Timing for cranioplasty to improve neurological outcome: A systematic review

Abstract Introduction Cranioplasty is a surgical technique applied for the reconstruction of the skullcap removed during decompressive craniectomy (DC). Cranioplasty improves rehabilitation from a motor and cognitive perspective. However, it may increase the possibility of postoperative complications, such as seizures and infections. Timing of cranioplasty is therefore crucial even though literature is controversial. In this study, we compared motor and cognitive effects of early cranioplasty after DC and assess the optimal timing to perform it. Methods A literature research was conducted in PubMed, Web of Science, and Cochrane Library databases. We selected studies including at least one of the following test: Mini‐Mental State Examination, Rey Auditory Verbal Learning Test immediate and 30‐min delayed recall, Digit Span Test, Glasgow Coma Scale, Glasgow Outcome Scale, Coma Recovery Scale‐Revised, Level of Cognitive Functioning Scale, Functional Independence Measure, and Barthel Index. Results Six articles and two systematic reviews were included in the present study. Analysis of changes in pre‐ and postcranioplasty scores showed that an early procedure (within 90 days from decompressive craniectomy) is more effective in improving motor functions (standardized mean difference [SMD] = 0.51 [0.05; 0.97], p‐value = 0.03), whereas an early procedure did not significantly improve neither MMSE score (SMD = 0.06 [−0.49; 0.61], p‐value = 0.83) nor memory functions (SMD = −0.63 [−0.97; −0.28], p‐value < 0.001). No statistical significance emerged when we compared studies according to the timing from DC. Conclusions It is believed that cranioplasty performed from 3 to 6 months after DC may significantly improve both motor and cognitive recovery.


| INTRODUC TI ON
Decompressive craniectomy (DC), consisting in the partial removal of the skullcap, is widely used in the management of neurological emergencies as it allows a decrease in brain swelling and intractable intracranial hypertension (Hofmeijer et al., 2009). DC is performed for a variety of reasons, but the most common are tumor removal and the reduction in increased intracranial pressure due to malignant ischemic or hemorrhagic stroke (Hofmeijer et al., 2009;Vahedi et al., 2007). Cranioplasty (CP) is a neurosurgical procedure aimed to repair the skull defect following craniectomy.
The search for materials and strategies to provide more comfortable and reliable surgical procedures is a challenging topic, both in clinical and in economical terms. However, none of the currently available materials meets the criteria required for an ideal implant (Zanotti et al., 2016).
Recently, promising results following this procedure in both motor and cognitive outcomes have been reported. Thus, this link between the repair of the cranial defect and the changes in cerebrovascular and cerebrospinal fluid hydrodynamics seems to have positive effects on neurological functions (Bijlenga, Zumofen, Yilmaz, & Creisson, 2007).
If on one hand CP may lead to notable improvements (Sancisi et al., 2009;, on the other hand it may increase the possibility of infections, the risk of hydrocephalus (especially when performed later), and the possibility of developing the "trephined" syndrome , especially when operation time exceeding 90 min (Cho & Kang, 2017). Indeed, although the mortality rate after cranioplasty is rather low, research suggests that 1 out of 3 people has overall complications (Zanaty et al., 2015), especially seizures and infection . Timing of cranioplasty is therefore crucial even though the literature is divided. According to several studies, it should be performed from 3 to 12 months following DC, based on the presence of infections or postoperative complications. Indeed, in order to prevent the development of devitalized autograft or allograft infections it is recommended to wait from 3 to 6 months before reconstructive surgery, even one year if there is an infected area (Aydin, Kucukyuruk, Abuzayed, Aydin, & Sanus, 2011). On the contrary, an early intervention (i.e., within 3 months) seems to reduce neurological complications, especially in patients with severe acquired brain injury, since a lesion in the postacute period might be negative for motor and cognitive recovery (Huang, Lee, Yang, & Liao, 2013). Although the timing to perform cranioplasty largely depends on personal clinical experience rather than evidence-based data, it could be useful to estimate a suitable threshold to perform cranioplasty.
In this article, we want to review current literature on motor and cognitive effects of an early cranioplasty after decompressive craniectomy, also focusing on the optimal timing to perform it.

| Data sources and keywords
A systematic review and meta-analysis in line with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were performed.
Articles published up to July 2017 were searched on the PubMed, Web of Science, and Cochrane Library databases, without language restrictions. A follow-up search was done in January 2018. Databases were queried using key words, and their combinations as follows: "Recovery AND Cranioplasty"; "Rehabilitation AND Cranioplasty"; "Timing AND Cognitive AND Cranioplasty"; "Timing AND Motor AND Cranioplasty"; "Early AND Cognitive AND Cranioplasty"; "Early AND Motor AND Cranioplasty"; "Cognitive recovery AND Cranioplasty"; "Motor recovery AND Cranioplasty."

| Study selection and search strategy
All studies reporting motor and/or cognitive recovery after cranioplasty for the patients with cranial defects after DC were included. Systematic reviews that investigated the effects of cranioplasty timing on motor and cognitive recovery in patients underwent cranioplasty were also included. Reports of less than ten subjects, comments, letters, editorial articles, and studies included mainly patients <18 years old were excluded.
At first, search results were summarized and duplicate citations were deleted, together with non-English articles. Then, titles were screened for relevance to motor and cognitive recovery after cranioplasty. Next, abstracts of the remaining articles were read and those not meeting the eligibility criteria were excluded.
The full text of all potential articles was evaluated in depth. In case of uncertainty, or when the abstract was not available, the entire article was read. Two reviewers performed independently the selection of the articles included in this systemic review. The Cohen's kappa score for inter-rater agreement in study selection was computed (Sands & Murphy, 1996). Discrepancy was resolved through discussion.

| Data extraction and outcomes
Data from the studies were collected in an electronic sheet including age, gender, pathology, craniectomy to cranioplasty time interval, surgical site, and pre-and postcranioplasty assessment. Concerning the latter, given that our primary outcome was to compare effects of early and late cranioplasty on the cognitive and motor recovery, we selected studies including as assessment tools Mini-Mental State Examination (MMSE; Folstein, Folstein, & McHugh, 1975) Doyle, 1989), the Glasgow Outcome Scale (GOS; Wilson, Pettigrew, & Teasdale, 1998), the Coma Recovery Scale-Revised (CRS-R; Giacino, Kalmar, & Whyte, 2004), and the Level of Cognitive Functioning Scale (LCF; Sander, 2012). Concerning the motor recovery, we selected studies including Functional Independence Measure (FIM; Keith, 1987) or Barthel Index (BI; Collin, Wade, Davies, & Horne, 1988).
In absence of at least one of the aforementioned assessments, administered both at baseline and at follow-up, we excluded the article from the meta-analysis for inadequate study design.

| Data analysis
The meta-analysis was performed using the metafor package of R (version 3.4.0; the R Foundation for Statistical Computing, Vienna, Austria), setting at α = 0.05 the statistical significance. Statistical averages and relative percentages of all patient characteristics were combined, when and if appropriated. The main analysis concerned the effects of early versus late cranioplasty on motor and cognitive recovery, assessed by comparing the changes in pre-and postcranioplasty scores. For studies reporting multiple test assessment, only the primary outcome was included in the analysis. For studies reported multiple evaluation times before CP, we considered as pre-CP evaluation the one closest to the date of the procedure.
We also performed a subgroup analysis by subdividing the studies according to the time interval from DC to CP: within 3 months and within 6 months. Where the article included both the early and the late cranioplasty groups, we considered the patient's subdivisions of the original study. Otherwise, we subdivided the patients into two groups choosing a threshold according to the median time interval between DC and CP.
Since many studies used different outcome scales, as well as had different sample dimensions, the treatment effect of an intervention was estimated by pooling the standardized mean difference (SMD) with 95% confidence interval (CI). Heterogeneity was quantified by the estimated between-study variance τ 2 , I 2 . When the level of heterogeneity was higher than 75%, we considered the results obtained by the application of the random-effects model. Risk of bias, at outcome level, was graphically investigated by funnel plot.
The kappa score for inter-rater agreement in study selection was 0.88 indicating an "almost perfect agreement," (Landis & Koch, 1977) with a percentage of agreement between the two reviewers of 99.3%.

| Study characteristics
In Only two studies reported complications after cranioplasty (Corallo et al., 2017;Songara et al., 2016), and in both cases, they were observed in patients belonging to the late group.

| Timing effects in cognitive domain
Given that there was no significant heterogeneity for any analyses, a fixed-effects analysis was used. Figure 2 shows meta-analyses of early CP versus late CP, subdivided by type of cognitive outcome.

| Subgroup analysis between 3 and 6 months
We subdivided the studies according to the time interval from DC to CP. Three studies set at 3 months the threshold between early and late CP, including 79 participants (37 in the early and 42 in the late group). Results of the meta-analysis ( Figure 4)

| Timing effects in motor domain
Three studies assess the motor recovery by means of the FIM scale for 77 participants (35 early, 42 late) with an absent statistical heterogeneity (τ 2 = 0 and I 2 = 0%). We found that an early procedure was significantly effective in improving the motor functions compared to a late procedure (SMD = 0.51 [0.05; 0.97], p-value = 0.03), as showed in Figure 5.
As all studies set at 3 months the threshold between early and late CP, we did not perform the subgroup analysis.

| D ISCUSS I ON
The optimal cranioplasty timing is a controversial matter. This choice mainly depends on the presence of complications, as well as the time needed for the recovery.
Several studies define "early cranioplasty" as a cranioplasty performed within 91 days from decompressive craniectomy (Malcolm et al., 2018(Malcolm et al., , 2016Xu et al., 2015). Notably, Xu et al. (2015) sustain that early cranioplasty may reduce the duration of surgery by reducing difficulties in dissecting the scalp flap and fitting the bone flap.
Nonetheless, this early procedure cannot reduce the complications and may even increase the risk of hydrocephalus. Indeed, Tasiou et al. reported that delayed cranioplasty should be preferred to minimize the risk of infection that may be caused by intervening in a still contaminated wound (Tasiou et al., 2014). Malcolm et al. (2016) showed that early cranioplasty, with almost certain hydrocephalus management, has similar complication rates to late cranioplasty.
In the last few years, researcher interest is moving toward the association of cranioplasty with the recovery of consciousness and cognitive function as well as the timing of performing cranioplasty (Huang et al., 2013;Shahid et al., 2018;Songara et al., 2016). Rish et al. (1979) reported that cranioplasty performed within 6 months after DC is as-  (Malcolm et al., 2018). Similarly, many recent studies recommend early cranioplasty because of its association with clinical improvement (Bender et al., 2013;Chibbaro et al., 2011;Liang et al., 2007;Quah et al., 2016), which can be performed as early as 2 weeks postcraniectomy (and in any case not later than 6 months) to lower the overall cost of care by eliminating the need for additional hospital admissions (Beauchamp et al., 2010). Indeed, it would seem that the majority of neurocognitive changes tend to be at their maximum initially and then decline gradually (Di Stefano et al., 2016), given that ipsilateral low cerebral blood flows increased and reached normal levels after CP (Erdogan et al., 2003), raising the recovery of motor and cognitive functioning (Su et al., 2017).
These contradictory results may be attributed to several factors.
First of all the heterogeneity of the population studied, but also the study design features, the choice of surgical approach and operational factors (Sancisi et al., 2009). Thus, our review was aimed at shedding some light on the ongoing debate concerning the right timing to perform cranioplasty and to observe positive effects on cognitive and motor functions. The main question was whether it is reasonable to suggest performing cranioplasty within 90 days from craniectomy to improve the neurological recovery.
Our results showed that such timing is "optimal" only when considering motor outcomes. Indeed, in all studies included in this work, we observed greater positive effects on motor function in the early than late cranioplasty group. On the contrary, to observe a significant cognitive recovery CP should be performed later, although Kim et al. (2017) reported a strong evidence of effects on cognitive functions within 90 days. However, its retrospective study design may lead to a minor reliability since data collected and the measured outcomes are not planned before the study began. Indeed, the follow-up assessment was performed not "after" but "within" 4 weeks; hence, the is not so important in motor functions, it is fundamental in the cognitive domain. To this aim, the assessment should be specifically based on the site and side of lesion, as some brain areas are more strictly related to specific cognitive functions than others (Redolfi et al., 2017).
With regard to the memory tests, the findings suggest that late CP leads to better overall effects. Notably, when focusing on the Digit Span test results, two studies (Corallo et al., 2017;Di Stefano et al., 2016) showed a more significant recovery after 6 months from CP, whereas one study , which has the highest weight, after only 3 months. Thus, we could suppose that a CP performed between 3 and 6 months leads to more significant cognitive recovery, maybe by the restoration of physiological cerebrospinal fluid circulation that, in turn, allows an efficient restoration of blood circulation and, consequently, of the large-scale neuronal networks responsible for cognition (Corallo et al., 2017;Rish et al., 1979). Indeed, before CP, most of the cognitive abnormalities may be due to changes in cerebrovascular and cerebrospinal fluid hydrodynamics, as per the "sinking skin flap syndrome." (Coelho et al., 2014;Erdogan et al., 2003;Juul et al., 2000;Maekawa et al., 1999;Mah & Kass, 2016;Winkler et al., 2000). However, it is possible that the early group of Honeybul (Liang et al., 2007) had an improved outcome after CP in a shorter time due to less severely injured patients than those reported in Corallo et al. (2017) and Di Stefano et al. (2016). Moreover, we have to underline that in Honeybul et al.
(2016) the follow-up assessment was performed within 3 days from CP, as opposed to Corallo et al. (2017) and Di Stefano et al. (2016) who performed it after 1 month, thus explaining the substantial difference that might influence the test scores. Indeed, the difference in follow-up assessment times after CP is another important issue to discuss, since it can affect the measurement. After all, in many studies the greatest improvements were evident many months after cranioplasty and most of the clinical improvement due to cranioplasty is secondary to prolonged effects on brain physiology, rather than immediate changes (Jasey et al., 2018). However, neurorehabilitation programs (if performed) might affect outcomes after longer times (Jolliffe, Lannin, Cadilhac, & Hoffmann, 2018), reinforcing cranioplasty effects on spontaneous cognitive recovery. Su et al. (2017) observed synergetic effects of cranioplasty on TBI patients with rehabilitation training, both in the motor and in the cognitive domains.
Moreover, it is well known that an early neuropsychological rehabilitation that has been performed for an adequate time can affect the outcomes in both severe brain injured and patients with disorder of consciousness (Sancisi et al., 2009).
It is noteworthy to highlight that in postcoma patients, results showed very strong evidence of effects of cranioplasty on cognitive functions, but independently from the timing. Unfortunately, our meta-analysis included only two studies; thus, the findings might not correctly reflect reality. After all, the current literature is poor of studies investigating cranioplasty effects on cognitive functions by means of specific neuropsychological assessment, and, to the best of our knowledge, this is the first attempt to do such analysis.
To summarize, cranioplasty performed within 30 days after initial craniectomy may minimize infection, seizure, and bone flap resorption, whereas waiting >90 days may minimize hydrocephalus but may increase the risk of seizure (Morton et al., 2018;Thavarajah, Lacy, Hussien, & Sugar, 2012). Moreover, at 6-month follow-up patients with severe brain injury got better functional outcomes after early than late CP (Yang, Song, Yoon, & Seo, 2018).
A limitation of the study consists in the fact that we did not include the key word "complications" in our database search, although it has been reported that postsurgical complications after cranioplasty may influence the motor and cognitive recovery and the outcome. Thus, further research is needed to address this important issue.

| CON CLUS IONS
Despite the limitations of this meta-analysis, findings confirm that cranioplasty may improve cognitive and motor recovery. Although 6 months is considered the minimum time to reduce complications, cranioplasty performed within 3 months from decompressive craniectomy may lead to greater effects on motor functions, while for the cognitive domain that the best choice seems to be from three to 6 months, especially if the patient underwent neuropsychological rehabilitation. Future prospective larger sample studies are needed to standardize the best timing of performing CP in patients with different disorders, also by using specific psychometric approaches in order to improve functional recovery and thus patient's quality of life.

ACK N OWLED G M ENTS
The authors wish to thank Antonina Donato for having revised the manuscript in the English language and Nunzio Muscarà for his useful suggestions.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no financial or other conflict of interests in relation to this research and its publication.