Skip to main content
Free AccessTonsillectomy

Craniofacial Contribution to Residual Obstructive Sleep Apnea after Adenotonsillectomy in Children: A Preliminary Study

Published Online: by:21


Study Objectives:

Pediatric obstructive sleep apnea (OSA) is frequently associated with adenotonsillar hypertrophy, and the fact that about 30% of affected children continue to show OSA after adenotonsillectomy (AT) suggests the presence of some other predisposing factor(s). We hypothesized that abnormal maxillofacial morphology may be a predisposing factor for residual OSA in pediatric patients.


A total of 13 pediatric OSA patients (9 boys and 4 girls, age [median (interquartile range)] = 4.7 (4.0, 6.4) y, body mass index (BMI) z score = -0.3 (-0.8, 0.5)) who had undergone AT were recruited for this study. Maxillomandibular size was measured using an upright lateral cephalogram, and correlations between size and the apnea hypopnea index (AHI) values obtained before (pre AT AHI) and about 6 months after AT (post AT AHI) were analyzed.


AHI decreased from 12.3 (8.9, 26.5)/h to 3.0 (1.5, 4.6)/h after AT (p < 0.05). Residual OSA was seen in 11 of the 13 patients (84.6%) and their AHI after AT was 3.1 (2.7, 4.7)/h. The mandible was smaller than the Japanese standard value, and a significant negative correlation was seen between maxillomandibular size and post AT AHI (p < 0.05).


These findings suggest that the persistence of OSA after AT may be partly due to the smaller sizes of the mandible in pediatric patients. We propose that the maxillomandibular morphology should be carefully examined when a treatment plan is developed for OSA children.


Maeda K, Tsuiki S, Nakata S, Suzuki K, Itoh E, Inoue Y. Craniofacial contribution to residual obstructive sleep apnea after adenotonsillectomy in children: a preliminary study. J Clin Sleep Med 2014;10(9):973-977.


The prevalence of obstructive sleep apnea (OSA) in children has been estimated to be 1% to 3%,1,2 and unstable sleep brought about by this disorder may cause decreased neurocognitive performance and behavioral impairments, both of which may lead to degraded school performance.38 Most cases of OSA in children have been believed to be associated with adenotonsillar hypertrophy, and adenotonsillectomy (AT) has been widely accepted as the first choice of treatment. However, many reports have suggested that 20% to 40% of pediatric OSA patients continue to show OSA after AT.912 This phenomenon implies that other predisposing factors may be present in patients with pediatric OSA.

An abnormality of the maxillofacial morphology, especially small mandibular size relative to oropharyngeal soft tissue, has been revealed to play an important role in the development of OSA in adults.13 Similarly, a small mandible has been reported to be responsible for the familial aggregation of OSA in both adult and pediatric patients.14 These findings raise the possibility that an abnormal maxillofacial morphology may be a predisposing factor for the residual OSA in the pediatric population. Thus, the purpose of this preliminary study was to clarify whether the persistence of OSA after AT is related to an abnormality of the maxillofacial morphology in pediatric OSA patients. In addition, we discuss whether the design of this preliminary study is reasonable to support further studies from which a definitive conclusion can be drawn.


Current Knowledge/Study Rationale: The persistence of obstructive sleep apnea (OSA) after adenotonsillectomy (AT) in children may be associated with the maxillofacial morphology, as in adult OSA patients. Therefore, we evaluated the relationship between the severity of residual OSA and the maxillofacial morphology by using lateral upright cephalography.

Study Impact: A significant negative association between maxillomandibular size and post-treatment AHI suggests that OSA likely remains after AT when pediatric OSA is associated with a smaller mandible. Therefore, an approach to encourage mandibular growth (e.g., orthodontic treatment) could be considered.


Patient Enrollment and Polysomnography

The study protocol was approved by the ethics committee of the Neuropsychiatric Research Institute, Tokyo, Japan. In the protocol, we stated that we used retrospective clinical data from another institute (Department of Internal Medicine, Takaoka Clinic, Nagoya, Japan). Written informed consent was obtained from each participant after the aim and potential risks were fully explained.

The Japanese patients enrolled in this study were diagnosed as having OSA based on the following criteria: (1) an apnea hypopnea index (AHI) ≥ 1 (/h) on overnight polysomnogram (PSG)15 using the Alice4 (Respironics, Inc., Murrysville, PA); (2) symptoms related to pediatric OSA (snoring, nocturnal enuresis, difficulty in waking up in the morning, emotional labiality or instability); and (3) obviously hypertrophic adenoid and/or palatine tonsil, both of which were diagnosed by an expert board-certified sleep disorder otolaryngologist. At a minimum, grade I adenoid and/or grade II tonsil based on the classification scheme described by Broadsky16 was required for the indication of AT treatment.17 Patients with families who consented to the proposal for AT underwent the operation at Takaoka Clinic, Nagoya, Japan (N = 43). The following conditions were set as exclusion criteria: having genetic syndrome, central nervous system abnormality, craniofacial syndrome, pulmonary disease, hormone deficiency, and neurodevelopmental disability. Additionally, patients with excessive deviation of the left and right mandibular condyles in both the sagittal and vertical planes and an unclear silhouette of the condyle on the lateral cephalogram were excluded. Consequently, 13 of 43 consecutive OSA patients (30.2%, 9 boys and 4 girls) were selected for the analyses during the period of May 2001 to March 2011. Based on results of clinical interviews, none of the patients had a family history of OSA.


Tonsillectomy was performed with a combination of sharp and blunt dissection primarily using electrocautery, and adenoidectomy was performed using a microdebrider.17 Postoperative PSG was performed in the routine clinical setting a few months after surgery (median [interquartile range] = 7.0 [6.0, 8.0] months). Children with an AHI < 1/h were considered to be cured.15


An upright lateral cephalogram was taken in the routine clinical setting for each patient to confirm the size of adenoid tissue before AT. Cephalographic images were obtained for each patient in the upright position with a natural head posture using a pair of earpieces at the end of expiration.18 The exposure parameters were arranged to clearly visualize the adenoid tissue and specific bony landmarks. Cephalometric parameters that reflected the size of the mandible were defined in this study (Figure 1). To exclude age-related differences in maxillofacial size and to compare the sizes with those in normal children, some parameters (Ar-Pog, Ar-Go, Go-Me) were normalized by age using the standard cephalometric values in Japanese children.19 The normalized cephalometric values (Ar-Pog, Ar-Go, Go-Me) were expressed as follows:

Figure 1
Figure 1

(A) Definitions of maxillofacial variables. Sella (S) = the center of the hypophyseal fossa, Nasion (N) = the junction of the frontonasal suture at the most posterior point on the curve at the bridge of the nose, Point B (B) = the most posterior point to a line from the infradentale to the pogonion on the anterior surface of the symphyseal outline of the mandible; Articulare (Ar) = the point of intersection of the images of the posterior border of the ramus process of the mandible and the inferior border of the basilar part of the occipital bone; Pogonion (Pog) = the most anterior point on the mandibular symphysis; Gonion (Go) = the most outward point on the angle formed by the junction of the ramus and body of the mandible on its posterior, inferior aspect; Menton (Me) = The most inferior point on the mandibular symphysis; SNB = angle between the N–S line and the line from Point B to N, maxillary length (Ar-A) = distance between Ar and A; mandibular length (Ar-Pog) = distance between Ar and Pog; mandibular height (Ar-Go) = distance between Ar and Go; mandibular body length (Go-Me) = distance between Go and Me; mandibular CSA (Ar-Go-Me) = cross-sectional area of a triangle enclosed by Ar-Go-Me. (B) Definitions of Adenotonsillar variables. AN ratio = represents relative values of adenoidal hypertrophy. A: distance from the point of maximum thickness of the adenoids, along a line perpendicular to the line that is drawn along the straight part of the basiocciput, N: distance from the posterior nasal spine (PNS) to the anterior-inferior edge of the spheno-basiooccipital synchondrosis. PNS-A = distance from PNS to the surface of the adenoid shadow along the line drawn from PNS to basin (Ba). TP ratio = relative value of tonsillar hypertrophy. T: width of the tonsil at the level of the upper axis, P: the airway space at the level of the upper axis.

Normalized cephalometric value = [(cephalometric value of study sample) – (mean value of the standard cephalometric value in Japanese children)] / (standard deviation of the standard cephalometric value in Japanese children).

Adenotonsillar hypertrophy was also evaluated.20 All radio-graphs were hand-traced on acetate paper over a light-viewing box. Cephalometric analyses were also made in a blinded manner by a single investigator who had no knowledge of the patients' information.


Prior to data acquisition, the intra-rater reliability of cephalometric measurements was assessed by intraclass correlation coefficient (ICC) statistics21 based on 3 calculations with 20 cephalometric measurements. The measurement errors for the variables were considered to be within the acceptable range based on an ICC of 0.9970 (95% CI: 0.9951-0.9982) for cephalometric measurements. Pre AHI (AHI value before AT), age, mandibular CSA, and PNS-A were not normally distributed according to the Kolmogolov-Smirnov test.

The differences between the pre and post AHI (AHI value after AT) values was assessed by the Wilcoxon signed-rank test. A Spearman rank correlation coefficient analysis was used to evaluate the relationship between the severity of OSA and the maxillofacial morphology by assessing the correlations between pre and post AT AHI and age, body mass index (BMI) z-score and cephalometric variables, all of which were obtained from a lateral cephalogram taken before surgery. A p-value < 0.05 was considered to indicate statistical significance for all of the analyses.

To verify the adequacy of the sample size and to clarify the statistical power of the results in our study, we also conducted a sample size calculation and post hoc power analysis using G*Power software ( According to previous reports, the inclusion of either approximately 10 participants or 10% of the final study size is appropriate for pilot studies.23,24 Since the calculations showed at least 134 patients would be needed to detect a correlation at a medium effect size (0.3) and a statistical significance level of 5%, 13 patients were considered to be sufficient for this preliminary study. The results of a post hoc power analysis for a Spearman rank correlation coefficient analysis with pre AT were as follows: BMI z-score = 0.597, Ar-Pog = 0.647, Ar-Go = 0.617 and Age = 0.630, and those with post AT were Ar-Go = 0.595 and Go-Me = 0.588.


Thirteen of 43 patients who met our inclusion/exclusion criteria were selected for the analyses. Among the 13 patients, the median (interquartile range) age was 4.7 (4.0, 6.4) years, and the BMI z-score was -0.3 (-0.8, 0.5). The interval between the 2 PSG evaluations was 7.0 (6.0, 8.0) months. During the study, obvious weight gain was not seen in any of the patients. Pre AT AHI was 12.3 (8.9, 26.5)/h, while post AT AHI was 3.0 (1.5, 4.6)/h. Although a significant reduction of AHI was seen after AT compared to before AT (p = 0.001), 11 patients had residual OSA (AHI > 1/h).

The median values of cephalometric variables are shown in Table 1. The cephalometric variables that were normalized by the standard cephalometric values in Japanese children (Ar-Pog, Ar-Go, and Go-Me) were 0.4 (0.2, 1.0), 1.2 (0.5, 1.7), and -1.1 (-1.5, -0.8), suggesting that the study samples had smaller mandibles than normal Japanese children. The cross-sectional area of the mandible (mandibular CSA) was 646.1 (636.1, 795.1) mm2.

Table 1 Patient demographics

Table 1

BMI z-score, Ar-Pog, and Ar-Go were negatively correlated with pre AT AHI; and Age, Ar-Go, and Go-Me were negatively correlated with post AT AHI. (BMI z-score: ρ = -0.593, p = 0.033, Ar-Pog: ρ = -0.802, p = 0.001, Ar-Go: ρ = -0.714, p = 0.006, Age: ρ = -0.702, p = 0.008, Ar-Go: ρ = -0.587, p = 0.035, Go-Me: ρ = -0.557, p = 0.048, Table 2 and Figure 2).

Table 2 Correlations between AHI pre- and post-AT and descriptive variables including age, BMI z-score, and cephalometric variables

Table 2
Figure 2: The relation of the mandible height (Ar-Go) and body length (Go-Me) to the apnea hypopnea index (AHI) at post adenotonsillectomy (AT).

Linear correlations were observed between post AT AHI and both Ar-Go and Go-Me. Black dots indicate pre AT AHI, and white dots indicate post AT AHI.


This is the first demonstration that pediatric OSA patients who did not achieve a sufficient improvement in their disorder with AT tended to have a smaller mandible. Although many reports have shown relationships between residual OSA after AT and various background factors such as high BMI and allergic disease,17,25 no previous reports have suggested that maxillo-mandibular morphology is a factor for residual OSA after AT in children.

The present study showed that mandibular length and height (Ar-Pog and Ar-Go) had a significant negative correlation with the pre-AHI value, while the height and body length of the mandible (Ar-Go and Go-Me) were correlated with post-AHI. One possible explanation for this finding is that the pre-AHI value may reflect the additive effects of adenotonsillar hypertrophy and anatomical factors, especially maxillomandibular size, on upper airway patency. On the other hand, after adenotonsillar hypertrophy was removed by AT, the residual OSA was thought to reflect only the effect of anatomical factors. These results might suggest that the maxillomandibular morphology may highly affect the upper airway patency.

Our results are summarized in Figure 3. In the anatomical balance theory, the skeletal tissues form a “rigid box,” and the upper airway can be regarded as a “collapsible tube” encircled by soft tissue including the adeontonsillar tissue and tongue within the rigid box.26,27 The cross-sectional area of the tube, which reflects upper airway patency, depends on the amount of soft tissue relative to the skeletal size.2830 AT can be successful when the upper airway is narrowed only by a hypertrophic adenoid and palatine tonsil (Figure 3A and 3B). However, when a patient has a smaller mandible in addition to a hypertrophic adeontonsillar tissue, AT cannot lead to a sufficient improvement of OSA. The narrowing of the upper airway due to excessive soft tissues with a smaller skeletal size cannot be ameliorated by AT alone (Figure 3C and 3D). Our results suggest that it may be important to evaluate the maxillomandibular morphology prior to AT to predict treatment outcome. Furthermore, Japanese adult OSA patients tend to have small mandible rather than maxillary constriction, both of which are considered as the pathogenesis of OSA.31,32 Therefore, the use of orthodontic treatment to facilitate the growth of the mandible33 may become a radical treatment option for residual OSA after, and even before, AT in these patients.

Figure 3: Schematic illustrations of the status of upper airway patency associated with the success/failure of AT treatment.

Schematic diagram shows the upper airway and its surrounding soft tissue including the adeontonsillar tissue and skeletal tissue in the horizontal plane at the mandibular level. An upper airway with adenotonsillar hypertrophy but without a skeletal problem (A) can be clearly enlarged by AT (B), since the only obstacle to maintaining upper-airway patency, a hypertrophic adeontonsillar tissue, can be overcome by AT. An upper airway with both adenotonsillar hypertrophy and a skeletal problem (C) can not be sufficiently enlarged by AT alone (D), since the treatment can not change the size of the skeletal tissue.

One of the significant study limitations is patient recruitment. We examined only a small number of patients; thus, the results were of low power, although the number was sufficient for a preliminary study. Further studies with larger sample sizes (e.g., multicenter studies) would be needed and ideal to confirm our results. Another limitation is that we evaluated the maxillomandibular morphology only by cephalogram. Other studies have used 3-dimensional imaging techniques such as computerized tomography (CT) in adult OSA patients.34 However, CT examination, which requires exposure to a larger amount of X-ray and the need for a longer period of immobility during the scan than a cephalometric examination, is not easily applicable to children. In addition, although CT examination provides more precise information about soft tissue than cephalometric examination, cephalometric equipment is used more commonly than CT equipment, and cephalometric evaluation is more acceptable than CT evaluation for both medical and dental personnel.

In conclusion, our preliminary result suggests that residual OSA after AT was related to a smaller maxillomandibular size in pediatric OSA patients. An evaluation of the maxillomandibular morphology prior to starting treatment may be important for the development of a treatment plan for children with OSA.


This was not an industry supported study. This study was supported by Grants-in-Aid for Young Scientists from the Ministry of Education, Culture, Sports, Science & Technology [# 21792107] and the Japan Society for the Promotion of Science (JSPS) [# 24792126] to Dr. Maeda, by a Grant-in-Aid for Scientific Research from the JSPS [# 25515010] to Dr. Tsuiki, by a Grant-in-Aid for Scientific Research from the JSPS [# 25461180] to Dr. Itoh, and by a Grant-in-Aid for Scientific Research from the JSPS [# 22591312 and 50213179] to Dr. Inoue. The sponsors had no role in the design of the study, the collection and analysis of the data, or preparation of the manuscript. This work was performed at the Neuropsychiatric Research Institute, Tokyo, Japan. The authors have indicated no financial conflicts of interest.


  • 1 Ali NJ, Pitson DJ, Stradling JRSnoring, sleep disturbance, and behaviour in 4-5 year olds. Arch Dis Child; 1993;68:360-6, 8280201.

    CrossrefGoogle Scholar
  • 2 Hultcrantz E, Löfstrand-Tideström B, Ahlquist-Rastad JThe epidemiology of sleep-related breathing disorder in children. Int J Pediatr Otorhinolaryngol; 1995;32Suppl:S63-6, 7665302.

    CrossrefGoogle Scholar
  • 3 Rhodes SK, Shimoda KC, Waid LRet al.Neurocognitive deficits in morbidly obese children with obstructive sleep apnea. J Pediatr; 1995;127:741-4, 7472827.

    CrossrefGoogle Scholar
  • 4 Gozal DSleep-disordered breathing and school performance in children. Pediatrics; 1998;102:616-20, 9738185.

    CrossrefGoogle Scholar
  • 5 Blunden S, Lushington K, Kennedy D, Martin J, Dawson DBehavior and neurocognitive performance in children aged 5-10 years who snore compared to controls. J Clin Exp Neuropsychol; 2000;22:554-68, 11094391.

    CrossrefGoogle Scholar
  • 6 Owens J, Spirito A, Marcotte A, McGuinn M, Berkelhammer LNeuropsychological and behavioral correlates of obstructive sleep apnea syndrome in children: a preliminary study. Sleep Breath; 2000;4:67-78, 11868122.

    CrossrefGoogle Scholar
  • 7 Blunden S, Lushington K, Kennedy DCognitive and behavioural performance in children with sleep-related obstructive breathing disorders. Sleep Med Rev; 2001;5:447-61, 12531153.

    CrossrefGoogle Scholar
  • 8 Lewin DS, Rosen RC, England SJ, Dahl REPreliminary evidence of behavioral and cognitive sequelae of obstructive sleep apnea in children. Sleep Med; 2002;3:5-13, 14592247.

    CrossrefGoogle Scholar
  • 9 Brietzke SE, Gallagher DThe effectiveness of tonsillectomy and adenoidectomy in the treatment of pediatric obstructive sleep apnea/hypopnea syndrome: a meta-analysis. Otolaryngol Head Neck Surg; 2006;134:979-84, 16730542.

    CrossrefGoogle Scholar
  • 10 Mitchell RB, Kelly JOutcome of adenotonsillectomy for obstructive sleep apnea in obese and normal-weight children. Otolaryngol Head Neck Surg; 2007;137:43-8, 17599563.

    CrossrefGoogle Scholar
  • 11 Friedman M, Wilson M, Lin HC, Chang HWUpdated systematic review of tonsillectomy and adenoidectomy for treatment of pediatric obstructive sleep apnea/hypopnea syndrome. Otolaryngol Head Neck Surg; 2009;140:800-8, 19467393.

    CrossrefGoogle Scholar
  • 12 Marcus CL, Moore RH, Rosen CLet al.A randomized trial of adenotonsillectomy for childhood sleep apnea. N Engl J Med; 2013;368:2366-76, 23692173.

    CrossrefGoogle Scholar
  • 13 Ferguson KA, Ono T, Lowe AA, Ryan CF, Fleetham JAThe relationship between obesity and craniofacial structure in obstructive sleep apnea. Chest; 1995;108:375-81, 7634870.

    CrossrefGoogle Scholar
  • 14 Guilleminault C, Partinen M, Hollman K, Powell N, Stoohs RFamilial aggregates in obstructive sleep apnea syndrome. Chest; 1995;107:1545-51, 7781344.

    CrossrefGoogle Scholar
  • 15 Marcus CL, Omlin KJ, Basinki DJet al.Normal polysomnographic values for children and adolescents. Am Rev Respir Dis; 1992;146:1235-9, 1443877.

    CrossrefGoogle Scholar
  • 16 Brodsky LModern assessment of tonsils and adenoids. Pediatr Clin North Am; 1989;36:1551-69, 2685730.

    CrossrefGoogle Scholar
  • 17 Tagaya M, Nakata S, Yasuma Fet al.Children with severe or moderate obstructive sleep apnoea syndrome show a high incidence of persistence after adenotonsillectomy. Acta Otolaryngol; 2012;132:1208-14, 23025449.

    CrossrefGoogle Scholar
  • 18 Miyamoto K, Ozbek MM, Lowe AA, Fleetham JAEffect of body position on tongue posture in awake patients with obstructive sleep apnoea. Thorax; 1997;52:255-9, 9093342.

    CrossrefGoogle Scholar
  • 19 Ozaki M, Ishi K, Yoshida Y, Motokawa WStudies on craniofacial growth, based on chronological age in children. Jpn J Pediatr Dent; 1991;29:594-606.

    Google Scholar
  • 20 Shintani T, Asakura K, Kataura AThe effect of adenotonsillectomy in children with OSA. Int J Pediatr Otorhinolaryngol; 1998;44:51-8, 9720681.

    CrossrefGoogle Scholar
  • 21 Clayton D, Hills MStatistical Models in EpidemiologyOxford: Oxford University Press; 1993.

    Google Scholar
  • 22 Faul F, Erdfelder E, Buchner A, Lang AGStatistical power analyses using G*Power 3.1: tests for correlation and regression analyses. Behav Res Methods; 2009;41:1149-60, 19897823.

    CrossrefGoogle Scholar
  • 23 Nieswiadomy RMFoundations of Nursing Research 4th edNew Jersey: Pearson Education; 2002.

    Google Scholar
  • 24 Lackey N, Wingate AL, Brink PJ, Wood MJThe pilot study: one key to research success. Advanced Design in Nursing Research2nd edCA: Sage; 1998.

    Google Scholar
  • 25 Weinstock TG, Marcus CL, Amin RSet al.Obstructive sleep apnea severity and associated co-morbidities in candidates for adenotonsillectomy: the childhood adenotonsillectomy (CHAT) study. Am J Respir Crit Care Med; 2012;185:A1055.

    Google Scholar
  • 26 Watanabe T, Isono S, Tanaka A, Tanzawa H, Nishino TContribution of body habitus and craniofacial characteristics to segmental closing pressures of the passive pharynx in patients with sleep-disordered breathing. Am J Respir Crit Care Med; 2002;165:260-5, 11790665.

    CrossrefGoogle Scholar
  • 27 Isono S, Tanaka A, Tagaito Y, Ishikawa T, Nishino TInfluences of head positions and bite opening on collapsibility of the passive pharynx. J Appl Physiol; 2004;97:339-46, 15020573.

    CrossrefGoogle Scholar
  • 28 Tsuiki S, Isono S, Ishikawa T, Yamashiro Y, Tatsumi K, Nishino TAnatomical balance of the upper airway and obstructive sleep apnea. Anesthesiology; 2008;108:1009-15, 18497601.

    CrossrefGoogle Scholar
  • 29 Ito E, Tsuiki S, Namba K, Takise Y, Inoue YUpper airway anatomical balance contributes to optimal continuous positive airway pressure for Japanese patients with obstructive sleep apnea syndrome. J Clin Sleep Med; 2014;10:137-42, 24532996.

    LinkGoogle Scholar
  • 30 Tsuiki S, Ito E, Isono Set al.Oropharyngeal crowding and obesity as predictors of oral appliance treatment response to moderate obstructive sleep apnea. Chest; 2013;144:558-63, 23493981.

    CrossrefGoogle Scholar
  • 31 Maeda K, Tsuiki S, Fukuda T, Takise Y, Inoue YIs maxillary dental arch constriction common in Japanese male adult patients with obstructive sleep apnoea?Eur J Orthod; 2014;36:403-8, 23956331.

    CrossrefGoogle Scholar
  • 32 Tsuiki S, Maeda K, Inoue YRapid maxillary expansion for obstructive sleep apnea: a lemon for lemonade?J Clin Sleep Med; 2014;10:233, 24533009.

    LinkGoogle Scholar
  • 33 Franchi L, Pavoni C, Faltin K, McNamara JA, Cozza PLong-term skeletal and dental effects and treatment timing for functional appliances in Class II malocclusion. Angle Orthod; 2013;83:334-40, 22931200.

    CrossrefGoogle Scholar
  • 34 Lorenzoni DC, Bolognese AM, Garib DG, Guedes FR, Sant'anna EFCone-beam computed tomography and radiographs in dentistry: aspects related to radiation dose. Int J Dent; 2012;2012:813768, 22548064.

    CrossrefGoogle Scholar