A retrospective study of orthodontic treatment with pre-treatment gingival recessions

Background: The development of gingival recessions (GR) after orthodontic treatment (OT) is well described in the literature. However, there is a lack of information about the changes in pre-existing GR during OT. Methods: This retrospective study aimed to examine the change of GR before and after OT. Intraoral photographs of 993 patients treated with xed appliances between 2005 and 2017 were evaluated. Patients who had at least one GR on the vestibular surface of maxillary/mandibular teeth mesial to rst molars were included. GR was measured on good quality plaster models before and after OT. The change of GR (T0-T1) was categorised into the three groups: worsened, did not change and improved (this group consisted of GR, which became better but did not disappear, and fully healed, which disappeared after OT)). The change of GR by ≥ 0.5 mm was recorded. Statistical analysis was performed using SPSS Statistics (Version 22.0. Armonk, NY, USA). Results: Fifty-one (5.1%) patient had ≥ 1 GR before OT. Of them, 37 (72.5%) patients were included in the nal analysis according to the inclusion criteria. GR before OT was found on 114 buccal/labial surfaces. At tooth level the median GR improvement was 0.55 (Q1; Q3: 0.12; 0.96) mm (p<0.001). GR improved in 67 teeth (58.7%), did not change in 41 (36.0%), and worsened in 6 (5.3%). Of the 67 GR, which improved, full healing was observed in 15 (22.4%) teeth. There was a lower chance for GR improvement in cases with pre-treatment open bite (OR 3.35; 95% CI: 1.14, 9.84; p=0.03) and dental Class III patients (OR 2.20; 95% CI: 0.92, 5.28; p=0.03). Conclusions: Based on the ndings of the present retrospective study, it may be concluded that orthodontic treatment induces changes in gingival recessions. In more than half of the teeth, the improvement in gingival

The overall prevalence of GR affecting at least one tooth ranges from 11% to 90% in the adult population, being more frequent in adults aged 50 years and above [8][9][10]. The severity of GR has been observed to increase with age [8,11]. GR is more prevalent on vestibular surfaces and in mandibular teeth [9]. It is usually observed on one or several teeth; however, GR can be more widespread when the gingiva of multiple teeth is affected [12].
Although the pathogenesis of gingival recession remains unclear, there are several predisposing and precipitating factors, e.g. coronally attached frenulum and muscles, abnormal tooth position, overhanging restorations, proclination of teeth, fenestration, dehiscence, thin mandibular alveolar bone or mechanical trauma [13]. Thin periodontal/gingival biotype, absence of attached gingiva and reduced thickness of the alveolar bone due to abnormal tooth position are considered as the main risk factors for the development of GR [2,13]. As labial bone and crestal soft tissue thickness are highly associated, they may also in uence the outcome of GR during different kinds of orthodontic movements [14,15].
Several clinical characteristics have been described for de ning gingival biotype: crown width and length ratio, the width of keratinised tissue, papilla height and gingival thickness, which may be assessed by transgingival probing, ultrasonic measurement or probe visibility test [16]. However, method of visual inspection of the probe transparency through the gingiva is mostly recommended [17]. Studies have shown that subjects with a thin biotype are more prone to gingival marginal alterations, irrespective of the type of orthodontic movement. Thin periodontal biotype and proclination of teeth have been shown to correlate with loss of keratinised tissue [15]. It has also been found that the risk for the development of GR grows by 9.7% with each year after orthodontic treatment (OT) [18]. Canines, rst premolars and rst molars are most prone to GR after OT in the maxilla, while the highest risk of GR in the mandible was observed for the central incisors and rst premolars [18].
The management of GR involves the elimination of aetiological and predisposing factors and, if needed, surgical root coverage [3,13]. As untreated malocclusions have also been found to in uence gingival problems (including GR), orthodontic therapy may favour tooth positions and be advantageous in the interdisciplinary treatment plan of GR management [19]. It has been described in the literature that moving a facially positioned tooth or its root in the lingual direction to a more centred position within the alveolar process, the apico-coronal dimension of gingival tissue will increase in width [2,20].
In the face of an increasing number of adult patients in orthodontic practice and the scarcity of the literature discussing recession changes, the investigation of the possible advantages of OT in the interdisciplinary management of pre-existing GR becomes essential. The literature up to date mainly focuses on the development of new GR after OT and mainly in patients before adulthood.
The present retrospective study aimed to examine the change in pre-existing GRs during orthodontic treatment.
Null hypothesis: No change in GRs may be observed during orthodontic tooth movement.

Selection
The present study was a retrospective clinical study. Ethical approval for the study was obtained (No. BEC-OF-67).
Intraoral photographs of 993 patients treated between January 2005 and November 2017 were analysed.
Flowchart of included patients is presented in Figure 1. Patients were treated by two orthodontists (authors EZ and DS) in the Department of Orthodontics, Faculty of Odontology at Lithuanian University of Health Sciences (LUHS) and the private practice in Kaunas, Lithuania. The presence of GR was identi ed on frontal, right and left buccal intraoral photographs (in occlusion) by one investigator, who was blinded (GA). The diagnosis was con rmed in patients' records and then measured on the plaster models ( Figure 2). The presence of gingival recession was con rmed if the marginal border of the gingiva was apical to the cement-enamel junction.
Patients with ≥1 GR before OT were selected and included in the study. Inclusion criteria were: 1) ≥1 GR before OT on labial surfaces of maxillary/mandibular teeth mesial to rst molars; 2) dual arch OT with xed orthodontic appliances; and 3) good quality plaster models available before and after OT. Exclusion criteria were: 1) periodontal disease; OT without xed orthodontic appliances; 3) systemic disease or medication that could in uence the treatment outcome; 4) combined orthodontic-orthognathic treatment; and 5) surgical treatment of GR during OT. Only patients with the records of good oral hygiene without bleeding on probing (BoP) were included. Professional oral hygiene was performed every 3-6 months during OT for all patients.
Tooth extraction (no extraction, extractions in the maxillary or mandibular dental arch, or extractions in both dental arches).

Plaster models
All measurements were performed on plaster models before (T0) and after (T1) OT by the blinded calibrated observer (GA).
The extent of gingival recession was measured in millimetres, using a digital calliper (Fino, Germany) before (T0) and after (T1) OT. Measurements were recorded to the nearest 0.01 mm ( Figure 2). The change of GR (T0-T1) was categorised into three groups (worsened, did not change, improved (this group consists of healed GR, which became better but did not disappear, and fully healed-which disappeared after OT).
Recession change of ≥0.5 mm was recorded as improvement or worsening [21].
Sagittal relationship (between maxillary and mandibular canines). Class I was considered when upper canine displacement from the ideal position was less than 1 mm.
Crowding (yes/no) was determined according to the plaster models before OT.
Position of teeth with GRs in the dental arch was assessed and classi ed to: 1) labial tooth position due to crowding, 2) labial/buccal root inclination, and 3) unfavourable occlusal contacts with antagonist teeth (cusp-to cusp or deep overbite).

Intraoral photographs
Gingival biotype (normal/thick or thin) was determined according to intraoral photographs before OT based on tooth crown morphology and capillary transparency ( Figure 3). The gingiva was identi ed as thin when triangular crowns and interproximal contacts close to the incisal edge were present, and the contours of tooth roots and the capillary network were easily visible [23,24]. The gingiva was recorded as thick/normal when square-shaped tooth crowns, large interproximal contact located more apically, and the contours of tooth roots and the capillary network were not visible [16,23].
The primary outcome variable was the change in the extent of GR in relation to the cement-enamel junction (CEJ) from before to after OT (T0-T1).
As the outcome of xed appliance orthodontic treatment may be in uenced by some speci c treatment (e.g. duration of orthodontic treatment, extraction of teeth) or patient-related factors (e.g. age, sex, gingival biotype, crowding), they were also analysed ( [25,26].

Data preparation
Firstly, analyses were performed on the patient as the unit of measure. Measurement analysis included all GRs at patient (cluster) level. GR change T0-T1 was evaluated by the mean millimetre change. Later, the analysis was performed at tooth (individual) level, taking into account possible clustering effects, to nd out whether the number of teeth with GR changed within the patient during the treatment [27].
Also, analyses were performed at the tooth level. Three groups were created according to the change of GR: 1) improved -when the most apical point of GR approached cement-enamel junction (the distance between the most apical point of GR and cement-enamel junction decreased (recession change was ≥0.5 mm and T0-T1 was possitive); 2) did not change -when the distance between the most apical point of GR and cement-enamel junction changed by less than 0.5 mm; 3) worsened -when the distance between the most apical point of GR and cement-enamel junction increased (change was ≥0.5 mm and T0-T1 was negative ).

Statistical analysis
Statistical analysis was performed using SPSS Statistics (Version 22.0. Armonk, NY, USA).
The interdependence of qualitative characteristics was evaluated using the chi-squared (χ 2 ) criterion. The Kolmogorov-Smirnov test was used to determine the normality of the parameter distribution. If variable distribution met the normality assumption, the Student's (t) criterion was applied to compare the quantitative sizes of two independent groups. When the variable did not meet the distribution normality condition, a signi cance level was veri ed by the Mann-Whitney U nonparametric method. For quantitative dependent variables, we used the paired test when distribution was normal, and the Wilcoxon nonparametric test when the test of normality of the investigated variables was denied.
The probability of the event given a certain risk factor was calculated using univariate and multivariate logistic regression analysis, including an odds ratio (OR) and its con dence interval (95% CI).
The difference between groups was considered statistically signi cant when p<0.05.
The intra-examiner error was estimated by performing gingival recession measurements twice with 1-week interval on plaster models before (T0) and after (T1) OT. Measurement error ranged between 0.16 and 0.37 mm. For the assessment of the reproducibility of the measurements made by two investigators (GA and EZ), the inter-class correlation coe cient (ICC) was calculated. ICC revealed high agreement between 2 investigators (0.98; 95% CI: 0.97; 0.99, p = 0.0001).

Results
Baseline data Fifty-one (5.1%) patient of the 993 assessed treated by two orthodontists during 12 years had at least one vestibular GR before OT. Of those, 37 (72.5%) patients, eight males (21.6%) and 29 females (78.4%) were included in the nal analysis according to the inclusion criteria. The mean age of the included patients was 28.7 (95% CI: 26.01, 31.40) years, and most of them were adults (n=33). Orthodontic treatment lasted for a mean of 21.4 (95% CI: 18.99, 23.82) months. The characteristics of the included patients are presented in Table   1.
Gingival recession change at the patient level: Median GR before OT was 1.47 mm (Q1; Q3: 1.13; 1.69); the number of teeth with GR within 37 patients before OT is shown in  Table 2.
Median GR change in the different groups of teeth is presented in Figure 7. The percentage of improved GR was highest on maxillary canines (84.6%); however, the highest amount of GR improvement in millimetres was observed on maxillary premolars and mandibular premolars (p<0.001). Healing of GR was mostly observed on maxillary incisors (46.7%). No recession was completely healed on mandibular incisors.
Median GR change in relation to different pre-treatment factors could be seen in Table 3. A statistically signi cant difference in median GR change was found between dental Angle I and III. The smallest median change of GR was found in Angle III (p=0.004) and open bite cases (p=0.001). Teeth with thin gingival biotype were the least prone to GR improvement compared to thick/normal gingival biotype (p=0.01).
As signi cant changes in GR were found, the in uence of speci c patient and treatment factors were analysed. Sagittal dental discrepancy: GR showed less tendency for healing in dental Angle Class III patients on one or both sides (p=0.006). Also, the odds ratio for the GR improvement was 6 times less (95% CI: 1.11, 6.00, p=0.026) in cases with dental Class III.

Treatment-related factors
No difference of median GR change at patient level was found for different orthodontic treatment duration groups (<18 vs >18 months) (p=0.43), or extraction and non-extraction cases (p=0.89).

Discussion
The present retrospective study aimed to evaluate the change of GRs, which were present before OT. As the focus on mucogingival factors is increasing in the orthodontic literature due to the increasing number of adult subjects in orthodontic practice, it is essential to analyse whether the improvement in GR may be in uenced by OT. The present study is one of the very few studies in the eld to assess changes of pretreatment GRs and related factors retrospectively. The available literature is more focused on the factors inducing GR after OT rather than changes [18,28,29].
In the present study, only 5.1% of patients with pre-existing GR were identi ed from the 12-year clinical material of 2 orthodontists showing rather low prevalence. The included patients were different in the type of dental malocclusion, the number of gingival recessions and other variables, partly explaining the current scarcity of the prospective studies (Table 1, Figure 4). Most of the included cases were adult patients (n=33), showing that GRs are more prevalent in adult patients and con rming ndings in the literature [11]. GRs in teenagers have been described to be associated with atypical tooth position; however, there were very few subjects <18 years in the present study (n=4) [30].
The measurements of 114 GRs in 37 patients were performed calculating the distance between the most apical marginal gingival contour and cement-enamel junction using a digital calliper, which is more accurate than using a periodontal probe. In the recent article, authors found that measurements on digital models were the most accurate; however, their calliper was not digital, and this could in uence the results [31]. Other authors compared pre-treatment and post-treatment clinical crown heights; however, this kind of measurement in adults could be affected by incisal wear and/or restorations during OT [18,32]. At tooth level, due to the small numbers of recession measurements, we performed qualitative analysis and classi ed improvement only when it was clinically signi cant (change from T0 to T1 was ≥0.5), so actual percentage with GR improvement could be higher. Changes of clinical periodontal variables are usually small and are in line with the ndings of the present study [33].
The results of the present study revealed a positive impact of OT on the change in GR (58.8% GRs improved).
The mean change of GRs was similar at the patient (0.45 mm) and tooth (0.55 mm) levels con rming the positive impact of OT. A high percentage of improved GRs could be in uenced by gingival enlargement during OT. However, gingival enlargement is mostly prevalent in teenagers with compromised oral hygiene; this study comprised mostly of adults with good oral hygiene, so the in uence of this factor is not discussed. Tooth group was found to be important in GR changes. The present study showed the most signi cant GR improvement in millimetres in maxillary premolars and mandibular premolars (Figure 7). The study by Melsen et al. (2005) showed equal amounts of improved and worsened GRs (42.3%) [28]. It is worth mentioning that these authors examined recessions only on mandibular incisors, which improved in more than half in our study; however, none of them fully healed. Mandibular incisors might be described most frequently because of the highest prevalence of new GRs after OT on their labial surfaces. The incomplete healing of GRs in the mandibular incisors revealed in the present study may be explained by the fact that the labial alveolar bone is anatomically very thin already before OT, and may predispose the development of dehiscences after the orthodontic movement of these teeth [34]. The percentage of improved GRs in the present study was the largest on maxillary canines (84.6%), which is in contrast to the results of Boke et al. (2014), where GRs worsened after orthodontic treatment [23]. That could be in uenced by negative torque, usual in Roth prescription of canine brackets, leading to more signi cant labial movement of the canine root. In the present study, all patients were treated with maxillary canine torque of 0° or +7°. Careful selection of torque for GR treatment, bearing in mind the position of the root in the alveolus, could have in uenced our results. However, this is only a speculation, as the present study had a retrospective design.
As the study patients had different malocclusions, we found it essential to analyse patient-related factors, which could have in uenced GR change. Pre-treatment overbite had an impact on GR change: patients with reduced overbite had a tendency for GR worsening in comparison to normal/deep overbite cases, where improvement of GR was found. The results are in line with the results by Zimmer et al. (2007), where an average improvement of 2 mm in GRs was observed in maxillary incisors [32]. The sample of this study consisted mainly of patients with a traumatic deep bite. Therefore, a signi cant improvement in GR was related to the elimination of the causative factor -mechanical load. Based on the results of the present study and earlier studies, it may be expected that the treatment of deep overbite may favour the improvement of GR ( Figure 6: a-c). Enhos et al. (2012) also found that patients with hypo-divergent vertical growth pattern (deep bite) have a lower prevalence of dehiscence than those with a normo-divergent or hyper-divergent (open bite) growth pattern [35].
GR improvement in the present study was similar in pre-treatment increased and decreased overjet at the patient level and tooth level (Table 3) Gingival biotype at patient level did not result in signi cantly different GR change; however, at tooth level, a smaller improvement was found in cases with thin biotype. Some authors suggest that GRs are less likely to improve after OT in cases with thin biotype [38]. However, in the study by Boke et al. (2014), no association between GR change and gingival biotype was reported suggesting that gingival thickness is not the most critical factor [23].
The sagittal dental relationship has also been found to in uence GR change. The literature mostly describes Angle class III cases as risk factors to GR worsening after OT, mainly due to retroclination of lower incisors to camou age malocclusion [29]. Sperry et al. observed that Class III patients with excessive dental compensations had more than three times as many teeth with labial GRs after OT in comparison to patients with Class I or Class II [39]. The results of the present study support these ndings: no signi cant GR improvement was found in Angle III patients. Patients with a Class III canine relationship had 2.6-times less chance of GR improvement than those with Class I or II. This observation can be explained by the anatomically thin buccal cortical plate, and the presence of dehiscences and fenestrations in the mandibular incisor region found in all types of untreated sagittal malocclusions [34]. Therefore, lingual movement of the lower incisor crowns, in order to compensate Class III malocclusion, may result in labial movement of incisor roots thereby causing or worsening GR [29]. Maxillary anterior teeth usually undergo proclination due to dentoalveolar compensation in Class III patients, which has also been found to induce the development or worsening of GR [38]. Therefore, the net effect is that Class III patients have a risk of GR worsening during OT.
Even if a small percentage (5.3%), most of the worsened GRs were in Class III patients with small OB and OJ before OT, and conversely patients with improved GRs were mostly with thick/normal gingival biotype and normal or deep bite.
An etiologic factor of the recession was classi ed to labial tooth position due to crowding, labial/buccal root inclination and unfavourable occlusal contacts with antagonist teeth (Table 3) [12]. Interestingly, GR improvements were somewhat similar despite etiologic factor suggesting that recession improvement may be due to correction of malocclusion, the better position of tooth roots in the alveolar envelope by lingually moving labially positioned teeth, or tooth roots (by changing inclination) and also resolving unfavourable occlusal contacts [12,20]. Therefore orthodontic movements should be carefully planned before the treatment to reach the aforementioned results [40].
Possible confounding factors such as surgical treatment of GR during OT, periodontal disease and systemic disease or medication that could in uence treatment outcome were excluded from the study. There were also no in uence of sex, age or treatment duration on GR change.
The clinical relevance of the present study is that orthodontic movements, together with some related factors, may induce GR change. The results of the present article show the merit of OT in adult patients with pretreatment GRs therefore con rming the necessity of interdisciplinary collaboration. The tendency of GRs to be improved may be expected in the normal/deep overbite cases, however more in maxillary teeth. However, GRs in open bite and Class III cases, also on mandibular anterior teeth, if present before OT, may need soft tissue augmentation, especially when planning unfavourable camou age movements in relation to the alveolar bone envelope. This decision is more critical in thin gingival biotype cases [20].

Limitations
The present study was of the retrospective design, and the measurements were made on plaster models, which were performed directly after bracket debonding, and this could in uence the results. Using plaster models can be problematic as they can be damaged and result in inaccurate measurements. Also we can not guarantee that during 12 years exactly the same materials were used for impressions. However, only good quality plaster models were included in the present study, and GR measurements were performed with a digital calliper. In the prospective study, we would recommend to wait 3-6 months after OT and perform direct measurements of GR or to perform measurements using digital models [31].
Another limitation is that gingival biotype evaluation was performed on intraoral photographs. In the prospective design, it could be performed clinically [31].
The sample size was small, however many years would be needed to collect clinically uniform groups for comparison.
The need for prospective and follow up studies is warranted to con rm the results of the present study.

Conclusions
Based on the ndings of the present retrospective study, it may be concluded that orthodontic treatment induces changes in gingival recessions. In more than half of the teeth, the improvement in gingival recession was observed. Class III malocclusion and reduced pre-treatment overbite had a negative impact on GR improvement.      Example of biotype assessment from digital photographs. a, b, c -the gingiva was identi ed as thin when triangular crowns and interproximal contacts close to the incisal edge were present, and the contours of tooth roots and the capillary network were easily visible; d, e, f -The gingiva was recorded as thick/normal when square-shaped tooth crowns, long interproximal contact located more apically, and the contours of tooth roots and the capillary network were not visible.

Figure 4
Number of teeth with pre-treatment gingival recessions in 37 patients.

Figure 5
The prevalence of pre-treatment gingival recessions in different groups of teeth.