (stainless steel)	Modified Radiographic Union Score for Tibia (mRUST) score		3 months	9.0 ± 1.9 (6.3–12.3)	6.9 ± 2.0 (4.3–11.7)	0.015
						6 months	11.4 ± 2.6 (7.7–16.0)	10.5 ± 2.5 (6.0–15.7)	0.374
				Loss of alignment			1/10 (10%)	1/21 (5%)	0.548
				Unplanned surgeries (nonunion repairs)			0/10 (0%)	3/21 (15%)	0.296
Guzzini et al.17	Prospective observational study	Ankle fractures (plate fixation)	87 46 CFR PEEK 41 Metal (stainless steel)	OMAS (Olerud-Molander Ankle Score)		6 months	79.25 ± 10 (69.25–89.25)	78.8 ± 15 (63.8–93.8)	>0.05
						12 months	85.1 ± 5 (80.1–90.1)	84.1 ± 6.7 (77.4–90.8)	>0.05
						24 months	91.1 ± 4.16 (86–95.26)	88.7 ± 4.7 (84–93.4)	>0.05
				American Orthopaedic Foot and Ankle Society (AOFAS) Ankle-Hindfoot Scale		6 months	77.25 ± 9 (68.25–86.25)	75.8 ± 15 (60.8–90.8)	>0.05
						12 months	86.4 ± 7 (79.4–93.4)	83.1 ± 6.7 (76.4–89.8)	>0.05
						24 months	92.1 ± 4.16 (87.94–96.26)	90.1 ± 4.7 (85.4–94.7)	>0.05
				Visual Analogue Scale (VAS) pain score		6 months	2.8 ± 2.1 (0.7–4.9)	3.1 ± 1.5 (1.6–4.6)	>0.05
						12 months	2.4 ± 1.1 (1.3–3.5)	2.1 ± 1.7 (0.4–3.8)	>0.05
						24 months	1.4 ± 1.1 (0.3–2.5)	1.5 ± 0.7 (0.8–2.2)	>0.05
				Talo-crural angle		24 months	9.3° ± 0.9 (8.4°–10.2°)	10.4° ± 0.8 (9.6°–11.2°)	NR
				Restoration of joint line			98% (45/46)	95% (39/41)	NR
				Satisfactory rotation of the peroneal malleolus on its axis			100% (46/46)	100% (41/41)	NR
				Removal of hardware for local infection			6% (3/46)	9% (4/41)	NR
Ziran et al.18	Retrospective observational study	Diaphyseal Tibial fractures (Intramedullary nailing)	56 26 CFR PEEK 30 metal (titanium)	RUST (Radiographic Union Scale in Tibia)		8 weeks	19% (5/26)	0% (0/30)	<0.0001
						12 weeks	69% (18/26)	17% (5/30)	<0.0001
						16 weeks	92% (24/26)	57% (17/30)	<0.0001
						20 weeks	96% (25/26)	87% (26/30)	<0.0001
						24 weeks	96% (25/26)	97% (29/30)	>0.05
				Nonunion			0	0	>0.05
				Infection			0	0	>0.05
				Nonunion/infection			1/26	1/30	>0.05
				Barometric pain			8% (2/26)	23% (7/30	0.065
				Ankle/knee pain			8% (2/26)	20% (6/30)	0.109
				Removal of implant			8% (2/26)	3% (1/30)	0.269
				Associated injuries			35% (9/26)	13% (4/30)	>0.05

KEY.

+/- Standard Deviation.

(range).

NR: not recorded in literature.

3.5. Upper limb trauma

Of the nine studies included in this review, four of them compared the use of CFR PEEK plates against conventional metallic plates in proximal humerus fractures.15, 16, 17, 18 When we aggregated functional scores at the last follow-up across the four studies (Table 6), there was a small signal of better improvement with CFR PEEK plates (SMD 0.22, 95% CI -0.03 to 0.47, p = 0.08, low certainty). Schliemann et al.¹⁸ found that CFR PEEK plates provided substantially functionally better outcomes than titanium plates at 24 months postoperatively. Both the Oxford Shoulder Score (OSS) and Constant-Murley (CM) Score (SMD 0.44, 95% CI 0.02 to 0.79, p = 0.04, low certainty) were superior. There was a greater odds of adverse events in the metal group (OR 2.34, 95% CI 0.73 to 7.55, p = 0.15, low certainty). Although the authors stated that the patients were matched, they did not report sufficient detail to determine if confounding bias was sufficiently controlled for. Therefore, confounding may have driven these outcomes.

Of the four studies above, one is a randomized controlled trial by Ziegler et al.¹⁵ which showed no statistically significant difference between Disabilities of the arm, shoulder and hand (DASH) score, Simple Shoulder Test (SST) and the Oxford Shoulder Score (OSS) at multiple time points (p = 0.81). There was; however, a disproportionate number of 4-part fractures between the groups (Table 2). Padolino et al.¹⁶ showed that CFR patients had roughly a 15° better improvement in both anterior and lateral elevation (combined p = 0.01), but the mean CM score was only a couple of points higher in the CFR PEEK (and not significantly different). Also, Katthagen et al.¹⁷ found a small to medium effect size difference in the CM Scores that favored CFR-PEEK shoulders but the estimate was imprecise due to the small sample size of 42 patients (21 patients in each group)(SMD 0.35, 95% CI -0.26 to 0.96, p = 0.25, very low certainty, underpowered).

Ziegler et al.¹⁵ and Padolino¹⁶ et al. also reported radiological outcomes for the fixation of proximal humeral fractures. Ziegler et al.¹⁵ reported consolidations by 12 weeks, no adverse events, and identical radiographic parameters such as the head-shaft angle between groups. Padolino et al.¹⁶ noted that there was a significantly higher rate of tuberosity resorption in the metal plate group (OR 4.5, 95% CI 1.01 to 20.1, p = 0.04, low certainty) but did not find a significant difference in the other radiological parameters measured.

Schliemann et al.¹⁸ reported a 17.2% reduction in the adverse event rates with CFR PEEK plates concerning screw cutout, loss of reduction, and varus deformity (ARR 17.2%, 95% CI -5.8 to 40.3%, p = 0.15, low certainty). However, the loss to follow-up by 2 years was 20.1% in the CFR group (vs 0% in the metal cohort), which may have introduced substantial attrition bias (possibly favouring the CFR PEEK group).

A small randomized controlled trial by Perugia et al.¹⁹ compared internal fixation of distal radius fractures with CFR PEEK and Titanium volar locking plates reporting a range of clinical and radiological outcomes. They found no difference in the range of motion, grip strength, DASH score, pain score, and a range of different radiological parameters at a mean follow up of a mean of 16 months.

3.6. Lower limb trauma

Three observational studies of the 9 reviewed investigated plating of lower limb fractures. Two studies (Mitchell et al.²⁰ and Byun et al.²¹) enrolled distal femoral fractures and compared CFR PEEK versus standard metal plate fixation. The combined estimate indicated a higher odds of unplanned reoperation with metal implants (OR 4.98, 95% 0.76 to 32.80, p = 0.10, low certainty). However, despite the lower pooled event rate with CFR, Mitchell et al.²⁰ reported that the time to union was 6.4 weeks later than standard metal implants (95% CI -4.38 to 17.18, p = 0.24, very low certainty). This is likely driven by confounding bias. On the contrary, Byun et al.²¹ found that CFR PEEK produced greater callus formation at three months in distal femoral fractures, but not later timepoints.

Guzzini et al.,²² conducted a prospective observational study investigating plate fixation of lateral malleolus ankle fractures. They reported small, clinically insignificant improvements in functional scores and radiographic parameters in CFR patients across all time points (i.e. 6 months to 2 years).

A retrospective observational study by Ziran et al.²³ compared the outcomes of CFR PEEK intramedullary (IM) nails versus titanium alloy IM nails for fixation of tibial shaft fractures. The CRF group healed, in general, dramatically faster up until 24 weeks post-injury. For instance, at 12 weeks 69% versus 17% healed and at 16 weeks 92% versus 57% had united (p < 0.001, Table 6). This translates to have a 9-fold odds increase of being healed by 20 weeks post-injury with CFR implants (95% CI 1.83 to 46.1, p = 0.007, low certainty).

4. Discussion

The objective of this review was to evaluate the clinical outcomes of CFR PEEK material used for orthopaedic extremity trauma internal fixation surgery compared to metal plates. To our knowledge, this is the first systematic review on the subject which compares outcomes of the two materials in this clinical application.

Advances in biomaterials that mimic the properties of bone while achieving stable fixation have immense potential. We found that in both the proximal humerus and tibial shaft there was low certainty evidence of a plausible effect on improved healing rates with CFR PEEK implants. Specifically, in the humerus, there was a small improvement in functional outcomes that marginally surpassed SMD = 0.2, which is widely accepted as a clinically meaningful “small effect size”.²⁴ This may have been driven by a 2-fold reduction in adverse healing events with CFR PEEK proximal humeral plates. In the tibia, there was a 9-fold odds of having bone healing by 20 weeks post-injury.

In one of the trials, Ziegler et al.,¹⁵ found no meaningful differences for plate fixation of proximal humerus fractures in their randomized trial with only 32 versus 31 patients. In contrast, Schliemann et al.¹⁸ in their prospective observational study of 58 patients found that OSS and CM Scores were significantly superior, the CM by 12 points (SMD 0.41 95% CI 0.02 to 0.79, p < 0.001). This is in keeping with a moderate effect on functional improvement which is likely related to the 2-fold increase in healing complications in the metal plate group. Another paper, Padolino et al.,¹⁶ showed a slightly better functional result in the CFR group that is unlikely to be clinically meaningful (SMD 0.15, p > 0.05). However, upon closer inspection of the data, there is some indication of the complexity that may be happening. In keeping with the plausible biological advantage of the CFR PEEK plates (lower young's modulus compared to metal), there was a 5-fold greater odds of tuberosity resorption in the metal plate group (p = 0.04). However, there was a 1.5-fold odds increase in the CFR PEEK group of failing to attain an anatomic calcar reduction (p > 0.05). It would appear that the biological advantage of CFR PEEK implants may come at the cost of lack of versatility (inability of the plate to be bent/moulded at the time of surgery due to the lack of its malleability) in some contexts, which may be a major factor why the results are heterogeneous throughout this review. For this reason, we implore future studies to report issues around reduction and fixation quality and detailed reporting on adverse events related to these contributing factors. Specifically, if the adverse events occurred in those with suboptimal fixation/reduction. This will aid in us differentiating between the biological effect of CFR PEEK and the offsetting versatility effect.

Perugia et al.,¹⁹ in their randomized controlled trial on the optimal fixation of distal radius fractures, the reporting suggests no significant difference but the weakness of their approach required us to re-analyze their work. First, the study is small with only 15 patients per arm; however, there was a 2-fold odds in the CFR PEEK group of having a DASH score above 30 and above 50 (p > 0.05, underpowered). The mean difference in the DASH scores was 3.1 patients favoring CFR PEEK, which is a difference of 20.0%. The minimally clinical important difference (MCID) that is reported ranges from 10.8 to 15.0%, and therefore, the prior difference may reach the threshold in this context. In addition, when this study is aggregated with the four proximal humerus fixation studies, it increases the small effect on functional improvement to reach statistical significance (SMD 0.23, 95% CI 0.0 to 0.47, 5 studies, p = 0.05, very low certainty). This would equate to a mean change in the DASH score of 5.0 points which is in a gray area for MCID between those that report no change (-3.0 to 3.0 point change) and those with some better (7–14 point change).²⁵ It is certainly lower than Slobogean's MCID threshold for proximal humerus fractures of 15.0 points.²⁶ In all likelihood, after close review of the data, it suggests that some patients are likely benefiting from CFR PEEK's biological properties while others may be adversely affected due to technical issues such as the inability to bend the CFR PEEK plate to adequately fit the anatomy of the bone in complex fracture cases.

For lower limb fracture fixation, there is a paucity of both comparative and non-comparative studies in the literature. Mitchell et al.²⁰ and Byun et al.²¹ who have investigated the treatment of distal femoral fractures both showed an aggregate 5-fold odds reduction in adverse healing events (p = 0.10, low certainty). However, they report discrepant results on time to union. Byun et al. found a large improvement in the RUST score at 3 months (SMD 1.07, 95% CI 0.47 to 1.66, p < 0.001, low certainty) that was mitigated by one year. However, Mitchell et al.,²⁰ showed a faster time to union in the CFR PEEK plating group. Mitchell's study had confounding biases that were likely enormously influential, with the CFR PEEK group being a mean 15 years older (mean age 71.7 years old) and also had a higher proportion of fractures with intraarticular extension. Our recommendation, in addition to conducting larger studies, is for studies to perform adjusted analyses for confounding factors (e.g. linear regression for functional outcomes and logistic regression for adverse events). Ideally, one should choose the confounders for adjustment apriori, which should be done; however, in contexts where the sample size is limited, one may consider adding to the model factors that are 1) clinically likely to impact outcomes and 2) disproportionate between the study arms. This can be done as a sensitivity analysis to maximize power for the main outcome while correcting for confounding bias.

It is important to recognise that the mechanical properties of CFR PEEK implants are dependent on the composition of the percentage of CFR and PEEK, carbon fibre lengths, and orientations within the material. This will lead to different mechanical properties depending on implant design and manufacturer. In addition, the process of creating the CFR PEEK such as compression, injection or composite flow moulding and annealing can all affect the material's mechanical and wear properties and characteristics.²⁷ For example, studies containing the Arthrex (Arthrex, Naples, FL, USA) CFR PEEK plates are made by composite flow molding as compared to the Carbofix CFR PEEK (CarboFix Orthopedics Ltd., Herzeliya, Israel) which are made by compression molding of layers of continuous Carbon Fibres in a matrix of PEEK. How the various compositions of CFR implants impact their performance concerning healing and also durability is unknown. This diversity within the CFR PEEK cohorts could have certainly imposed heterogeneity within this review. The metal implants are likely to have less diversity in the mechanical properties in comparison. However, we have insufficient data available to properly evaluate these differences. As more data becomes available, a follow-up to this report may be of value to analyze the impact of CFR composition with advanced meta-analytic techniques.

None of the reviewed clinical studies have reported on the biological response to wear particles created at the screw-plate interface. Previous tribological studies looking at wear from CFR PEEK materials have shown not to be worse than other metals regularly used in medical fixation implants.28, 29, 30 Further studies are required to study the effects of CFR PEEK material in the clinical context of fracture fixation that assesses wear, tissue-based reactions, debris production and any plausible adverse reaction with bone.

Many limitations in this review are related to the quality of available literature. To address this we recommend, 1) more complete data collection to include demographic, injury details, classification, specific boney location, other injury information, other important prognostic data (e.g. Weight bearing status), indication for the implant used (if not randomized), outcome data at important time points (e.g. bony union at 6, 12, 16, 20 and 24 weeks) 2) attempt to include all patients and not exclude patients based on follow-up. In other words, use all the data until patients drop out and then provide the outcomes of the limited dataset for later time points. 3) Identification of confounding bias by assessing the differences between groups by intuitive comparison of percentages and means and avoid relying on p-values to flag confounding 4) Reporting using the commonly used functional outcomes. Investigators should consider published MCIDs in their interpretation of results. 5) Clear definitions for binary outcomes such as defining nonunion of the tibia diaphysis as failure to progress to healing by six months post-injury. 6) Improvement in blinding practices between implants that are easily discernible on unmodified plain films. 7) Attempt to increase study sizes by increasing the recruitment window and by collaboration.

5. Conclusion

Low to very low certainty evidence suggests a small plausible improvement in functional recovery with CFR PEEK implants. This may be mediated through a small reduction in major adverse events related to fracture healing and stability. There are a number uncorrected confounding bias, unblinded - likely discrepant - adjudication of outcomes (especially assessment of time to union), and modifications to study protocols that were employed - without explanation - that likely favoured CFR PEEK implant-outcomes, by reducing the risk of implant failure. Patterns in the data suggest a plausible improvement in expediency to union with CFR PEEK constructs, which may be offset by inferior implant versatility. Very low certainty evidence suggests that the latter issue may have resulted in a small increase in adverse events related to suboptimal reduction and or fixation. Future studies must improve methodological quality and reporting standards as per our recommendations in this report. There is insufficient evidence to support the widespread use of CFR PEEK implants in fracture fixation at this time. Given the paucity of high quality clinical data on CFR PEEK for trauma implants, high powered well designed randomised clinical trials are needed before clinically relevant recommendations regarding the use of CFR PEEK can be made.

Contributions from authors

K Theivendran: Conceiving the study, formulating the review question, study selection, data analysis & writing the manuscript, F Arshad: Study selection, data analysis, writing the manuscript, UK Hanif: Study selection, data analysis, writing the manuscript, A Reito: Manuscript review, X Griffin: Manuscript review, CJ Foote: data analysis, writing the manuscript.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declaration of competing interest

All authors have no conflicts of interests.