Treatment of unicameral bone cyst: systematic review and meta analysis

Purpose Different treatment modalities have been utilized to treat unicameral bone cyst (UBC), but evidence has not been fully described to support one treatment over another and the optimal treatment is controversial. The aim of this quantitative systematic review was to assess the effectiveness of different UBC treatment modalities. Methods We utilized Pubmed to isolate retrospective studies on patients with UBC who received any kind of treatment. The included studies needed to have a minimum sample size of 15 patients, and have provided data on radiographic healing outcome. Results Sixty-two articles were selected for the meta-analysis from a total of 463 articles. The cumulative sample size was 3,211 patients with 3,217 UBC, and male to female ratio was 2.2:1. The summary or pool estimate of methylprednisolone acetate (MPA) injection resulted in a healing rate of (77.4 %) that was comparable to bone marrow injection (77.9 %). A higher healing rate was observed with MPA injection when inner wall disruption was performed. The pool estimate of bone marrow with demineralized bone matrix injection was high (98.7 %). UBC healing rate after surgical curettage was comparable whether autograft or allograft was utilized (90 %). UBC treatment with flexible intramedullary nails without curettage provided almost 100% healing rate, while continuous decompression with cannulated screws provided 89 % healing rate. Conservative treatment indicated a healing rate of 64.2, 95 % CI (26.7–101.8). Conclusions Active treatment for UBC provided variable healing rates and the outcomes were favorable relative to conservative treatment. Due to the heterogeneity of the studies and reporting bias, the interpretation of these findings should be handled with caution.

The purpose of treatment is to restore bone strength, cortical thickness and obliteration of the cyst; and based on these criteria, healing of UBC was classified by Neer [3] and Capanna [6]. Complete filling of the cyst with restoration of cortical thickness was described as healed cyst. When small radiolucent areas persist with good bone strength, the cyst was described as partially healed. When the UBC had continuous bone lucency and thin or broken cortex and did not respond to the treatment, the cyst was considered persistent. Cyst recurrence was defined when a cavity and expansile remodelling and thinning of the cortex develop after full obliteration.
Various treatment modalities have been utilized to treat UBC, including surgical and nonsurgical interventions with variable reported healing rates. UBC has frequently been examined in the literature, yet no clearcut evidence supports one modality over another. The aim of this current study was to review the literature and examine the published articles referenced in Pubmed about the healing outcome of UBC management with every treatment modality. A qualitative systematic review was performed to screen all the articles and to select eligible articles for a meta-analysis on UBC treatment outcome. We postulated that treatment of UBC leads to better rates of cyst healing.

Search techniques and terms:
The studies were identified from Pubmed until 2012 using the following search terms: unicameral bone cyst, simple bone cyst and solitary bone cyst. The articles were searched by a single investigator (MK), but were reviewed by two investigators (MK, AK) independently and data were extracted during the review. The articles selected were Fig. 1 Schema illustrating the details of the articles for the systematic review and the metaanalysis reviewed and screened for pertinent information based on specific criteria (Fig. 1).

Eligibility criteria
The extracted articles were examined based on: (a) nonexperimental retrospective studies on unicameral bone cyst in any anatomical location and any kind of treatment, (b) studies conducted in any geographic location, (c) studies with abstracts published in English, (d) studies that examined consecutive patients with sample size equal or greater than 15 patients, (e) studies with follow up more than 6 months, and (f) studies that demonstrated the radiographic outcome of treatment that can be classified into (healed, partially healed, persistence or recurrence).
Studies with sample size smaller than 15 were excluded in order to avoid sparse data bias; implying negative finding as a result of small sample size and limited statistical power. Case reports were excluded, as they did not examine consecutive patients. When more than one study was conducted in the same institution, the articles were examined for a possibility of duplicate report on the same patient sample, and data were retrieved from the most complete or recent study.

Study quality assessment
The included studies were assessed for study design, sampling techniques and data on patients' characteristics and radiographic healing status. We determined the objectivity of each study, whether the aims or the purposes were clearly stated, and whether the studies presented factual results. We also examined for adequate statistical analysis and how the outcomes were determined. Studies were assessed for confounding factors that might have influenced the outcome and for any bias, including selection, information and misclassification biases. This quality assessment provided comparable measures to the preferred reporting for systematic reviews and meta-analysis (PRISMA statement) [7].

Study variables
Each study was identified by the first author's last name and the year of publication. Study populations were stratified by sex, age at treatment, cyst location, cyst activity (active or latent), follow-up time and radiographic healing classified by Neer [3] and Capanna [6] criteria. The cohort of each study was categorized into two groups based on the final healing status of UBC: success-good outcome included healed and partially healed UBCs, and failurepoor outcome included persistence and recurrent UBCs.

Treatment groups
All treatment modalities were included in this study (conservative and surgical intervention). Studies that utilized injection as a method of treatment were categorized in one treatment group and subgroups were identified based on the material utilized for injection, including; methylprednisolone acetate (MPA) injection, bone marrow injection, combined MPA with bone marrow, combined bone marrow with demineralized bone matrix (DBM), and combined bone marrow with MPA and DBM. The technique of injection procedure was reviewed in each article to identify whether inner wall disruption was performed or not, and based on this criterion, every injection subgroup was subcategorized into injection with and without inner wall disruption. Studies on surgical treatment were examined and subgroups were identified, including: curettage only, curettage with bone graft (autograft versus allograft), curettage with bone substitution (calcium sulphate pellets, calcium phosphate pellets, Hydroxyapetite, ceramic substitute), curettage with myoplasty, cyst excision, flexible intramedullary nailing (IM nail) and continuous decompression with cannulated screws.

Comparative studies
Retrospective comparative studies that presented outcome of different treatment modalities were selected for further evaluation by combining the patients of each treatment group. Comparative assessment performed included: MPA vs. bone marrow injection, MPA injection with vs. without inner wall disruption, curettage with autograft vs. curettage with allograft, MPA injection vs. curettage with autograft, MPA injection vs. curettage with allograft.

Anatomical location
We identified the studies that provided a list of UBC cases demonstrating age, sex, cyst location treatment modality and radiographic healing outcome. The UBC cases were categorized based on the anatomic location for further assessment.

Sample size and power estimation
Four hundred and sixty-three articles published between 1942 and 2012 were identified for the qualitative systematic review. The number of studies included in the quantitative systematic review (meta-analysis or quantitative evidence synthesis) was represented by K = 62. The combination of study sizes (UBC) that constituted this meta-analysis was represented by n = 3,217 cysts. To estimate the power of this meta-analysis, we used 95 % confidence interval (5 % error), effect size of 20 % (0.2), and a random effect meta-analytic technique (DerSimonian and Laird). With these parameters, and with the combined sample sizes from the individual studies (n) for each treatment group, we obtained sufficient power ([80 %) to detect a difference between success (good) and failure (poor) outcomes in terms of UBC healing.

Statistical analysis
Frequency and percentages were employed to summarize nominal variables, while mean and standard deviation were used to describe the continuous variables, including age and follow-up.
We tested the hypotheses to determine the effectiveness of different treatment modalities. Both the fixed method of Peto and Mantel-Haenszel, and the random effect method of Dersimonian-Laird meta-analytic techniques were utilized. The fixed effect method was applicable to the summary point estimate when the studies that constituted the meta-analysis were assessed to be homogenous (implying heterogeneity test I with a significance level[0.05). Unless the homogeneity test was performed, the fixed method was inapplicable. The random effect method was utilized when the homogeneity assumption was not met.
The significance level was \0.05 and all tests were twotailed. All analyses were performed using STATA 12.0 (StataCorp LP, College Station, TX).
The 62 studies of the meta-analysis generated 3,211 patients with 3,217 UBCs. There were 1,709 boys and 765 girls. While some studies did not provide enough data on gender and the proportions did not add up to 100 %, the estimated male to female ratio was 2.2:1. The clinical presentation was with a pathologic fracture in 1,107 UBCs, pain in 204 cysts, and incidental finding in 133 UBCs (the clinical presentation was not specified in all studies). Seven hundred and twenty-four UBC were classified as active and 535 cysts were latent. The overall average age at treatment was 10.9 ± 2.9 years, and the average follow up was 6.1 ± 4.2 years (not all the studies provided data on age at surgery and follow-up).
The most common location of UBC was in the humerus (59.2 %), followed by the femur (25.9 %). The calcaneus was affected in 6 %, the tibia in 4 %, the fibula in 1.5 %, the radius in 1.1 %, the pelvis in 0.9 %, and the other locations were affected in 1.4 % ( Table 1).

Outcome of treatment based on the anatomic location
The outcome of treatment based on UBC anatomic location was also examined, including (humerus, femur, calcaneus, fibula, tibia and radius). Some studies did not present healing status for each anatomic location of the cyst; therefore, the numbers and percentages did not add up to 100 % of the total number of UBC of the meta-analysis. Tables 3, 4, 5, and 6 present the findings on healing outcome stratified by sex, activity level and treatment modalities. A few studies examined the fibula, radius, tibia and pelvic locations.
Quantitative results (Pool estimates) Table 7 illustrates the percentages and 95 % CI of each treatment modality representing the successful outcome (healed and partially healed UBC). The combination of studies in almost all the categories of the meta-analysis indicated satisfactory healing of UBC after each specific treatment. The summary or pool estimate of overall injection category was 81.3, 95 % CI (77.9-84.8), implying a statistically significant 81 % success rate for overall injection (p \ 0.05). MPA injection resulted in a healing rate of 77.4, 95 % CI (72.7-82.2) (Fig. 6). The effect of inner wall disruption with MPA injection was examined and higher healing rate was observed when the disruption was performed. The pool estimate of bone marrow injection was comparable to MPA injection, 77.9, 95 % CI (65.9-89.9) (Fig. 7). Only one study reported on the outcome of bone marrow injection with inner wall disruption and success rate was 78.6 % [39]; therefore, no pool estimate was performed for bone marrow injection with inner wall disruption. The pool estimate of bone marrow with DBM injection was high, 98.7, 95 % CI (95.7-101.7) relative to injection with either MPA or bone marrow.  Surgical curettage was examined according to the utilization of bone graft or bone substitute, and the pool estimates of all the subcategories were comparable. Other surgical managements were also assessed, including cyst excision, continuous decompression with IM nail or cannulated screws, and cyst wall trepanation (Table 3). The pool estimate of conservative treatment was 64.2, 95 % CI (26.7-101.8) (Fig. 8).

Discussion
The effect of various treatment modalities on UBC management has been repeatedly examined in the literature, but the natural history of UBC and the optimal treatment remain unclear. The conventional knowledge about UBC is that it is a self-limited benign bone lesion [370]; however, the healing rate without treatment (observational management) is undetermined and most of the studies in this review demonstrated better results with active treatment.
With the purpose to better understand the role of each treatment modality of UBC, we aimed in this quantitative systematic review to synthesize the literature and assess the MPA methylprednisolone, DBM demineralized bone matrix, IM nail Intramedulary nail, Screws continuous decompression with cannulated screws a The exact location of the cyst was not determined in all the papers; therefore, the percentages did not add up to 100 % MPA methylprednisolone, DBM demineralized bone matrix, IM nail Intramedulary nail, Screws continuous decompression with cannulated screws a The exact location of the cyst was not determined in all the papers; therefore, the percentages did not add up to 100 % natural history of UBC, as well as evaluate the various treatment methods.
The main purpose of meta-analysis is to improve the power by combining homogeneous studies and arriving at a summary effect or pool estimate. We performed both systematic review and meta-analysis to examine the effect of several treatment modalities on UBC. Specifically, we tested the hypothesis regarding the effectiveness of different treatment modalities on UBC healing. Our metaanalysis has some relevant findings: overall active treatment improves UBC healing, regardless of anatomic site and the type of treatment compared to observational management. In this meta-analysis, observational treatment leads to a healing rate of 64 %. However, due to the variability of the studies, the rate of healing after conservative treatment can be as low as 27 % and as high as 100 %. The healing rate after conservative treatment varied in different anatomic location. Higher failure rate of conservative treatment was found in calcaneal and humeral UBCs compared to femoral UBCs (although the number of femoral UBCs was small, 11 out of 211 cysts were treated conservatively).
The indication to treat UBC is to prevent a pathologic fracture and to manage symptoms, especially pain. Historically, surgical curettage and cyst excision with bone graft were the optimal choices to treat UBCs. Based on this meta-analysis, UBC healing rate after surgical curettage was comparable (90 %) whether autograft or allograft was utilized. The utilization of MPA in the treatment of UBC was first described by Scaglietti [454]. The role of MPA is to reduce the production of cyst fluid from the inner wall DBM demineralized bone matrix, Screws Continuous decompression with cannulated screws a The exact location of the cyst was not determined in all the papers, therefore the percentages did not add up to 100 %   cyst, which enhances bone healing [352]. On the other hand, bone marrow injection was proposed to accelerate healing, given the osteogenic potential of red marrow [465,466]. In this meta-analysis, the healing rate of UBC was comparable after injection with bone marrow or MPA (77 % The utilization of IM nails or cannulated screws for the treatment of UBC was also described, and the rationale was that theoretically, UBCs develop from venous blockage and increase of fluid pressure inside the cyst [5]. In this meta-analysis, we found that treatment with IM nails without curettage provided almost 100 % healing rate of UBC in long bones. IM nails were described as an efficient acute treatment for UBCs that present with a fracture [11,16,67]. Cyst wall drilling and trepanation also provided a healing rate of almost 100 %; however, this healing rate was estimated based on only two studies [9,49].
Although this meta-analysis may be interpreted as a comparison between different treatment modalities such as MPA and bone marrow injection, this is incorrect. Studies that tested hypothesis with respect of comparative effectiveness of treatment were not available; therefore, we were unable to use formal meta-analysis method to examine the differences in outcomes of treatment with different treatment modalities. Future studies designed to compare different modalities of treatment are recommended, and will provide the opportunity for these questions to be addressed.
Our systematic review is limited in the sense that different studies presented with different sampling designs and different study protocols. For example, data were not  Also, the variability of surgical techniques and type of bone graft (autograft or allograft) has a tendency of introducing a confounding that may lead to bias in our outcomes estimation and interpretation. In addition, we were unable to examine the need for further intervention and recurrence rate in each treatment modality, as data were not clearly specified in the reviewed articles. Also UBC activity as described by the distance from the growth plate may be an important factor that may affect on cyst healing after different treatment modalities. Some of the studies provided data on cyst activity, but only a few of these provided data on the outcome of treatment stratified by cyst activity. The other limitations are the heterogeneity of the patients in terms of UBC anatomic locations, cyst size, sex variability, and duration of treatment and follow-up. We could have performed subgroup meta-analysis in terms of the impact of sex or anatomic site of the UBC on the outcome implying cyst healing. However, since many studies did not provide relevant data, performing such subgroup analysis could have ended up in biased results regarding such effects. Considering this inability to perform the observed subgroup analyses, we recommend future studies to provide complete data on patients' sex and specific anatomic location. Heterogeneity remains an important factor to be considered in the conduct and interpretation of meta-analysis. There is no meta-analytic study without some influence of heterogeneity. Several studies used in our meta-analysis showed significant heterogeneity; however, our meta-analysis is not completely driven by this variability. A metaanalytic finding could be reliable and valid despite heterogeneity when appropriate analysis is used in the summary estimate. We applied the random effect meta-analysis given the heterogeneity of studies included in this metaanalysis. Consequently, by adjusting for the differences between studies with the random effect method approach, we have produced relatively standard and valid meta-analytic results on the effect of treatments on the outcomes of UBC. Whereas qualitative systematic review is not intended to quantify results of reviewed literature, we attempted to summarize proportions with regard to UBC healing, and caution is recommended in the interpretation of the results of our systematic review.
We are unaware of previous meta-analysis in this direction to either support or negate our findings. Clinicians will be interested in assessing the possibility of the results of this meta-analysis in developing guidelines regarding the treatment of UBC. Our meta-analysis of published literature on UBC treatment indicates improved healing rates among treated patients relative to the observational management. Healing rate was found to be comparable in studies that utilized bone marrow injection or MPA, and higher rate of healing was found when DBM was added. Surgical curettage resulted in healing rate of 90 % with the utilization of autograft, allograft or any bone substitution material. Healing rate was also high with the utilization of IM nails.
Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.