Effects of Low-Level Laser Therapy as an Adjunct for Orthodontic Retention: A Systematic Review of Human and Animal Studies

Background: Low-level laser therapy (LLLT) has been veried effective in tooth-movement acceleration and pain alleviation during active orthodontic treatment, but its function remains inconclusive post-treatment. This systematic review aims to evaluate the effects of LLLT as an adjunct retention regimen following active orthodontic tooth movement (OTM). Methods: This review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Six databases (Cochrane Central Register of Controlled Trials, MEDLINE, Embase, Pubmed, Scopus, ProQuest) were comprehensively searched for human and animal studies published till December 2019 and screened according to our eligibility criteria. The risk of bias was assessed based on the Cochrane Handbook for Systematic Reviews of Interventions and Systematic Review Center for Laboratory Experiment Tool. Two independent reviewers performed all procedures in duplicate. Any disagreement was resolved by discussion or consultation with a third reviewer. Results: A total of 394 records were identied from the initial search. Following screening, 15 full-text articles were reviewed for eligibility (ĸ>0.90), and ultimately, eight studies (three human studies and ve animal studies) were included in this review. The key outcomes considered were ‘preventing tooth relapse’ and ‘rehabilitating root resorption’. Two controlled clinical trials (CCTs) and one animal study supported the preventive effects of LLLT on the relapse of post-orthodontic tooth positions. In contrast, two animal studies reported opposing ndings. Regarding the rehabilitation of root resorption, evidence supported the reparative potential of LLLT in orthodontic force-induced root resorption. Overall, there was a high risk of bias among studies, except for one randomised controlled trial. Due to the substantial heterogeneity among studies in terms of their types, participants, designs, LLLT settings and variables of interest, it was not feasible to conduct a meta-analysis; therefore, a qualitative synthesis is presented. Conclusion: The quality of evidence for LLLT contributing to the maintenance of orthodontic outcomes or a better treatment prognosis remains low. There is considerable controversy over the effects of LLLT on orthodontic relapse. However, the use of LLLT after OTM has promising reparative effects for root resorption

factors for the application of LLLT after the removal of active orthodontic forces by analysing current evidence of the effects of LLLT in post-OTM treatments, especially in terms of orthodontic relapse, dental or periodontal health and patients' self-perception.
Question speci cation formatted in PICOS Population: Subjects who have undergone active OTM Intervention: LLLT Comparison: Placebo Outcomes: Primary outcomes (post-OTM tooth stability, dental or periodontal health status and participant's subjective perceptions); secondary outcomes (histological or biochemical changes) Study designs: Randomized-controlled trial (RCTs), clinical-controlled trials (CCTs), and animal experiments Methods This systematic review was performed and reported following the instructions of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guideline [43,44]. The protocol has been prospectively registered on the PROSPERO online database (CRD49019132133). Headings, free text words and their synonyms were applied as search terms, including 'orthodontic/appliance/force', 'retention/maintenance/stability/relapse' and 'low-level laser/low-intensity laser/soft laser/photomodulation'. The detailed search strategy is presented in Appendix 1.

Study selection
The eligibility criteria are listed as follows.
Inclusion criteria 1. Randomised controlled trials (RCTs) or controlled clinical trials (CCTs) involving patients with and without irradiation by low-level lasers after active orthodontic treatment.
2. Experimental animal studies involving subjects with and without low-level laser application following the termination of an active orthodontic f Exclusion criteria 1. Studies with no subjects having received LLLT after active orthodontic treatment completion.
2. Studies involving subjects who have severe maxillofacial deformities or have undergone any orthopaedic or surgical procedures.
3. Studies involving subjects with systematic diseases or under any pharmaceutical therapies that could potentially affect bone and soft tissue remodelling.

4.
In-vitro studies, case reports, reviews, personal opinions and technique description articles without sample reporting.

Human or animal studies without any demonstrations of post-OTM effects.
Accordingly, all titles and abstracts obtained from the electronic search were screened by the two reviewers independently. Full articles were retrieved for the nal assessment and their reference lists were also screened based on the aforementioned criteria. During the process, any disagreement between the two reviewers was resolved by discussion or consultation with a highly experienced reviewer (YQY). Cohen's Kappa values were computed to verify the two-round inter-reviewer reliabilities, which would be considered acceptable if not lower than 0.6.

Data extraction and analysis
The following data were extracted: general information ( rst author and year of publication), study type and design, participants and target teeth (sample size and characteristics), orthodontic regimen (active and post-active strategy and period), low-level laser protocols (types, wavelength, beam size, mode, output power, dosage density, time of onset, duration, frequency and method of delivery), assessments (approach, region of interest, outcome variables and timepoints); primary outcomes related to post-OTM tooth stability, dental or periodontal health status and participant's subjective perceptions; and secondary outcomes related to histological or biochemical changes.

Assessment the risk of bias
The assessment for the risk of bias of all human studies was performed in RevMan5.3 [45] using the Cochrane Risk of Bias Tool [46]. Seven domains were considered: (1) random sequence generation; (2) allocation concealment; (3) blinding of participants and personnel; (4) blinding of outcome assessment; (5) incomplete outcome data; (6) selective reporting; and (7) other bias. For the animal studies, the risk of bias was assessed based on the Systematic Review Center for Laboratory Experiment Tool (SYRCLE tool) [47]. Ten domains covering the aspects of selection, performance, detection, attrition and reporting were considered to grade the quality of evidence.

Data synthesis
The data of interest from human and animal studies were synthesised separately because of the substantial differences between them. Within each study type, data on individual settings, eligibility criteria, study design, intervention, outcome variables and statistical methods were further analysed. If heterogeneity was acceptable, quantitative synthesis and meta-analysis of the retrieved data were performed; otherwise, a narrative description is presented.

Characteristics of selected studies
The process of study search and selection is illustrated in Fig. 1. The electronic search until March 2020 yielded a total of 386 relevant records from the six databases; another eight records were identi ed by screening bibliographies. After removing duplicates, the remaining 298 studies were analysed by title and abstract, which left 15 articles for full-text comprehension. According to the eligibility criteria, seven of these articles were further excluded for speci c reasons.
Only tooth relapse and dental health in terms of root resorption were investigated out of all post-OTM outcomes, and no studies reported any impacts of LLLT on patients' subjective perceptions. For orthodontic relapse, two of the studies were CCTs [48,49] and the others were animal experiments [51][52][53]. These studies covered two types of tooth movement, rotational and transitional relapse. The other prognosis was orthodontically induced in ammatory root resorption (OIIRR), which was reported in one RCT [50] and two animal studies [54,55]. Considering the substantial heterogeneity exhibited in outcomes, no meta-analysis could be performed; therefore, a qualitative synthesis of LLLT's post-OTM effects was performed in a narrative manner.
Effects on post-OTM stability of tooth position Two CCTs [48,49] and three animal studies [51][52][53] ful lled the inclusive criteria investigating the effects of LLLT on post-OTM stability of tooth position, and the results they documented were inconclusive. Detailed information from all studies on post-OTM stability is summarised in Table 1. CCT: clinical controlled trial; SMD: split-mouth design; N: number of participants; n: number of evaluated teeth; FA: xed appliances; CSF: circumferential supr metalloproteinase; LLLT: low-level laser therapy Table 1 Characteristics of included studies on orthodontic relapse

Clinical studies
The favourable preventive effects of LLLT on orthodontic relapse were reported in both CCTs, but one showed a statistically signi cant difference between groups [49] and one did not [48]. One study showed that teeth irradiated with a low-level laser (GaAlAs, 810 nm, continuous wave, 35.7 J/cm 2 ) had a nearly 60% decrease in the amount of post-OTM relapse compared with their control counterparts (p < 0.05) [49]. The other study using a low-level laser (GaAs, 904 nm, pulse wave, 4.9 J/cm 2 ) detected a favourable reduction in orthodontic relapse but it did not reach statistical signi cance, except a signi cant increase in the alveolar crest height and a substantial retardation of bone density reduction in areas under irradiation were found [48].
Although the two studies consistently reported LLLT's positive effects on orthodontic relapse, their methodologies appeared to diverge. First, two types of postactive tooth movement were discussed. One study observed transitional tooth movement [48], whereas the other study examined the impacts on rotational relapse [49]. Second, the two studies adopted different retention regimens. One left the xed appliances attached for the rst 45 days and then used Hawley's retainers for the following 6 months [48], whereas the other did not use any post-OTM retainers [49]. Third, the adopted LLLT parameters were quite different.
One study used irradiation with a GaAl laser at 904 nm in pulse waves and a 4.9-J/cm 2 dosage density [48], whereas the other applied a GaAlAs laser at 810 nm in continuous mode and a 35.7-J/cm 2 dosage density [49]. Importantly, the assessment timepoints were also not comparable. One study undertook a long-term observation for 1.5 years [48], whereas the other had an observation period of 30 days post-OTM [49].

Animal studies
The animal studies investigating the effects of LLLT on post-OTM tooth stability include one that assessed rotation [51] and two that observed transitional movement [52,53] in rodent and canine models, respectively.
For rotational relapse, Kim et al. [51] found that a GaAlAs laser (808 nm, pulse wave, 4.63-6.47 J/cm 2 ) deteriorated post-OTM stability of tooth position by approximately 15% when not using any retainers (p < 0.05). For transitional relapse, one rodent study showed a positive effect of LLLT on tooth stability (GaAlAs, 830 nm, continuous wave, 23 J/cm 2 ), but the effect did not reach statistical signi cance (p > 0.05) [52]. In contrast, another rodent study found a negative effect of LLLT, indicating that incisors which underwent LLLT (GaAlAs, 780 nm, continuous wave, 20 J/cm 2 ) exhibited higher rates of relapse compared with the control counterparts (p < 0.05) [53]. Interestingly, both rodent studies [52,53] found that the number of osteoblasts was increased surrounding the laser-irradiated teeth, and one of them found an increase in the expression of matrix metalloproteinases in the LLLT group [53].

Effects on orthodontic-induced in ammatory root resorption (OIIRR)
To date, the number of studies investigating the effects of LLLT on OIIRR after OTM is limited, but the existing evidence is relatively conclusive and of high quality. One RCT [50] and two animal experiments [54,55] showed some level of rehabilitative effect of LLLT on root surfaces. The detailed information of all included studies is summarised in Table 2. CCT: clinical controlled trial; SMD: split-mouth design; N: number of participants; n: number of evaluated teeth; FA: xed appliances; CSF: circumferential supr MMP: matrix metalloproteinase; LLLT: low-level laser therapy Table 2 Characteristics of included studies on orthodontically-induced in ammatory root resorption Clinical studies The one RCT that prescribed LLLT (AlGaInP, 660 nm, continuous wave, 3.6 J/cm 2 ) applied it to one side of patients' upper rst premolars immediately after the removal of a buccal tipping force during retention, whereas their counterparts on the opposite side were subject to a placebo laser. Six weeks post OTM, the mean total crater volume on the root surfaces of laser-irradiated teeth was 0.033 ± 0.039 mm 3 less than that of the placebo-irradiated teeth. However, this difference did not exhibit statistical signi cance (p > 0.05) [50].

Animal studies
Of the two animal studies both adopted rodent models, one used a parallel study design involving 30 subjects and 30 maxillary rst molars for the outcome assessment [54], whereas the other study applied a split-mouth design including 20 subjects with 40 paired teeth in total [55]. They both found that teeth underwent GaAlAs laser irradiation (820 nm, continuous wave, 4.8 J/cm 2 [54], and 810 nm, continuous wave, 75 J/cm 2 [55]) showed less OIIRR compared with their control counterparts (p < 0.05). In addition, both studies found that the number of clastic cells and the ratio of receptor activator of nuclear factor kappa-B ligand to osteoprotegerin (RANKL/OPG) decreased signi cantly in LLLT-irradiated teeth compared with their control counterparts, indicating that this positive impact was associated with the downregulation of clastogenic activities [54,55]. Moreover, one study determined that the numbers of osteoblasts and broblasts surrounding the teeth in the laser-irradiated group were also remarkably increased, especially when LLLT was delivered after the active orthodontic treatment compared with irradiation during OTM [54]. This suggests that LLLT is more likely to have a restorative effect than a preventive effect on root resorption.

Quality evaluation
Clinical studies The quality assessment of the three human studies was performed according to the guidelines of the Cochrane Risk of Bias Tool. The two CCTs [48,49] were ranked as having a high risk of bias with emphasise on the lack of blinding and randomisation. Additionally, they conducted no power analysis or sample size calculation, which further diminished the quality of evidence. In comparison, the risk of bias of the RCT for the effects of LLLT on OIIRR ranked considerably lower, as all seven evaluation domains were ful lled [50] (Fig. 2) Animal studies The qualities of the animal studies were assessed using the SYRCLE tool. All animal studies were ranked as possessing a high risk of bias due to their lack of ful lment of or unclear declaration of randomisation, balancing subjects' characteristics, blinding during intervention and assessment, or avoidance of data attrition [51][52][53][54][55]. The schematic representation of the assessment results of risk of bias in animal studies is provided in Fig. 3.

Discussion
Of all the included studies evaluating the effects of LLLT on post-OTM tooth stability, two discussed rotational relapse. In clinical practice, orthodontically derotated teeth are more likely to return to their original state, even when orthodontic retainers are routinely administered. Past studies have revealed that soft tissue turnover, i.e. the remodelling of collagen and elastic bres, plays a vital role in the occurrence of rotational relapse [56]. Based on this nding, some researchers hypothesised that the biomodulation of soft tissues by LLLT might be a promising approach to prevent post-OTM rotation. However, according to the results of the two relevant studies, the impacts of LLLT could be either positive or negative depending on various factors to the maintenance of orthodontically de-rotated tooth position. In line with the biphasic dosage-response theory [57], the rst factor is the dosage density. Jahanbin et al. [49] found that a GaAlAs laser with 810-nm wavelength could alleviate the degree of rotational relapse when the dosage density was high at 35.7 J/cm 2 . In contrast, Kim et al. [51] used the same type of laser (GaAlAs, 808 nm) with a low dosage density (4.63-6.47 J/cm 2 ) and found that it decreased post-treatment tooth stability. One systematic review of past in vitro studies reported that laser with a dosage density of less than 16 J/cm 2 could promote broblast growth, proliferation and osteogenic differentiation, where laser with an extremely high dosage density exhibited inhibitory effects [25]. It is possible that the effects of LLLT on rotational relapse also follow the same rules, converting broblasts from predominantly anabolic activities to catabolic ones corresponding to a shift from adverse effects to positive ones along with the increase in dosage density [42]. However, this interpretation only applies to teeth free of movement after the immediate termination of active forces. As Kim et al [58] suggests, LLLT could act differently to the orthodontic outcomes for teeth prescribed with retainers and those without by stimulating soft tissue metabolism. Therefore, whether the adjunctive LLLT has extra bene ts on the e cacy of conventional retention appliances is still unclear. Finally, there are other confounders that hinder any generalisation of the effects of LLLT on rotational relapse, including the substantial heterogeneities in the characteristics of subjects and the initial status of the experimental teeth. In addition, both articles have a high risk of bias because of a limited number of recruited subjects and no sample size calculations. Further investigations with a higher quality of evidence are thus highly warranted.
The other two post-OTM outcomes discussed by the remaining six studies, i.e. transitional relapse and root resorption, are both closely related to the activities of osteoblast-like cells and osteoclast-like cells for hard tissue remodelling. On one hand, after the termination of active forces, alveolar processes generate some hyalinised areas in response to the released mechanical forces, which then trigger osteoclast recruitment and bone resorption in the direction of tooth relapse on the previous tension side. Meanwhile, more anabolic activities such as osteoblast proliferation and differentiation occur on the opposite side, leading to bone regeneration against the direction of tooth relapse to compensate for previous bone resorption [16,17,60]. On the other hand, pathological OIIRR occur during OTM when osteoclastic-like cells accumulate near the root surfaces [61,62]. After orthodontic force removal, a physiological repair would follow involving the deposition of uncalci ed-cementoid matrix, broblast-like cells and cementoblast cells as well as the detachment of clastic cells [63]. Past cellular [28] and molecular investigations [64] have documented LLLT's capacity to modulate the activities of osteoblasts and osteoclasts with bone-related biomarkers such as RANKL and OPG. This provides a biological explanation for adopting LLLT to prevent transitional relapse and OIIRR. However, this theory is yet to be supported.
Among the three studies on transitional relapse [48,52,53], two failed to observe any signi cant reduction in the amount of post-OTM displacement for LLLTirradiated teeth after the termination of active forces [48,52]. One study showed a statistically signi cant detrimental effect of LLLT on tooth position maintenance [53]. This inconsistency in the results on transitional relapse might also be related to the diversity of laser types and parameter settings, similar to the results on rotational relapse. However, one past review [28] reported a general susceptibility of osteoblast-like cells to multiple laser parameters without a clear speci city, unless the dosage densities are extremely high. Thus, the disagreement between the three studies is possibly dominated by the variation in their retention regimens. Two studies implemented LLLT immediately after OTM [48,52], whereas the other one delivered irradiation after a 1-week retention [53]. Because the lag of adjacent alveolar reconstruction is the primary reason for transitional relapse, the effects of LLLT on tooth position maintenance might omit the critical period and recede with time.
Comparably, the results of the three studies on OIIRR are more consistent [50,54,55]. The two animal studies reported favourable effects of LLLT with statistically signi cant reductions in OIIRR for teeth that received LLLT during the post-OTM period compared with their counterparts [54,55], and one clinical trial showed a general tendency for a positive effect of LLLT but without a statistical signi cance [50]. These bene cial effects of LLLT in the reduction of OIIRR are in line with previous ndings [65,66] suggesting that LLLT could enhance the development of roots or their reparative process by stimulating the proliferation of cementoblast cells and the formation of secondary cementum. The two rodent studies applied near-infrared light at 810 and 820 nm [54,55], whereas the clinical study adopted an AlGaInP laser at a 660-nm wavelength [50]. Considering the penetration depth of LLLT's PBM is largely dependent on the wavelength [67] and that the longer is the wavelength the deeper can photons travel through biological surfaces [68], it might be possible that the nonsigni cant differences found in the clinical study were because the PBM effect cannot occur at the depth of root surfaces. The second reason might be the frequency of repeated laser applications, as longer time intervals and total irradiation spans in the clinical study might vary the effects on OIIRR [50]. Finally, the substantial differences in the study subjects and retention regimens might also blur the interpretation of the above inconsistencies and whether the effects reached statistical signi cance.
It is clear that current evidence is still insu cient to deduce the effects of LLLT on the prognosis of orthodontic treatment after the active OTM stage, in terms of outcomes for both tooth relapse and OIIRR. One hinderance in generalising these results is the aforementioned methodological discrepancies, and another is the fact that the underlying cellular and molecular mechanisms are yet to be fully clari ed. A classical theory for the PBM effect considers the activities of cytochrome C oxidase (CCO) in the respiratory chain, which are boosted by photons in the red and infrared wavelengths that penetrate the mitochondria [69].
By inducing high levels of adenosine triphosphate (ATP) production and vital second messengers such as nitric oxide and reactive oxygen species, LLLT regulates various metabolic activities such as cell proliferation, migration, adhesion and apoptosis. However, this theory cannot explain the inconsistencies between some therapeutic laser wavelengths and the absorption spectra of CCO. Another hypothesis, the 'water oscillator paradox', was proposed by Santana-Blank et al. [70] implicating intracellular water dynamics also play an essential role in PBM effects. Recently, Wang et al. [71] found that heat/light-gated ion channels seem to be the primary photoreceptor for 980-nm wavelength lasers, whereas CCO is the primary photoreceptor for the 810-nm wavelength. Most studies reviewed here used LLLT in the 808-830-nm wavelength [49,51,52,54,55], except for one study that used a 660-nm laser for OIIRR [50] and two that used 780-nm [53] and 904-nm [48] lasers for transitional relapse. Considering that chromophores might alternate with different wavelengths, the optimal dosage for achieving favourable PBM effects can vary and may signi cantly in uence post-OTM outcomes.
This systematic review is the rst to comprehensively evaluate the effects of LLLT in post-OTM scenarios, aiming to justify the application of LLLT for plausible orthodontic prognosis. According to our results, there is great controversy over the effects of LLLT on orthodontic relapse, but its use in post-OTM applications for reparative effects on root resorption are generally recommended. However, great heterogeneity was noted in the assessed outcomes, study subjects, OTM and post-OTM regimens and LLLT parameters. Moreover, most studies suffered from limitations including small sample sizes, high risk of bias, relatively short observation periods and a shortage in demonstrations of cellular and molecular changes. Therefore, more well-designed studies with broader LLLT parameters and more consistent orthodontic and post-OTM settings are highly anticipated in the near future.

Declarations
Ethical approval and consent to participate The protocol has been prospectively registered on the PROSPERO online database (CRD49019132133).

Consent for publication
Not applicable