This article has Open Peer Review reports available.
Gait and lower limb muscle strength in women after triple innominate osteotomy
© Kolk et al.; licensee BioMed Central. 2015
Received: 25 July 2014
Accepted: 10 March 2015
Published: 24 March 2015
In adult patients with developmental hip dysplasia, a surgical procedure (triple innominate osteotomy) of the pelvic bone can be performed to rotate the acetabulum in the frontal plane, establishing better acetabular coverage. Although common clinical hip scores demonstrate significant improvements after surgery, they provide only overall information about function. The purpose of this study was to quantify the long-term outcome of triple innominate osteotomy in more detail using gait analyses and muscle strength measurements.
We performed gait analyses at self-selected walking speed as well as isometric hip and knee muscle strength tests in twelve women who had undergone a unilateral triple innominate osteotomy (age: 34 ± 12 y, time post surgery: 80 ± 18 m). We compared the results to reference values obtained from eight healthy peers (age: 33 ± 10 y).
The patients exhibited slight asymmetries in step length (smaller steps) and stance time (longer stance) as well as lower hip abduction moments in the operated limb in early stance compared to the non-operated limb. However, there were no differences in gait compared to healthy controls, even though the patients showed reduced bilateral hip abduction strength compared to controls.
Our results indicate that the patients’ gait pattern had generally recovered very well, despite slight asymmetries in spatiotemporal parameters. Subtle deviations in hip abduction moments were observed during gait, whereas hip abduction strength was substantially reduced. Hence, the patients walked at a higher percentage of their maximal capacity. They may, therefore, be prone to fatigue and adopt compensatory gait strategies more quickly than healthy peers when walking long distances.
Developmental dysplasia of the hip is characterized by a shallow, obliquely-oriented acetabulum and malposition of the proximal femur. Together, these anatomical deformations lead to an increased local load because of a decreased weight-bearing area . This is painful for patients, can cause secondary osteoarthritis, and limits patients in their daily activities. To relieve the symptoms, a surgical procedure (Triple Innominate Osteotomy; TIO) can be performed [2,3]. In this surgery, the acetabulum is rotated in the frontal plane to increase its coverage over the femoral head . This procedure reduces pain and further joint degeneration and, thereby, greatly improves the quality of life of patients . In addition to pain relief and radiographic improvement, functional outcome is also important in determining the success of hip dysplasia surgery.
Most patients who underwent TIO achieve good to excellent functional results based on clinical functional scores [4,6,7]. However, drawbacks of these scoring systems are that they are subjective , may have ceiling effects , and only provide information about the overall level of function. They do not allow one to assess which joints or muscle groups underlie functional abnormalities. Thus, in order to obtain a quantitative and objective evaluation of function after TIO, it is necessary to employ more detailed and quantitative measurement techniques such as gait analysis [10-13]. In addition to gait analysis, muscle strength measurements are valuable to assess the biomechanical properties of the operated limb .
To our knowledge, there have been no reports on the impact of TIO at such a detailed functional level. The information obtained from gait analysis could provide a better assessment of the long-term effects of this surgery. It can be used to inform the, usually young-adult, patients about their expected walking ability after the operation . It also allows comparing the functional outcome of TIO surgery to that of other treatment options such as Ganz’s osteotomy , for which gait analysis studies have been performed [17-19]. Several decrements (e.g. in walking speed and hip flexion pull-off power) after Ganz’s osteotomy have been found [17,18]. Since the surgical approach is slightly different between TIO (usually a modified Smith-Peterson approach) and Ganz’s osteotomy (usually either a classic Smith-Peterson or an abductor-sparing direct anterior approach), abnormalities in gait parameters after these respective operations may be related to the approach. Furthermore, the identification of specific gait and strength deficits might be useful for improving surgical techniques or postoperative rehabilitation protocols that may further enhance the functional outcome.
This study aimed to investigate the long-term outcome of TIO in terms of biomechanics of gait and muscle strength. Specifically, our objectives were to determine deviations in (1) spatiotemporal parameters, hip joint angles and joint moments during gait, and (2) hip and knee muscle strength in patients who had undergone TIO between two and ten years ago. We compared the results between the operated and non-operated limb and between patients and healthy controls.
Twelve women who had undergone TIO and eight healthy women in the same age range participated in this study. The indication for TIO was symptomatic hip dysplasia with a spherical femoral head, and without major damage to the cartilage . All patients were operated in the Sint Maartenskliniek, Nijmegen, The Netherlands by the same experienced surgeon. The surgical procedure (modified Tönnis osteotomy) has been described in detail . Patients between 18 and 70 years who had undergone a unilateral TIO between January 2003 and June 2010 were eligible for inclusion. This time window ensured that rehabilitation had been completed. Rehabilitation consisted of regular mobilization of the hip joint and strengthening exercises for the muscles of the hip region, and lasted for 3–12 months, depending on the patient’s individual speed of recovery. In total, 54 patients were screened, of whom 34 were eligible for inclusion. Of those, 18 patients were willing to participate, and 12 were included after screening for exclusion criteria. We excluded patients that had any disease or other condition that could affect their gait, including severe hip pain, as well as patients with a Body Mass Index (BMI) > 30 kg/m2 because of difficulties in marker placement for gait analysis. Selection was unspecific regarding gender, but all 12 included patients were women.
The study procedure was approved by the local ethical committee of the region Arnhem-Nijmegen, The Netherlands (study code 2012/065). A written informed consent was obtained from each subject.
The participants were invited to the gait laboratory of the Radboud university medical center, Nijmegen, The Netherlands for a combined gait and muscle strength assessment session. We also obtained the Oxford [21,22] (range 0–48) and Harris Hip Score  (range 0–100) from the patients during a break in the session.
The participants walked barefoot on an 8 m long walkway at self-selected comfortable walking speed. An integrated data collection was performed including three-dimensional motion capture with synchronized force plate recordings. A six-camera digital optical motion capture system (Vicon MX, Oxford, UK) was used to record the position of 35 retro-reflective markers placed on the lower limb and torso (100 Hz). The standard Vicon Plug-in-Gait marker set was used, with additional markers placed on the anterior side of the thigh and lower leg at 1/3 and 2/3 segment length, and on the fifth metatarsal head of the foot. Two custom-made force plates (AMTI, Watertown, MA, USA), embedded level in the laboratory floor measured ground reaction forces (1000 Hz) during the stance phase of the gait cycle.
No specific instructions were given other than ‘walk naturally’ to prevent participants from targeting the force plates. Trials were repeated until six successful trials had been recorded, where ‘successful’ was defined as a trial in which each foot cleanly struck one of the two force plates. The gait analysis data of one patient and one control subject had to be discarded due to technical problems, leaving a group of 11 patients and 7 controls available for the gait analysis part of the study.
Heel strike and toe off events were identified using thresholding of ground reaction force data (heel contact when F > 20 N, toe off when F < 20 N). Spatiotemporal parameters were subsequently calculated based on a combination of these events and the heel and toe marker position data.
A 21 degrees of freedom kinematic model (‘GaitLowerExtremityModel’, as available in the AnyBody Managed Model Repository 1.5.1) consisting of trunk, pelvis, thigh, shank, talus and foot segments was scaled to each subject based on the marker trajectories using the AnyBody Modeling System (version 5.3.1, AnyBody Technology A/S, Aalborg, Denmark). The marker trajectories were initially filtered with a 5 Hz 2nd order Butterworth low-pass filter. Force plate data were low-pass filtered at 12 Hz with a 2nd order Butterworth filter. The model was based on the kinematic part of the Twente Lower Extremity Model dataset . After this procedure, lower limb joint angles were obtained by solving the inverse kinematics using the optimized parameters. At each joint, ideal torque generators were added. The segment masses were scaled using common scaling laws , available in AnyBody. The model marker positions, segment lengths and knee joint axes were then optimized using a parameter optimization algorithm . Finally, the kinematics and ground reaction forces were used as input for an inverse dynamic analysis , in which joint moments were calculated in the local (ISB ) reference frames.
Our outcome variables of interest were spatiotemporal parameters, hip joint angles and moments in the sagittal and frontal planes. Each of these variables was averaged across trials to obtain subject ‘ensemble’ averages both for the operated and the non-operated limbs. The moments were normalized to the body weight (BW) and height (Ht) of the subject (%BW*Ht). In the sagittal plane, we determined the peak flexion and extension angles and moments. In the frontal plane, we used the adduction angle at 20% of the gait cycle in the analysis, since there was a peak in the angle at that particular phase of the gait cycle. Owing to the M-shaped curve of hip abduction moments during gait, we extracted two peak values, one in the first and one in the second half of the stance phase. For the healthy controls, we used the mean of the left and right limbs in the analyses.
For maximum knee flexion strength testing, subjects sat on the edge of a testing bench positioned close to a wall (Figure 1B). The knee and hip were flexed at 90°, and the end piece of the force transducer was applied just proximal to the malleoli. The other end was rigidly fixed to the wall, level with the end piece. During the contractions, manual pressure was applied by an assistant proximal to the top of the tested knee to prevent it from rising. Knee extension was tested in a custom-built chair with vertical backrest, with the knee and hip flexed at 90°, and with the end piece of the force transducer applied just proximal to the malleoli (Figure 1C). The subject was strapped to the chair to prevent the waist from rising.
Participants performed three maximum isometric contractions for four seconds each, separated by 30 seconds of rest. All subjects received the same verbal encouragements during the contractions to achieve maximum effort. The highest recorded maximal force was multiplied by the moment arm to the joint centre of the tested joint to obtain peak torque values. Raw data was low-pass filtered at 6 Hz with a 2nd order Butterworth filter and the peak torque was normalized by body weight.
Student’s t-tests were used for comparing group characteristics, spatiotemporal parameters, kinematics and muscle strength tests between the patient and control groups. Paired samples t-tests were used for comparisons between operated and non-operated limbs. Hip joint moments during gait were tested with repeated-measures ANOVAs; one with phase of gait cycle (early and late stance) and limb (operated and non-operated) as within-subjects factors, and another with phase of gait cycle (early and late stance) as within-subjects factor and group (operated limb of patients and controls) as between-subjects factor. Post-hoc paired samples t-tests were used if the ANOVA indicated significant differences between the operated and the non-operated limb. The significance level was set at P ≤ 0.05. For the muscle strength data, significance was set at P ≤ 0.01 because of the number of t-tests performed (12). IBM SPSS Statistics 18.104.22.168 was used for all statistical analyses.
Characteristics of patients after triple innominate osteotomy and controls
TIO patients n = 12
Controls n = 8
Time since surgery (months)
L: 3, R: 9
Harris hip score
Oxford hip score
Spatiotemporal parameters of patients after triple innominate osteotomy and controls
Walking speed (m/s)
Step length operated limb (m)
Step length non-operated limb (m)
Stride length (m)
Stance duration operated limb (% of gait cycle)
Stance duration non-operated limb (% of gait cycle)
Kinematic gait parameters
Kinetic gait parameters
Hip abduction moments during stance are shown in Figure 2D. The operated limb had a different abduction moment pattern than the non-operated limb during the stance phase, as demonstrated by a significant phase x limb interaction effect (F1,10 = 5.78, P = 0.037); in early stance, the abduction moment in the operated limb was non-significantly lower than in the non-operated limb (P = 0.129), whereas in late stance it was non-significantly higher (P = 0.192). The main effect of limb was not significant (F1,10 = 0.12, P = 0.742). The operated limb was not significantly different from controls (group: F1,16 = 0.00, P = 0.949; phase x group: F1,16 = 2.63, P = 0.125).
Sagittal plane hip moments during stance are shown in Figure 2B. The hip flexion and extension moments generated by the operated limb of the patients were not different from the unaffected limb (limb: F1,10 = 0.03, P = 0.861, phase x limb: F1,10 = 1.51, P = 0.247). The operated limb did not differ from controls (group: F1,16 = 2.62, P = 0.125, phase x group: F1,16 = 0.28, P = 0.601).
In this study, we found that patients who had undergone TIO (on average 80 months ago) did not have major abnormalities in their gait pattern compared to controls, but did exhibit subtle asymmetries between the operated and non-operated limbs. The operated limb had a smaller step length and longer stance time compared to the non-operated limb. In addition, the abduction moment pattern during the stance phase was significantly different between the operated and non-operated limbs. In terms of muscle strength, a bilateral abduction strength deficit was found. Muscle strength did not differ between the operated and the non-operated limb for any of the tested joints.
Interestingly, we found subtle asymmetries in hip abduction moments between the operated and non-operated limbs. This may represent a more cautious walking pattern, by enabling a more gradual transfer of the body’s centre of mass over the operated limb, thereby reducing the peak hip abduction moment in early stance. This does not seem to be related to a lack of hip abductor strength, as there was no difference in strength between the operated and non-operated leg. The reduced step length and prolonged stance duration that we found in the operated limb, albeit rather small, may also point to an integral strategy in the frontal plane reflecting a more cautious walking pattern.
Compared to healthy controls, we found significantly reduced hip abduction strength in patients who had undergone TIO. Hip abductor weakness is a common finding in patients with hip dysplasia, frequently resulting in Trendelenburg’s sign during gait [2,29]. Hence, the persisting weakness that we found may be reflective of severe muscle atrophy that had developed over many years prior to surgery and that did not fully recover despite postoperative rehabilitation. Alternatively, our patient group may have been less physically active than the controls, resulting in poorer muscle strength. We indeed observed a general trend of muscle strength reduction in all the tests performed, but the pronounced weakness in hip abduction argues against an overall reduction in physical activity being the only responsible factor.
Our finding that hip abduction strength was substantially lower in patients compared to controls, whilst no main effect of group could be demonstrated for hip abduction moments during gait implies that the patients walked at a higher percentage of their maximum capacity. This may eventually result in gait pattern changes when patients become fatigued, as it has been shown that human gait is relatively sensitive to weakness in the hip abductor muscle group . A gait pattern typical of abductor weakness (Trendelenburg’s sign) may then emerge, involving pelvic drop at the contralateral side. This increases hip joint stress in the operated hip due to a decrease in joint contact area , which may speed up the development of osteoarthritis in the operated hip.
The clinical results of TIO at comparable follow-up times are generally ‘good’, ‘excellent’, or ‘improved compared with before the operation’ [1,4,32], which is congruent with the results of our study. Hence, the functional status of our study population appears to be representative of the TIO group at large. To our knowledge, up to this point no studies have been published that employed gait analysis or maximum strength measurements to describe the functional outcome after TIO surgery. In contrast, results from gait analyses have been reported for patients after Ganz’s periacetabular osteotomy , which conceptually is a rather similar procedure to the modified Tönnis osteotomy used here . Our findings that patients after TIO did not demonstrate major gait deviations compares well to previous results from patients after Ganz’s osteotomy. After that surgery, no statistically significant differences from controls could be demonstrated in sagittal hip joint kinematics and kinetics [19,33] or frontal plane kinetics . Similarly, Karam et al. investigated spatiotemporal gait parameters before and one year after Ganz’s osteotomy  and found post-operative walking velocities, stride lengths and cadences very similar to our results. Thus, the relatively small differences in the surgical approach between TIO and Ganz’s osteotomy (even if it was abductor-sparing, as in the case of Sucato et al. ) do not appear to play a major role in long-term functional outcome.
Our study has certain limitations. First, we invited 34 patients to participate in the study, but only 18 responded, out of which 12 patients could be included. The non-responders and excluded patients might have had more pain or functional limitations than those included. Moreover, since there were only women in our study population, the results might not apply to men who underwent TIO. Second, although the study groups were similar with regard to age, gender, height, weight and BMI, we did not take confounding factors into account such as general activity level or participation in sports. As such, it is possible that the patients were less active than the controls and that, therefore, they had less muscle strength around the hip and knee joints. Third, we did not perform gait and strength measurements before the surgery. Such analyses could reveal whether the deviations in gait observed post-operatively in this study stem from pre-operatively adopted gait patterns. Fourth, we did not adjust the geometry of the operated hip joint in the AnyBody model. The joint’s centre of rotation in our patients might have been slightly altered by the TIO, thereby influencing the joint moment calculations.
This study showed that patients who underwent TIO generally recovered well from their operation with regard to gait, but did exhibit subtle asymmetries between the operated and non-operated limbs several years after surgery. Muscle strength deficits were found bilaterally in hip abduction. Although the differences observed in gait were subtle, they may become clinically important in young or active patients who engage in high-level activities over many years. Thus, in further research in this patient group it is recommended to focus on gait adaptations after prolonged walking.
We gratefully acknowledge financial support by the European Commission FP7 Programme for the TLEMsafe project (http://www.tlemsafe.eu). The funding agency played no role in design, collection, analysis, interpretation of data, writing process, or in the decision to submit the paper for publication. We thank drs. Arno M. ten Ham for useful comments on the manuscript draft.
- Hsin J, Saluja R, Eilert RE, Wiedel JD. Evaluation of the biomechanics of the hip following a triple osteotomy of the innominate bone. J Bone Joint Surg Am. 1996;78(6):855–62.PubMedGoogle Scholar
- Steel HH. Triple osteotomy of the innominate bone. J Bone Joint Surg Am. 1973;55(2):343–50.PubMedGoogle Scholar
- Tonnis D, Behrens K, Tscharani F. [A new technique of triple osteotomy for turning dysplastic acetabula in adolescents and adults (author’s transl)]. Z Orthop Ihre Grenzgeb. 1981;119(3):253–65.View ArticlePubMedGoogle Scholar
- de Kleuver M, Kooijman MA, Pavlov PW, Veth RP. Triple osteotomy of the pelvis for acetabular dysplasia: results at 8 to 15 years. J Bone Joint Surg Br. 1997;79(2):225–9.View ArticlePubMedGoogle Scholar
- Faciszewski T, Coleman SS, Biddulph G. Triple innominate osteotomy for acetabular dysplasia. J Pediatr Orthop. 1993;13(4):426–30.View ArticlePubMedGoogle Scholar
- Guille JT, Forlin E, Kumar SJ, MacEwen GD. Triple osteotomy of the innominate bone in treatment of developmental dysplasia of the hip. J Pediatr Orthop. 1992;12(6):718–21.View ArticlePubMedGoogle Scholar
- Nakamura S, Ninomiya S, Takatori Y, Morimoto S, Umeyama T. Long-term outcome of rotational acetabular osteotomy: 145 hips followed for 10–23 years. Acta Orthop Scand. 1998;69(3):259–65.View ArticlePubMedGoogle Scholar
- Learmonth ID, Young C, Rorabeck C. The operation of the century: total hip replacement. Lancet. 2007;370(9597):1508–19.View ArticlePubMedGoogle Scholar
- Wamper KE, Sierevelt IN, Poolman RW, Bhandari M, Haverkamp D. The Harris hip score: Do ceiling effects limit its usefulness in orthopedics? Acta Orthop. 2010;81(6):703–7.View ArticlePubMedPubMed CentralGoogle Scholar
- Rosler J, Perka C. The effect of anatomical positional relationships on kinetic parameters after total hip replacement. Int Orthop. 2000;24(1):23–7.View ArticlePubMedPubMed CentralGoogle Scholar
- Lindemann U, Becker C, Unnewehr I, Muche R, Aminin K, Dejnabadi H, et al. Gait analysis and WOMAC are complementary in assessing functional outcome in total hip replacement. Clin Rehabil. 2006;20(5):413–20.View ArticlePubMedGoogle Scholar
- Nantel J, Termoz N, Vendittoli PA, Lavigne M, Prince F. Gait patterns after total hip arthroplasty and surface replacement arthroplasty. Arch Phys Med Rehabil. 2009;90(3):463–9.View ArticlePubMedGoogle Scholar
- Kolk S, Minten MJ, van Bon GE, Rijnen WH, Geurts AC, Verdonschot N, et al. Gait and gait-related activities of daily living after total hip arthroplasty: a systematic review. Clin Biomech (Bristol, Avon). 2014;29(6):705–18.View ArticleGoogle Scholar
- Lin DH, Jan MH, Liu TK, Lin YF, Hou SM. Effects of anterolateral minimally invasive surgery in total hip arthroplasty on hip muscle strength, walking speed, and functional score. J Arthrop. 2007;22(8):1187–92.View ArticleGoogle Scholar
- van Hellemondt GG, Sonneveld H, Schreuder MH, Kooijman MA, de Kleuver M. Triple osteotomy of the pelvis for acetabular dysplasia: results at a mean follow-up of 15 years. J Bone Joint Surg Br. 2005;87(7):911–5.View ArticlePubMedGoogle Scholar
- Ganz R, Klaue K, Vinh TS, Mast JW. A new periacetabular osteotomy for the treatment of hip dysplasias. Technique and preliminary results. Clin Orthop Relat Res. 1988;232:26–36.PubMedGoogle Scholar
- Karam MD, Yubo G, McKinley T. Assessment of walking pattern pre and post peri-acetabular osteotomy. Iowa Orthop J. 2011;31:83–9.PubMedPubMed CentralGoogle Scholar
- Sucato DJ, Tulchin K, Shrader MW, DeLaRocha A, Gist T, Sheu G. Gait, Hip Strength and Functional Outcomes After a Ganz Periacetabular Osteotomy for Adolescent Hip Dysplasia. J Pediatr Orthop. 2010;30(4):344–50.View ArticlePubMedGoogle Scholar
- Pedersen ENG, Alkjaer T, Soballe K, Simonsen EB. Walking pattern in 9 women with hip dysplasia 18 months after periacetabular osteotomy. Acta Orthop. 2006;77(2):203–8.View ArticlePubMedGoogle Scholar
- Kooijman MA, Pavlov PW. Triple osteotomy of the pelvis. A review of 51 cases. Clin Orthop Relat Res. 1990;255:133–7.PubMedGoogle Scholar
- Dawson J, Fitzpatrick R, Carr A, Murray D. Questionnaire on the perceptions of patients about total hip replacement. J Bone Joint Surg Br. 1996;78(2):185–90.PubMedGoogle Scholar
- Murray DW, Fitzpatrick R, Rogers K, Pandit H, Beard DJ, Carr AJ, et al. The use of the Oxford hip and knee scores. J Bone Joint Surg Br. 2007;89(8):1010–4.View ArticlePubMedGoogle Scholar
- Harris WH. Traumatic arthritis of the hip after dislocation and acetabular fractures: treatment by mold arthroplasty. An end-result study using a new method of result evaluation. J Bone Joint Surg Am. 1969;51(4):737–55.PubMedGoogle Scholar
- Klein Horsman MD, Koopman HF, van der Helm FC, Prose LP, Veeger HE. Morphological muscle and joint parameters for musculoskeletal modelling of the lower extremity. Clin Biomech (Bristol, Avon). 2007;22(2):239–47.View ArticleGoogle Scholar
- Winter DA. The biomechanics and motor control of human movement: Normal, elderly and pathological. Canada: University of Waterloo Press; 1991.Google Scholar
- Andersen MS, Damsgaard M, MacWilliams B, Rasmussen J. A computationally efficient optimisation-based method for parameter identification of kinematically determinate and over-determinate biomechanical systems. Comp Meth Biomech Biomed Eng. 2010;13(2):171–83.View ArticleGoogle Scholar
- Damsgaard M, Rasmussen J, Christensen ST, Surma E, de Zee M. Analysis of musculoskeletal systems in the AnyBody Modeling System. Simul Mod Pract Theory. 2006;14(8):1100–11.View ArticleGoogle Scholar
- Wu G, Siegler S, Allard P, Kirtley C, Leardini A, Rosenbaum D, et al. ISB recommendation on definitions of joint coordinate system of various joints for the reporting of human joint motion–part I: ankle, hip, and spine. International Society of Biomechanics. J Biomech. 2002;35(4):543–8.View ArticlePubMedGoogle Scholar
- Romano CL, Frigo C, Randelli G, Pedotti A. Analysis of the gait of adults who had residua of congenital dysplasia of the hip. J Bone Joint Surg Am. 1996;78(10):1468–79.PubMedGoogle Scholar
- van der Krogt MM, Delp SL, Schwartz MH. How robust is human gait to muscle weakness? Gait Posture. 2012;36(1):113–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Westhoff B, Petermann A, Hirsch MA, Willers R, Krauspe R. Computerized gait analysis in Legg Calve Perthes disease–analysis of the frontal plane. Gait Posture. 2006;24(2):196–202.View ArticlePubMedGoogle Scholar
- Janssen D, Kalchschmidt K, Katthagen BD. Triple pelvic osteotomy as treatment for osteoarthritis secondary to developmental dysplasia of the hip. Int Orthop. 2009;33(6):1555–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Pedersen EN, Simonsen EB, Alkjaer T, Soballe K. Walking pattern in adults with congenital hip dysplasia: 14 women examined by inverse dynamics. Acta Orthop Scand. 2004;75(1):2–9.View ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.