2. MACQUARIE NEUROSURGERY
JOURNAL CLUB
Stephen Honeybul, David Anthony Morrison, Kwok M. Ho, Christopher R. P. Lind, and
Elizabeth Geelhoed
A randomised controlled trial comparing autologous
cranioplasty with custom-made titanium cranioplasty
Journal of Neurosurgery, January 2017, volume 126, pp81-90.
3. Authors
• Senior author: Stephen Honeybul, neurosurgeon Sir Charles Gairdner
Hospital
• David Morrison: biomedical engineer Royal Perth Hospital
• Kwok Ho: intensive care physician Royal Perth Hospital
• Christopher Lind: neurosurgeon Sir Charles Gairdner Hospital and RPH
• Elizabeth Geelhoed, professor of population health, UWA
5. Study Relevance
• Cranioplasty after decompressive craniectomy is performed to restore cerebral
protection; for cosmesis; and has been reported to improve neurologic function
• Benefits of patient’s own bone: cheap; biocompatible; radiolucent; ideal
contour.
• Disadvantages: high failure rate due to infection or bony resorption
• Titanium is an alloplastic material that has been previously used due to
strength and biocompatibility
• Computer-assisted design and manufacturing has made production of custom-
made plates viable
7. Aim
• To compare the long term aesthetic and functional outcomes, and the cost-effectiveness, of
primary autologous bone versus primary prefabricated titanium cranioplasty.
• Hypothesis: primary titanium cranioplasty improves cerebral protection by avoiding the frequent
problem of bone resorption requiring reoperation, and this will offset the higher initial outlay cost of
titanium plates
8. Trial Design
• Prospectively registered with ANZCTR (ACTRN12612000353897)
• Interventional randomised controlled trial
• Single centre, single surgeon
• Two arm, parallel group study
• Single Blinded (to patient and relatives)
9. Trial Design
Eligibility criteria:
• age >18
• decompressive craniectomy + own bone available
Exclusions:
• bone flap damaged or comminuted
• intracranial infection following craniectomy
• penetrating brain injury
10. Trial Design
Manufacture:
• based on high resolution CT
• rapid protocol 3D model of skull created
• plates manufactured from medical grade titanium, (thickness 0.6-1.0mm)
• patient identification and orientation laser marked on plate
• plate try-fitted on model to ensure smooth contour and unambiguous orientation
• standard steam sterilisation
12. Trial Design
Primary outcome measure:
• implant failure requiring reoperation
• either infected cranioplasty requiring removal of implant or bone flap resorption
Secondary outcome measures:
• adverse events (any time in the 12 months)
• cosmetic and functional outcome (assessed by S.H. and patient/primary carer)
• total hospital costs over 1 year period
15. Trial Design
Statistical Analysis Plan
• sample size calculation: taking p<0.05 as significant, and a power of 80% for the
study, 32 patients in each group to demonstrate a 25% difference
• this was based on S.H. experience: 32% complication rate with autologous
cranioplasty (due to infection or bone resorption) v. <5% complication rate with
titatium cranioplasty
• intention to treat analysis
23. Internal Validity
Statistics:
• power calculation was appropriate (80% with p=0.05)
• 25% change in outcome is comparable to the literature of the senior surgeon
• achieved their recruitment target
• sample size calculation seems dubious
Protocol prospectively registered with ANZCTR
• note the change from double to single blinded RCT
• ?Experimenter bias
24. Internal Validity
Randomisation:
• 6 weeks prior to cranioplasty
• Random number sequence generated by software program
• Randomisation numbers concealed in sequentially numbered sealed envelopes
• Baseline characteristics were similar
Blinding:
• The surgeon was not blinded…he was also the clinical assessor
Subjective outcomes, though no widely used method of measuring cranioplasty success
25. External Validity
• Representative population
• Relevant inclusion/exclusion criteria
• The trial design is replicable
• Surgical technique is detailed, documented and replicable
• Possible that 12 months is not long enough follow up
• All surgeries performed by a senior neurosurgeon, not usually the case in public hospitals
• Titanium plates not used in MUH
26. Conclusion
• More revisions in the autologous bone group, but not statistically significant (p=0.053)
• Titanium cranioplasty was associated with better functional and cosmetic outcomes
than autologous bone
• There was no significant difference in health care cost
• No primary infections
• Risk of bias from unblinding of surgeon/assessor
• Overall, thoughtful and well conducted trial, though prone to bias and some limitations
on external validity, so now independent verifiability important
Editor's Notes
Transparent title
Doesn’t mention company
asceptic necrosis causes bone flap resorption (28 of 92 patients in one series had major resorption, range from 4 to 50% in literature)
rates of infection 1-2% in W.A. 20% in some series.
paper in 2013 reporting on 127 titanium cranioplasties in RPH and CGH (Neurosurgery. 2013 Feb;72(2)248-56;256. Cranioplasty with custom-made titanium plates--14 years experience. Wiggins A, Austerberry R, Morrison D, Ho KM, Honeybul S)
8 references by S.H.
52 patients, 6 months follow up. mixed results, 2 HA infections, 5 titanium; reoperation in 7 patients in each group
hypothesis stated in discussion
Initially double blinded, but changed to single blinded after failure to get funding for a research nurse. So assessment of outcome was performed by surgeon.
company?
deals with hospital?
one surgeon limit confounders from different techniques and expertise.
introduces some bias to be discussed later
assessment of bone flap resorption performed by S.H. based on post-op CT and the 12 month CT (criteria shown later)
AEs: neurologic deterioration; post-op collections; increased seizure rate
functional assessment involved clinical palpation to assess how effectively the contour of the skull was restored
assessment performed by S.H.
and infection was defined as infected cranioplasty requiring removal and systemic antibiotics
overall outcome was graded based on the lowest score achieved, i.e. to be a complete success, it needs ‘complete success’ in all categories
quotes 2 of his own articles
ITT ‘once randomised, always analysed’
suspicious power calculation, based on such a large suspected difference in the groups
Recruitment from 2012 to 2015
Both deaths were due to withdrawal of treatment in patients with severe TBI, unrelated to cranioplasty. (4 and 6 months)
similar with respect to baseline characteristics
mostly male
majority TBI and majority unilateral craniectomy
roughly 90 days
though infection of a redo done for bone resorption
5 of these underwent titanium cranioplasty, 2 declined (because of thick hair and not much of a cosmetic effect)
No significant differences in complications nor in cumulative hospital stay
Statistically significant increased success rates in all 4 categories
Mann-Whitney test or chi-square test (so not normal distribution data)
could feel dura in complete failure
put cost of abs in autologous group, although it was a titanium plate that got infected
note that the mean difference per patient was not significantly different
S.H. seems to have a lot of time and effort invested in titanium plates, as this is a product is leaves me slightly suspicious. Esp. as he was so involved in this trial.
So rigorous in this regard
introduces observational bias and interviewer bias (as it is not clear how they obtained data from patients
Revision in particular, and infection, is more clinically important
No infections may be due to senior surgeon, taking more care because it’s a trial, expert at technique…not reflective of clinical practice.
Follows CONSORT checklist reasonably well