INTRODUCTION

Skeletal maturity assessment plays a key role in orthodontic diagnosis and treatment planning.¹ Assessment of growth spurt, at least partially, affects the efficiency and effectiveness of orthodontic treatment such as the decision of extraction vs nonextraction orthodontic therapy.¹ Accurate prediction of the growth spurt is critical for modifying the growth of jaw bones.² Determining the end of the growth spurt helps clinicians improve the outcomes of orthognathic surgery and dental implant therapy in growing patients.^3,4

The onset of orthopedic treatment is mainly based on assessing skeletal age.⁵ There has been controversial data regarding the validity of chronological age for skeletal maturation assessment. Although Safavi et al.⁶ reported a positive association between age and growth stages in their studied population, several authors disagreed, reporting that age is not a reliable indicator for growth maturation due to growth spurt variation among different individuals.⁷

Radiographic analyses have been used to assess skeletal maturation.⁸ Lateral cephalometric radiography has been introduced to predict the growth stage by using the method of cervical vertebral maturation staging (CVMS) that is based on assessing the morphological changes of the six cervical vertebrae (C1–C6).⁹ Although the use of the CVM method to predict skeletal age has been questioned by several clinicians, the modified version of CVMS that consists of five stages (CVMS I to V) by evaluating the second, third, and fourth cervical vertebrae is currently preferred and has been adopted by the American Board of Orthodontics.^5,10 Since a lateral cephalometric radiograph is routinely required for orthodontic patients, CVMS has been preferred to avoid extra radiation exposure from hand-wrist radiographs.⁹

Unlike the methods mentioned already that require radiographic exposure, new noninvasive analysis of biological mediators such as growth factors have been introduced to assess skeletal maturity.^11,12 An association between growth spurt and elevation in biomarkers of bone metabolism has been reported.¹¹ For example, alkaline phosphatase (ALP) was previously investigated in the gingival crevicular fluid (GCF) of growing subjects and was shown to increase in relation to the mandibular growth spurt, suggesting that ALP can be used as a clinical biomarker for the identification of pubertal growth.¹¹ Tarvade et al.¹³ reported higher levels of salivary ALP during the growth spurt assessing hand-wrist radiographs. Cabras et al.¹⁴ reported increased levels of proteins in whole saliva during adolescence that could be correlated with hormonal and growth maturation. Further studies have been recommended to investigate the role of salivary biomarkers in assessing growth status. To date, no published studies have investigated CVMS prediction based on salivary ALP activity, protein concentration, and the combination of chronological age with salivary ALP activity.

The aim of this study was to investigate the relationship between salivary ALP activity, protein concentration, and chronological age with skeletal maturation in growing patients as noninvasive biomarkers for skeletal maturation assessment.

MATERIALS AND METHODS

Seventy-nine subjects participated in this cross-sectional clinical study in the Department of Orthodontics, Tufts University School of Dental Medicine (TUSDM), Boston, MA, USA. Patients aged from 7 to 23 years old, who either were beginning orthodontic therapy or were under current orthodontic treatment and for whom lateral cephalographs within the last 6 months existed in their records, were included.¹⁵ Any subject diagnosed with a medical condition, systemic disease, or taking medication that affected growth and/or bone metabolism was excluded. In addition, those with self-reported pregnancy or lactation, non-English speakers and acute intraoral infection were excluded. The study protocol was approved by Tufts Health Sciences Institutional Review Board (#11986). Informed consent was obtained from legal guardians and subjects.

Subjects deemed eligible at a screening visit were scheduled for visit 2 and were instructed to refrain from drinks, food, and tooth-brushing for 90 minutes before saliva collection. If subjects fulfilled the above instructions, the screening visit and visit 2 were combined and unstimulated whole saliva was collected by drooling saliva into a tube for 5 minutes after swallowing. A sample volume of 1 to 5 mL was collected in preweighted tubes and stored immediately on ice. All samples were collected at the same time-period, from 9:00 AM to 12:00 PM to control for circadian changes in salivary flow and then centrifuged and stored at −80°C until analysis.

RESULTS

The distribution of the study population among the five CVMS is presented in Table 1. The median (IQR) age in CVMS I was 10 (2.75) years, in CVMS II 11 (2.0) years, in CVMS III 13 (4.0) years, in CVMS IV 14 (4.5) years, and in CVMS V 17.5 (6.25) years (Table 2).

Table 1 Distribution of Study Sample

View Fullscreen

Table 2 Age Distribution (Years) of the Study Sample

View Fullscreen

According to Landis and Koch's guidelines¹⁷ for interexaminer agreement, the agreement was moderate between rater 1 and rater 3 (0.43) and between rater 2 and rater 3 (0.49), while it was fair between rater 1 and 2 (0.33). Salivary ALP levels were the highest in CVMS I (Table 3). A significant difference in salivary ALP activity distribution among different CVMS was shown using the Kruskal-Wallis test (P = .002). ALP activity was significantly different between CVMS I and CVMS II (P < .001) and between CVMS I and CVMS V (P = .004) (Figure 1) using Dunn's test and Bonferroni correction.

Table 3 Salivary ALP Activity in 5 Different Cervical Stages*

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

Regarding protein concentration, the results showed higher protein values in CVMS III and CVMS V. The median (IQR) protein concentration in CVMS I was 0.79 (0.52) mg/mL, in CVMS II 1.15 (0.64) mg/mL, in CVMS III 1.44 (0.65) mg/mL, in CVMS IV 1.04 (0.60) mg/mL, and in CVMS V 1.50 (0.46) mg/mL. The protein concentration distribution was significantly different between stages using the Kruskal-Wallis test (P = .014). The protein concentration was significantly different between CVMS I and CVMS III (P = .005).

The nonparametric Mann-Whitney U test showed a significant difference in ALP activity between males and females (P = .007) (Figure 2). The median (IQR) ALP for females was 0.48 mU/mg (0.29 mU/mg), whereas the median (IQR) ALP for males was 0.70 mU/mg (0.54 mU/mg).

View Figure

• Download as Powerpoint
• Download Figure

A strong positive association between age and CVMS was found using the Spearman correlation (r_s = 0.763, P < .001) (Figure 3). Five candidate multinomial logistic regression models to predict CVMS, each using one or more independent variables, were compared (Table 4). In the first model (in which CVMS was predicted from ALP), the latter variable was a statistically significant predictor of the former (P = .002). In the second model, protein concentration was statistically significant (P = .021) in predicting CVMS. In the third model, age was significant (P < .001) in predicting CVMS. In the fourth model, both ALP activity (P = .002) and age (P < .001) were significant in predicting CVMS. In the fifth model, protein concentration was not statistically significant in predicting CVMS (P = .072), whereas age was significant (P < .001). McFadden's pseudo R² was used to assess the predictive strength of each model. According to Mokhtarian, a pseudo R² of 0.3 is indicative of a model with good fit.¹⁸ The ability of predicting CVMS correctly (model's overall correct classification rate) was also measured in all models (Table 4).

View Figure

• Download as Powerpoint
• Download Figure

Table 4 Multinomial Logistic Regression Models with McFadden's Pseudo R² and Correct Classification Rate By Model

View Fullscreen

DISCUSSION

New noninvasive biomarkers to assess skeletal maturation have been proposed since most craniofacial assessment methods are invasive, and require radiographic exposure and a long observation period. Saliva has been described as a “mirror of the body” and can be used as a diagnostic tool for the assessment of skeletal age.¹⁹ In the present study, salivary ALP activity, protein concentration, and chronological age were investigated as biomarkers for skeletal maturation assessment since they have been associated with bone metabolism and general body growth.^20,21 The current results suggested that salivary ALP activity may have a promising diagnostic value in skeletal maturity assessment.

Instructions were provided to study subjects before saliva collection to standardize sample collection. The results showed that salivary ALP activity peak was during the prepubertal period; it was higher at CVMS I and then declined with a significant difference between CVMS I and CVMS II and between CVMS I and CVMS V. In contrast to these findings, Perinetti et al.¹¹ reported that ALP activity peak in GCF was during the pubertal growth spurt and Tarvade et al.¹³ also found higher levels of salivary ALP during the growth spurt. Additionally, Tobiume et al.²² stated increased serum ALP activity during puberty. However, it is important to mention that, in the previously referenced studies,^11,13,22 ALP activity was not normalized to total protein concentration in saliva (ALP activity divided by protein concentration of each sample) as was done for the current samples. This methodology was considered to be critical in this study for fluid analysis to standardize the samples' analysis and allow equal comparisons between different patients.

Studies have reported that normal protein concentrations in saliva ranged between 0.72 and 2.45 mg/mL.²³ The current data showed a slightly wider range (0.22–2.53 mg/mL) and a range of 0.07 to 2.91 mU/mg of ALP activity in saliva. On the other hand, Perinetti et al.¹¹ reported that ALP activity in GCF ranged between 1.35 and 121.60 mU/mg. The different range in ALP activity between GCF and saliva could be because ALP activity was measured in different fluids and volumes but also that ALP activity in GCF was not normalized to total protein concentration¹¹ as was followed in the salivary measures. This omission could act as a confounding factor in data analysis due to different levels of salivary proteins in their samples.

Regarding gender differences in salivary ALP activity, the current findings showed a significant difference between males and females. The median in males' ALP activity (0.70 mU/mg) was higher compared to females' ALP activity (0.48 mU/mg) (Figure 2). Males have more growth potential and a longer growth spurt duration compared to females, and that could explain the increased enzymatic activity in males.²⁴ Similar to the current findings, Fleisher et al. reported that males' serum ALP activity in adolescents was higher than for females.²⁰

Total, but not normalized, GCF ALP activity was shown sensitive to the growth phase while total protein concentration in GCF was not a reliable indicator of different growth phases.²⁵ To date, no published studies have investigated prediction of skeletal maturation from salivary proteins. The current study results showed that skeletal growth prediction from salivary ALP activity was statistically significant (P = .002). Additionally, the ability for predicting CVMS correctly (model's overall correct classification rate) from ALP activity was 35.4%. ALP activity's ability for predicting CVMS was higher than the protein concentration (32.9%), but lower than age (51.9%) and the combination of ALP activity and age (53.2%) (Table 4). Salivary protein concentration was significant in predicting CVMS (P = .021) (Table 4). Thus, the data suggested that the ability of ALP activity to predict CVMS was modest in isolation. The three models including age (Models 3–5) exhibited the highest McFadden's pseudo R² statistics and correct classification rates, with the combination of ALP activity and age (Model 4) exhibiting the highest such values. In the fourth model, if adjusted for age, ALP activity was a statistically significant predictor of CVMS. In addition, salivary protein concentration was less sensitive as an indicator for maturation assessment compared to salivary ALP activity and age. This study provided some preliminary data about the value of salivary ALP activity in predicting skeletal maturity and more studies are needed on this topic.

Chronologic age has been used to evaluate skeletal maturation with limited reliability.⁷ Although Ramos et al.⁷ found a weak correlation between skeletal age and chronological age, the current data showed a significant positive correlation between chronological age and skeletal growth (Figure 3). These findings agreed with Litsas and Lucchese²¹ who also reported that skeletal growth was positively associated with chronological age. To date, no published studies have investigated CVMS prediction from chronological age. The data showed that chronological age was statistically significant in predicting CVMS (P < .001) (Table 4). Therefore, the combination of age and ALP may provide the best CVMS prediction compared to other models, suggesting that the use of new tools (biomarkers) in combination with traditional techniques (chronological age) can enhance assessment of skeletal maturation.

The data from this study can potentially help clinicians using noninvasive methods to identify the prepubertal stage in growing patients, a critical phase in determining the optimal timing of orthodontic treatment. The data suggested that the combination of high levels of salivary ALP activity and age could allow clinicians to make better predictions during skeletal assessment.

One of the limitations of this study was the interexaminer reliability for using the CVM method. As this method is subjective and poorly reproducible, we selected three experienced orthodontists as examiners to improve reliability of CVMS. Even with this approach, interexaminer reliability was low. Based on these findings, the use of CVM method to predict skeletal age was questionable in this study. Due to the presence of existing lateral cephalometric radiographs and ethical concerns with repeating them, saliva collection was not achieved exactly at the same time point of lateral cephalometric radiographs but within 6 months, following a similar approach as in a previous study.²¹ Although this study included a large number of subjects, the majority were Caucasian whites. Thus, the data sets a landmark for future studies with larger and more ethnically diverse sample populations to investigate the potential use of salivary ALP activity and other salivary proteins as biomarkers for skeletal maturity.

CONCLUSIONS

• Salivary ALP activity may be a promising diagnostic tool for prepubertal growth phase prediction.

• A strong positive association was shown between chronological age and CVMS.

• The combination of age and salivary ALP activity may provide the best CVMS prediction compared to other models.